index.xml

<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Geert Litjens</title><link>https://geertlitjens.nl/</link><atom:link href="https://geertlitjens.nl/index.xml" rel="self" type="application/rss+xml"/><description>Geert Litjens</description><generator>Wowchemy (https://wowchemy.com)</generator><language>en-us</language><copyright>© 2019</copyright><lastBuildDate>Tue, 11 Jan 2022 15:32:00 +0100</lastBuildDate><image><url>https://geertlitjens.nl/media/icon_hue0c2058e8f722cf54419e3dfd4f45926_15543_512x512_fill_lanczos_center_3.png</url><title>Geert Litjens</title><link>https://geertlitjens.nl/</link></image><item><title>PANCAIM</title><link>https://geertlitjens.nl/project/eu-pancaim/</link><pubDate>Tue, 11 Jan 2022 15:32:00 +0100</pubDate><guid>https://geertlitjens.nl/project/eu-pancaim/</guid><description>&lt;h2 id="overview">Overview&lt;/h2>
&lt;p>The central PANCAIM concept is to successfully exploit available genomic and clinical data to improve personalized medicine of pancreatic cancer. PANCAIM’s concept is unique as it integrates the whole spectrum of genomics with radiomics and pathomics, the three future pillars of personalized medicine. The integration of these three modalities is very challenging in the clinic, but also with AI. PANCAIM uses an explainable, data-efficient, two-staged AI approach. AI biomarkers transform the unimodal data domains into interpretable likelihoods of intermediate disease features. A second AI
layer merges the biomarkers and responds with an integrated assessment of prognosis, prediction and monitoring of therapy response, to assist in clinical decision making.
PANCAIM builds on four key concepts of AI in Healthcare: data providers, clinical expertise, AI developers, and MedTech companies to connect to data and bring AI to healthcare. Data quantity and quality is the main factor for successful AI. Partners provide eleven Pan European repositories of almost 6000 patients that are open to ongoing accrual. SME Collective Minds builds the GDPR data platform that hosts the data and provides a trustable connection to healthcare for even more and sustainable data. SME TheHyve builds tooling to connect to more genomic repositories (EOSC Health). Six
Pan European academic centers provide clinical expertise across all modalities and help realize a curated, high quality annotated data set. Partners also include expert AI healthcare researchers across all clinical modalities with a proven track record. Finally, Siemens Healthineers provides their AI expertise and tooling to bring AI into healthcare for clinical validation and swift clinical integration in 3000 health care institutes.&lt;/p>
&lt;p>
&lt;figure id="figure-integration-of-ct-and-pathology">
&lt;div class="d-flex justify-content-center">
&lt;div class="w-100" >&lt;img alt="Integration of CT and pathology" srcset="
/project/eu-pancaim/featured_huc33219bb2767acfe72c5a387084a2658_379429_8297772df108dec26ee750096995be04.webp 400w,
/project/eu-pancaim/featured_huc33219bb2767acfe72c5a387084a2658_379429_dadfd14345de9a5c898338f0c6bec109.webp 760w,
/project/eu-pancaim/featured_huc33219bb2767acfe72c5a387084a2658_379429_1200x1200_fit_q75_h2_lanczos_3.webp 1200w"
src="https://geertlitjens.nl/project/eu-pancaim/featured_huc33219bb2767acfe72c5a387084a2658_379429_8297772df108dec26ee750096995be04.webp"
width="480"
height="361"
loading="lazy" data-zoomable />&lt;/div>
&lt;/div>&lt;figcaption>
Integration of CT and pathology
&lt;/figcaption>&lt;/figure>
&lt;/p>
&lt;h2 id="tasks">Tasks&lt;/h2>
&lt;p>Pathology is a diagnostic gold standard for PDAC. However, the information it provides is limited to definitive confirmation of the diagnostic entity and the size and locoregional extent of the tumour. The only tumour-intrinsic feature is the grade of differentiation, a historical concept that is of no relevance for the management of PDAC. The rich information contained in tumour morphology integrates results of interactions at all levels - genetic, epigenetic, microenvironmental -, but is left totally unexplored. It is only with the advent of AI that attempts at deciphering this rich information have been undertaken. Recent studies on a variety of other cancers show that AI can extract information from routine pathology tissue sections that relate to underlying genomic aberrations and allows prognostic discrimination between patients that are currently lumped within the same clinical stage. With the wealth of data accessible in its repository, PANCAIM is in a unique position to decipher and fully exploit the prognostic and predictive information that hitherto has remained unmined in surgical and biopsy specimens from PDAC patients.&lt;/p>
&lt;p>Recent studies on a variety of other cancers show that AI can extract information from routine pathology tissue sections that relates to underlying genomic aberrations and allows prognostic discrimination between patients that are currently lumped within the same clinical stage. Although this has been shown and applied in for example, breast cancer [Jaber et al, Breast Cancer Research, 2020] and mesothelioma [Courtiol et al, Nature Medicine 2019], these insights have not yet been applied to PDAC. A key reason is the limited availability of large datasets that combine radiology, pathology, genetics, and clinical follow-up. With the wealth of data accessible in its repository, PANCAIM is in a unique position to decipher and fully exploit the prognostic and predictive information that hitherto has remained unmined in surgical and biopsy specimens from PDAC patients. Specifically, it is our ambition to use novel machine learning techniques such as neural image compression [Tellez et al, TPAMI 2019] to elucidate key pathomics features which can help predict prognosis, treatment response and genetic alterations in PDAC (Figure 9, Figure 8). By additionally applying for the latest advances in explainable artificial intelligence, such as attention-weighting and saliency mapping, PANCAIM furthers the acceptance and integration of these pathomic features in clinical practice.&lt;/p></description></item><item><title>IMI BigPicture</title><link>https://geertlitjens.nl/project/imi-bigpicture/</link><pubDate>Tue, 11 Jan 2022 15:31:52 +0100</pubDate><guid>https://geertlitjens.nl/project/imi-bigpicture/</guid><description>&lt;h2 id="overview">Overview&lt;/h2>
&lt;p>BIGPICTURE, a pathology-led consortium, has the vision to become the catalyst in digital transformation in Pathology. Our
mission is to create the first European GDPR compliant platform, in which both quality-controlled Whole Slide Imaging (WSI)
data and advanced Artificial intelligence (AI) algorithms will exist. The BIGPICTURE platform will be built on existing assets
of ELIXIR EU data infrastructure, including the federated European Genome-phenome Archive (EGA) technology for
managing the exchange of confidential information between contributors and users. The consortium will use Cytomine, an
established open-source, cross-platform framework to develop unique tools for access to WSI, including annotations and
visualisation of algorithm results, while we will develop new and generic models to facilitate AI development and mining of
WSI data. By engaging and building consensus with all the relevant stakeholders, we will contribute to the development of a
regulatory framework for digital slides and AI-based methods. Finally, BIGPICTURE envisions sustainability of its platform
through a community- based model which relies on reciprocity, value creation and inclusiveness.
To achieve our vision, we have brought together Europe’s leaders in the field of computational pathology who have access
to national and European high-performance computing infrastructures as well as Europe’s fully digitalised pathology
departments. Additionally, the consortium has currently access to approximately 4.5 million clinical WSI covering a wider
range of indications through 17 partners and 23 third parties from the largest European and international pathology and trial
groups. Our consortium is further strengthened by the presence of the European Society of Pathology, Digital Pathology
Association, FDA and 9 SMEs as partners, while we are further supported by professional societies, and patient advocates.&lt;/p></description></item><item><title>ICARUS</title><link>https://geertlitjens.nl/project/vidi-icarus/</link><pubDate>Tue, 11 Jan 2022 15:29:56 +0100</pubDate><guid>https://geertlitjens.nl/project/vidi-icarus/</guid><description>&lt;h2 id="overview">Overview&lt;/h2>
&lt;p>An ever-increasing amount of treatment options is available to prostate cancer patients. Although this is a positive development, it also increases the complexity of selecting the right therapy for the individual patient. The fusion of in and ex vivo information streams, such as from radiology and pathology, offers a promising avenue for improved models of disease physiology and progression, and consequently, better strategies for treatment selection. However, to build accurate models, large sets of fused radiology/pathology data are needed, which have been impossible to obtain due to the time-consuming and expensive nature of acquiring such datasets.&lt;/p>
&lt;p>In this project, we propose an artificial-intelligence-based platform that can automatically combine large archival sets of digitized histopathological slides and multi-parametric MRI (mp-MRI) and leverage them to build a disease model which will improve 1) identification of clinically significant prostate cancer, 2) selection of patients for active surveillance, and 3) predict lutetium-PSMA treatment success.&lt;/p>
&lt;p>
&lt;figure id="figure-flowchart-of-the-interlocking-work-packages">
&lt;div class="d-flex justify-content-center">
&lt;div class="w-100" >&lt;img alt="Flowchart of the interlocking work packages" srcset="
/project/vidi-icarus/featured_hu870ad888f693946160ad12cf0c6727a2_242380_235c226354bd2c55292402cf8a4cd756.webp 400w,
/project/vidi-icarus/featured_hu870ad888f693946160ad12cf0c6727a2_242380_ef114b7c9d66fbbee389f510f5a1db6b.webp 760w,
/project/vidi-icarus/featured_hu870ad888f693946160ad12cf0c6727a2_242380_1200x1200_fit_q75_h2_lanczos.webp 1200w"
src="https://geertlitjens.nl/project/vidi-icarus/featured_hu870ad888f693946160ad12cf0c6727a2_242380_235c226354bd2c55292402cf8a4cd756.webp"
width="760"
height="618"
loading="lazy" data-zoomable />&lt;/div>
&lt;/div>&lt;figcaption>
Flowchart of the interlocking work packages
&lt;/figcaption>&lt;/figure>
&lt;/p>
&lt;h2 id="tasks">Tasks&lt;/h2>
&lt;p>This project encompasses five objectives: I) establish a collection of combined mp-MRI and digitized prostatectomy specimens from 1350 patients, II) build an automated reconstruction algorithm for the generation of 3D tissue volumes from 2D digitized histopathology slides, III) develop registration techniques for spatial alignment of ex vivo histopathology to in vivo mp-MRI, IV) learn mp-MRI / histopathology correlations using unique deep streaming generative models, and V) Evaluate the learned correlations for improved diagnostics, active surveillance, and lutetium-PSMA treatment selection.&lt;/p>
&lt;p>The impact of the project will not just be improved diagnostic and treatment decisions for patients but can be the starting point of an entirely new field of cross-medical-specialty research; the developed platform can be leveraged for other cancer types and even non-oncological diseases.&lt;/p></description></item><item><title>AISCAP</title><link>https://geertlitjens.nl/project/erc-aiscap/</link><pubDate>Tue, 11 Jan 2022 15:29:38 +0100</pubDate><guid>https://geertlitjens.nl/project/erc-aiscap/</guid><description>&lt;h2 id="overview">Overview&lt;/h2>
&lt;p>Computational pathology, the application of advanced machine learning (ML) methods to digitized tissue sections, can revolutionize cancer care and research. Specifically, I propose a paradigm shift by moving away from the currently used manual grading systems towards ML-supported patient prognostication. However, significant knowledge gaps are hindering the field of computational pathology. We do not know how to: 1) effectively leverage global and local information in WSIs, 2) identify pan-cancer and cancer-specific prognostic features, and 3) make ML models explainable and interpretable.&lt;/p>
&lt;p>
&lt;figure id="figure-flowchart-of-the-interlocking-work-packages">
&lt;div class="d-flex justify-content-center">
&lt;div class="w-100" >&lt;img alt="Flowchart of the interlocking work packages" srcset="
/project/erc-aiscap/featured_hua9c1074d7efe068f74d45a2f6193ecd8_184892_956574a2efb802c9a8e60a7c864f7944.webp 400w,
/project/erc-aiscap/featured_hua9c1074d7efe068f74d45a2f6193ecd8_184892_633411acd04525656c0bd27e3ed63564.webp 760w,
/project/erc-aiscap/featured_hua9c1074d7efe068f74d45a2f6193ecd8_184892_1200x1200_fit_q75_h2_lanczos.webp 1200w"
src="https://geertlitjens.nl/project/erc-aiscap/featured_hua9c1074d7efe068f74d45a2f6193ecd8_184892_956574a2efb802c9a8e60a7c864f7944.webp"
width="760"
height="488"
loading="lazy" data-zoomable />&lt;/div>
&lt;/div>&lt;figcaption>
Flowchart of the interlocking work packages
&lt;/figcaption>&lt;/figure>
&lt;/p>
&lt;h2 id="tasks">Tasks&lt;/h2>
&lt;p>This ambitious project will address these critical knowledge gaps by building on the novel stochastic streaming gradient descent developed in my group. First, I will push SSGD to the next level by integrating hierarchical hyperparameter optimization and separable convolutions. Second, to identify pan-cancer and cancer-specific prognostic biomarkers, I will integrate innovative multi-task and cross-task learning algorithms with SSGD. Third, I will leverage the latest advances in concept learning and natural language processing to endow deep neural networks with unprecedented transparency and explainability. Last, I will validate our developed methodology in the largest dataset of oncological WSIs globally.&lt;/p>
&lt;p>By publicly releasing all developed tools and data, the proposed project will have a scientific multiplier effect on the fields of computational pathology, machine learning, and oncology. Specifically, the enhanced SSGD method can open new research areas for ML that require data across scales, such as remote sensing. My novel approach to ML explainability can encourage the adoption of innovative technologies, such as self-driving cars. Last, the derived specific and pan-cancer biomarkers will have a tremendous impact on the quest to understand cancer development and progression, and ultimately on public health and the economy&lt;/p></description></item><item><title>DeepGrading</title><link>https://geertlitjens.nl/project/ppp-deepgrading/</link><pubDate>Tue, 11 Jan 2022 15:29:21 +0100</pubDate><guid>https://geertlitjens.nl/project/ppp-deepgrading/</guid><description>&lt;h2 id="overview">Overview&lt;/h2>
&lt;p>Optimal treatment decisions for cancer patients are hampered by variability in grading among pathologists. When there is a suspicion for cancer, typically a tissue biopsy is taken. This biopsy is stained with hematoxylin and eosin (H&amp;amp;E) and evaluated by a pathologist for the presence and aggressiveness (i.e. grading) of the tumor. Professional organizations have drafted standardized guidelines on how this grading should be performed; however, there is still significant inter- and intra-observer variation among pathologists. Artificial intelligence, specifically through deep learning, has shown to increase efficiency and consistency in histopathological diagnosis, and, recently, to perform grading at the expert level. We have prototypes of such algorithms validated in the lab which could significantly impact clinical practice, but have not yet been tested in routine diagnostics. The aim of this project is three-fold.&lt;/p>
&lt;p>
&lt;figure id="figure-example-of-automated-prostate-cancer-grading-including-bar-plots-of-grades-assigned-by-multiple-pathologists">
&lt;div class="d-flex justify-content-center">
&lt;div class="w-100" >&lt;img alt="Example of automated prostate cancer grading, including bar plots of grades assigned by multiple pathologists" srcset="
/project/ppp-deepgrading/featured_hu89bde7ffdd6d32f799a92da8cbce97e7_165693_8d2064006c7afc74e65315ac2b5f696a.webp 400w,
/project/ppp-deepgrading/featured_hu89bde7ffdd6d32f799a92da8cbce97e7_165693_ae810ec51e0d1ebb380abef5920253ac.webp 760w,
/project/ppp-deepgrading/featured_hu89bde7ffdd6d32f799a92da8cbce97e7_165693_1200x1200_fit_q75_h2_lanczos.webp 1200w"
src="https://geertlitjens.nl/project/ppp-deepgrading/featured_hu89bde7ffdd6d32f799a92da8cbce97e7_165693_8d2064006c7afc74e65315ac2b5f696a.webp"
width="760"
height="376"
loading="lazy" data-zoomable />&lt;/div>
&lt;/div>&lt;figcaption>
Example of automated prostate cancer grading, including bar plots of grades assigned by multiple pathologists
&lt;/figcaption>&lt;/figure>
&lt;/p>
&lt;h2 id="tasks">Tasks&lt;/h2>
&lt;p>First, we will further develop our existing grading algorithms for prostate and breast cancer, to enhance robustness to variance in data sources and disease (sub)types. Second, we will conduct studies to determine the most efficient way to integrate algorithms into the routine workflow. Last, we will evaluate the most promising workflow prospectively in pilot lines at both peripheral and academic centers to assess performance. Currently, there are no algorithms commercially available which can do pathologist-level grading of cancer. Thus, this project can have both significant societal and economic impact. From a societal perspective, we can make expert grading available at locations without access to subspecialized pathologists. Outside the Netherlands, there are pathologist shortages in countries such as India and China where these algorithms could have even more impact. Thirona is already ISO13485 certified and has experience with CE certification and FDA 510(k) clearance. As such, upon completion of this project, algorithms could be commercially available quickly, offering the potential for a significant disruption of the health care market.&lt;/p>
&lt;p>
&lt;figure id="figure-examples-of-prostate-biopsy-transurethral-resection-and-prostatectomy">
&lt;div class="d-flex justify-content-center">
&lt;div class="w-100" >&lt;img alt="Examples of prostate biopsy, transurethral resection, and prostatectomy" srcset="
/project/ppp-deepgrading/surgical_entities_hu7557648230985fa00b2bf2b28982a3d2_347217_4d8fcd77f163e543bc103b6be72a9316.webp 400w,
/project/ppp-deepgrading/surgical_entities_hu7557648230985fa00b2bf2b28982a3d2_347217_44de861b65b54ee2e4ad3bc38931da22.webp 760w,
/project/ppp-deepgrading/surgical_entities_hu7557648230985fa00b2bf2b28982a3d2_347217_1200x1200_fit_q75_h2_lanczos.webp 1200w"
src="https://geertlitjens.nl/project/ppp-deepgrading/surgical_entities_hu7557648230985fa00b2bf2b28982a3d2_347217_4d8fcd77f163e543bc103b6be72a9316.webp"
width="760"
height="248"
loading="lazy" data-zoomable />&lt;/div>
&lt;/div>&lt;figcaption>
Examples of prostate biopsy, transurethral resection, and prostatectomy
&lt;/figcaption>&lt;/figure>
&lt;/p></description></item><item><title>Artificial intelligence for diagnosis and Gleason grading of prostate cancer: the PANDA challenge</title><link>https://geertlitjens.nl/publication/bult-22/</link><pubDate>Sat, 01 Jan 2022 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/bult-22/</guid><description/></item><item><title>Using deep learning for quantification of cellularity and cell lineages in bone marrow biopsies and comparison to normal age-related variation.</title><link>https://geertlitjens.nl/publication/eeke-21/</link><pubDate>Mon, 01 Nov 2021 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/eeke-21/</guid><description/></item><item><title>Artificial Intelligence for Diagnosis and Gleason Grading of Prostate Cancer in Biopsies-Current Status and Next Steps.</title><link>https://geertlitjens.nl/publication/kart-21-a/</link><pubDate>Thu, 01 Jul 2021 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/kart-21-a/</guid><description/></item><item><title>Mini Review: The Last Mile-Opportunities and Challenges for Machine Learning in Digital Toxicologic Pathology.</title><link>https://geertlitjens.nl/publication/turn-21/</link><pubDate>Tue, 01 Jun 2021 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/turn-21/</guid><description/></item><item><title>The Medical Segmentation Decathlon</title><link>https://geertlitjens.nl/publication/anto-21/</link><pubDate>Tue, 01 Jun 2021 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/anto-21/</guid><description/></item><item><title>Deep learning in histopathology: the path to the clinic.</title><link>https://geertlitjens.nl/publication/laak-21/</link><pubDate>Sat, 01 May 2021 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/laak-21/</guid><description/></item><item><title>Residual cyclegan for robust domain transformation of histopathological tissue slides.</title><link>https://geertlitjens.nl/publication/bel-21/</link><pubDate>Sat, 01 May 2021 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/bel-21/</guid><description/></item><item><title>Common Limitations of Image Processing Metrics: A Picture Story</title><link>https://geertlitjens.nl/publication/rein-21-a/</link><pubDate>Thu, 01 Apr 2021 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/rein-21-a/</guid><description/></item><item><title>Optimized tumour infiltrating lymphocyte assessment for triple negative breast cancer prognostics.</title><link>https://geertlitjens.nl/publication/balk-21/</link><pubDate>Thu, 01 Apr 2021 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/balk-21/</guid><description/></item><item><title>Detection of prostate cancer in whole-slide images through end-to-end training with image-level labels.</title><link>https://geertlitjens.nl/publication/pinc-21/</link><pubDate>Mon, 01 Mar 2021 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/pinc-21/</guid><description/></item><item><title>Neural Image Compression for Gigapixel Histopathology Image Analysis.</title><link>https://geertlitjens.nl/publication/tell-21/</link><pubDate>Mon, 01 Feb 2021 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/tell-21/</guid><description/></item><item><title>Common limitations of performance metrics in biomedical image analysis</title><link>https://geertlitjens.nl/publication/rein-21/</link><pubDate>Fri, 01 Jan 2021 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/rein-21/</guid><description/></item><item><title>End-to-end classification on basal-cell carcinoma histopathology whole-slides images</title><link>https://geertlitjens.nl/publication/geij-21/</link><pubDate>Fri, 01 Jan 2021 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/geij-21/</guid><description/></item><item><title>Tailoring automated data augmentation to H&amp;E-stained histopathology</title><link>https://geertlitjens.nl/publication/fary-21/</link><pubDate>Fri, 01 Jan 2021 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/fary-21/</guid><description/></item><item><title>Impact of rescanning and normalization on convolutional neural network performance in multi-center, whole-slide classification of prostate cancer</title><link>https://geertlitjens.nl/publication/swid-20-b/</link><pubDate>Tue, 01 Sep 2020 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/swid-20-b/</guid><description/></item><item><title>Artificial Intelligence Assistance Significantly Improves Gleason Grading of Prostate Biopsies by Pathologists</title><link>https://geertlitjens.nl/publication/bult-20-a/</link><pubDate>Sat, 01 Aug 2020 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/bult-20-a/</guid><description/></item><item><title>Streaming convolutional neural networks for end-to-end learning with multi-megapixel images</title><link>https://geertlitjens.nl/publication/pinc-20/</link><pubDate>Sat, 01 Aug 2020 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/pinc-20/</guid><description/></item><item><title>The 2019 International Society of Urological Pathology (ISUP) Consensus Conference on Grading of Prostatic Carcinoma.</title><link>https://geertlitjens.nl/publication/leen-20/</link><pubDate>Sat, 01 Aug 2020 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/leen-20/</guid><description/></item><item><title>Interview with Kijk on AI in Medical Imaging</title><link>https://geertlitjens.nl/media/kijk-gleason/</link><pubDate>Fri, 01 May 2020 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/media/kijk-gleason/</guid><description>&lt;p>Interview with Wouter Bulten and myself on AI in medical imaging for popular science magazine Kijk, with a specific focus on our &lt;a href="https://geertlitjens.nl/publication/bult-20/">automated Gleason grading algorithm&lt;/a>.&lt;/p></description></item><item><title>Visit Dutch Prostate Cancer Patient Foundation</title><link>https://geertlitjens.nl/talk/visit-patient-foundation/</link><pubDate>Fri, 21 Feb 2020 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/talk/visit-patient-foundation/</guid><description>&lt;p>AI in medical imaging is often pitched as either an aid or a threat to the physician. But what is the impact for the patient? Can they benefit from the introduction of AI and if so, in which way? In this talk I will sketch the impact of AI on prostate cancer diagnosis from the perspective of the patient.&lt;/p></description></item><item><title>NRC article on artificial intelligence in medicine</title><link>https://geertlitjens.nl/media/nrc-gleason/</link><pubDate>Sat, 01 Feb 2020 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/media/nrc-gleason/</guid><description>&lt;p>Article covering some of the latest advances of AI in medical imaging, with specific focus on our &lt;a href="https://geertlitjens.nl/publication/bult-20/">JAMA Oncology&lt;/a> paper on automated Gleason grading (on the cover of the Science section).&lt;/p></description></item><item><title>BNR Wetenschap Vandaag - Interview AI for prostate cancer grading</title><link>https://geertlitjens.nl/media/bnr-interview-gleason/</link><pubDate>Thu, 09 Jan 2020 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/media/bnr-interview-gleason/</guid><description>&lt;p>Interview together with Wouter Bulten on our publication in JAMA Oncology on using AI for automated Gleason grading.&lt;/p>
&lt;div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;">
&lt;iframe src="https://www.youtube.com/embed/YnW2rUtjnNM" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" allowfullscreen title="YouTube Video">&lt;/iframe>
&lt;/div></description></item><item><title>NOS Tech Podcast on AI for Gleason Grading</title><link>https://geertlitjens.nl/media/nos-techpodcast-gleason/</link><pubDate>Thu, 09 Jan 2020 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/media/nos-techpodcast-gleason/</guid><description>&lt;p>Interview in the NOS op 3 tech podcast on AI for medical imaging&lt;/p>
&lt;div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;">
&lt;iframe src="https://www.youtube.com/embed/fYaJ8xZrEFI?start=857" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" allowfullscreen title="YouTube Video">&lt;/iframe>
&lt;/div></description></item><item><title>Artificial intelligence to detect MYC translocation in slides of diffuse large B-cell lymphoma</title><link>https://geertlitjens.nl/publication/swid-20-c/</link><pubDate>Wed, 01 Jan 2020 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/swid-20-c/</guid><description/></item><item><title>Automated deep-learning system for Gleason grading of prostate cancer using biopsies: a diagnostic study</title><link>https://geertlitjens.nl/publication/bult-20/</link><pubDate>Wed, 01 Jan 2020 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/bult-20/</guid><description/></item><item><title>Efficient Out-of-Distribution Detection in Digital Pathology Using Multi-Head Convolutional Neural Networks</title><link>https://geertlitjens.nl/publication/linm-20/</link><pubDate>Wed, 01 Jan 2020 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/linm-20/</guid><description/></item><item><title>Multi-class semantic cell segmentation and classification of aplasia in bone marrow histology images</title><link>https://geertlitjens.nl/publication/eeke-20-a/</link><pubDate>Wed, 01 Jan 2020 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/eeke-20-a/</guid><description/></item><item><title>Predicting MYC translocation in HE specimens of diffuse large B-cell lymphoma through deep learning</title><link>https://geertlitjens.nl/publication/swid-20/</link><pubDate>Wed, 01 Jan 2020 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/swid-20/</guid><description/></item><item><title>Artificial Intelligence in Prostate Cancer Diagnostics</title><link>https://geertlitjens.nl/talk/prostate-cancer-academy/</link><pubDate>Tue, 05 Nov 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/talk/prostate-cancer-academy/</guid><description>&lt;p>Artificial Intelligence is starting a more important role in diagnostics, also in diagnostics of prostate cancer. In this talk I will sketch the current applications of AI and give directions for future use, also with respect to clinical research.&lt;/p></description></item><item><title>Lymph node detection in MR Lymphography: false positive reduction using multi-view convolutional neural networks</title><link>https://geertlitjens.nl/publication/deba-19/</link><pubDate>Fri, 01 Nov 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/deba-19/</guid><description/></item><item><title>No pixel-level annotations needed</title><link>https://geertlitjens.nl/publication/laak-19/</link><pubDate>Thu, 17 Oct 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/laak-19/</guid><description/></item><item><title>Interview in 'De zomeravond van...' van Omroep P&amp;M</title><link>https://geertlitjens.nl/media/omroep_pm_zomeravond/</link><pubDate>Thu, 08 Aug 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/media/omroep_pm_zomeravond/</guid><description>&lt;p>Interview for the show &amp;lsquo;De zomeravond van&amp;hellip;&amp;rsquo; for Omroep P&amp;amp;M, which is an interview program for (former) residents of the municipality Peel &amp;amp; Maas. It partly highlights my work, but also some personal items.&lt;/p>
&lt;p>Above you can find the full link to the video on the Omroep P&amp;amp;M website (via the Video button). For convenience I have also added a link to YouTube, which you can see here:&lt;/p>
&lt;div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;">
&lt;iframe src="https://www.youtube.com/embed/seXpZq9v1EU" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" allowfullscreen title="YouTube Video">&lt;/iframe>
&lt;/div></description></item><item><title>Learning to detect lymphocytes in immunohistochemistry with deep learning</title><link>https://geertlitjens.nl/publication/swid-19/</link><pubDate>Thu, 01 Aug 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/swid-19/</guid><description/></item><item><title>Neural Image Compression for Gigapixel Histopathology Image Analysis</title><link>https://geertlitjens.nl/publication/tell-19/</link><pubDate>Thu, 01 Aug 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/tell-19/</guid><description/></item><item><title>Quantifying the effects of data augmentation and stain color normalization in convolutional neural networks for computational pathology.</title><link>https://geertlitjens.nl/publication/tell-19-a/</link><pubDate>Thu, 01 Aug 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/tell-19-a/</guid><description/></item><item><title>State-of-the-Art Deep Learning in Cardiovascular Image Analysis.</title><link>https://geertlitjens.nl/publication/litj-19/</link><pubDate>Thu, 01 Aug 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-19/</guid><description/></item><item><title>Dealing with Label Scarcity in Computational Pathology: A Use Case in Prostate Cancer Classification</title><link>https://geertlitjens.nl/publication/derck-19/</link><pubDate>Sat, 01 Jun 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/derck-19/</guid><description/></item><item><title>Stain-Transforming Cycle-Consistent Generative Adversarial Networks for Improved Segmentation of Renal Histopathology</title><link>https://geertlitjens.nl/publication/bel-19/</link><pubDate>Sat, 01 Jun 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/bel-19/</guid><description/></item><item><title>Applications of Machine Learning for Clinical Practice</title><link>https://geertlitjens.nl/talk/nordic-digital-pathology-symposium/</link><pubDate>Thu, 16 May 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/talk/nordic-digital-pathology-symposium/</guid><description>&lt;p>Clinical pathology is at the forefront of a digital revolution. In addition to the digital workflow, pathologists will also come into contact with machine learning algorithms aimed at improving their diagnostic accuracy and efficiency. In this presentation I highlight some applications which will be among the first to see use in clinical practice.&lt;/p></description></item><item><title>Report on Omroep Gelderland on 'De Nieuwe Mens'</title><link>https://geertlitjens.nl/media/omroep_gelderland_interview/</link><pubDate>Mon, 06 May 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/media/omroep_gelderland_interview/</guid><description>&lt;p>Above you can find the full link to the article and the video on the Omroep Gelderland website (via the Video button). For convenience I have also uploaded the video to YouTube, which you can see here:&lt;/p>
&lt;p>
&lt;div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;">
&lt;iframe src="https://www.youtube.com/embed/3dt-E1vBr94" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" allowfullscreen title="YouTube Video">&lt;/iframe>
&lt;/div>
&lt;br>
In addition, a longer podcast was made by Vera Eisink from Omroep Gelderland which covers our research in a bit more detail. It is hosted on SoundCloud and can be listened here:&lt;/p>
&lt;iframe width="100%" height="300" scrolling="no" frameborder="no" allow="autoplay" src="https://w.soundcloud.com/player/?url=https%3A//api.soundcloud.com/tracks/614820420&amp;color=%23ff5500&amp;auto_play=false&amp;hide_related=false&amp;show_comments=true&amp;show_user=true&amp;show_reposts=false&amp;show_teaser=true&amp;visual=true">&lt;/iframe></description></item><item><title>A Single-Arm, Multicenter Validation Study of Prostate Cancer Localization and Aggressiveness With a Quantitative Multiparametric Magnetic Resonance Imaging Approach.</title><link>https://geertlitjens.nl/publication/maas-19/</link><pubDate>Wed, 01 May 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/maas-19/</guid><description/></item><item><title>Getting Started With Camelyon (Part 1)</title><link>https://geertlitjens.nl/post/getting-started-with-camelyon/</link><pubDate>Wed, 24 Apr 2019 13:41:43 +0200</pubDate><guid>https://geertlitjens.nl/post/getting-started-with-camelyon/</guid><description>&lt;p>This is the first part of a three part tutorial on how to get started with the &lt;a href="https://camelyon17.grand-challenge.org" target="_blank" rel="noopener">CAMELYON dataset&lt;/a>. This first part will focus on getting a basic convolutional neural network trained using &lt;a href="https://github.com/basveeling/pcam" target="_blank" rel="noopener">PatchCAMELYON&lt;/a>, TensorFlow 2.0, Keras and TensorFlow Datasets. Part 2 will cover applying your trained model to a whole-slide image and visualizing the results and Part 3 will cover how to use the full dataset to train a model at different resolution levels, sampling strategies, and data augmentation.&lt;/p>
&lt;p>To get started you need to setup a Python environment with NumPy, Matplotlib and TensorFlow 2.0. To use the PatchCAMELYON dataset with TensorFlow Datasets you will need to use my fork of the project for now as the pull request to add PatchCAMELYON to the master branch is not yet approved. To this end you need to do clone the repository and add it to your Python environment:&lt;/p>
&lt;pre>&lt;code class="language-bash">git clone https://github.com/GeertLitjens/tensorflow_datasets
cd tensorflow_datasets
python setup.py develop
&lt;/code>&lt;/pre>
&lt;p>After this step you should be able to import the relevant packages with the following cell&lt;/p>
&lt;pre>&lt;code class="language-python"># Import NumPy to handle array's and Matplotlib for plotting loss curves
import numpy as np
import matplotlib.pyplot as plt
# Import TensorFlow and relevant Keras classes to setup the model
import tensorflow as tf
from tensorflow.keras.layers import Input, Dense, Conv2D, MaxPool2D, Flatten, Dropout
from tensorflow.keras.models import Model
from tensorflow.keras.optimizers import SGD
from tensorflow.keras.utils import to_categorical
from tensorflow.keras.callbacks import ModelCheckpoint
&lt;/code>&lt;/pre>
&lt;p>The next cell will automatically download PatchCAMELYON from Zenodo and prepare the TensorFlow Datasets&lt;/p>
&lt;pre>&lt;code class="language-python">import tensorflow_datasets as tfds
pcam, pcam_info = tfds.load(&amp;quot;patch_camelyon&amp;quot;, with_info=True)
print(pcam_info)
&lt;/code>&lt;/pre>
&lt;blockquote>
&lt;pre>&lt;code class="language-python">&lt;/code>&lt;/pre>
&lt;/blockquote>
&lt;pre>&lt;code>tfds.core.DatasetInfo(
name='patch_camelyon',
version=1.0.0,
description='The PatchCAMELYON dataset for identification of breast cancer metastases in lymph nodes. This dataset has been extracted from the larger CAMELYON dataset of 1399 whole-slide images, which created for the CAMELYON challenges at ISBI 2016 and 2017.It contains 96x96 RGB patches of normal lymph node and tumor tissue in a roughly 50/50 distributions. It packs the clinically-relevant task of metastasis detection into a straight-forward image classification task, akin to CIFAR-10 and MNIST. This increases the ease of use by removing the complexity of handling large whole-slide images.',
urls=['https://github.com/basveeling/pcam', 'https://camelyon17.grand-challenge.org/'],
features=FeaturesDict({
'image': Image(shape=(96, 96, 3), dtype=tf.uint8),
'label': ClassLabel(shape=(), dtype=tf.int64, num_classes=2)
},
total_num_examples=327680,
splits={
'test': &amp;lt;tfds.core.SplitInfo num_examples=32768&amp;gt;,
'train': &amp;lt;tfds.core.SplitInfo num_examples=262144&amp;gt;,
'validation': &amp;lt;tfds.core.SplitInfo num_examples=32768&amp;gt;
},
supervised_keys=('image', 'label'),
citation='&amp;quot;&amp;quot;&amp;quot;
@ARTICLE{Veeling2018-qh,
title = &amp;quot;Rotation Equivariant {CNNs} for Digital Pathology&amp;quot;,
author = &amp;quot;Veeling, Bastiaan S and Linmans, Jasper and Winkens, Jim and
Cohen, Taco and Welling, Max&amp;quot;,
month = jun,
year = 2018,
archivePrefix = &amp;quot;arXiv&amp;quot;,
primaryClass = &amp;quot;cs.CV&amp;quot;,
eprint = &amp;quot;1806.03962&amp;quot;
}
@article{Litjens2018,
author = {Litjens, G. and Bándi, P. and Ehteshami Bejnordi, B. and Geessink, O. and Balkenhol, M. and Bult, P. and Halilovic, A. and Hermsen, M. and van de Loo, R. and Vogels, R. and Manson, Q.F. and Stathonikos, N. and Baidoshvili, A. and van Diest, P. and Wauters, C. and van Dijk, M. and van der Laak, J.},
title = {1399 H&amp;amp;E-stained sentinel lymph node sections of breast cancer patients: the CAMELYON dataset},
journal = {GigaScience},
volume = {7},
number = {6},
year = {2018},
month = {05},
issn = {2047-217X},
doi = {10.1093/gigascience/giy065},
url = {https://dx.doi.org/10.1093/gigascience/giy065},
eprint = {http://oup.prod.sis.lan/gigascience/article-pdf/7/6/giy065/25045131/giy065.pdf},
}
&amp;quot;&amp;quot;&amp;quot;',
redistribution_info=,
)
&lt;/code>&lt;/pre>
&lt;pre>&lt;code>
Now we have our dataset ready, it is time to define our model. The cell below defines a very simple VGG-like convolutional neural network using Keras.
```python
#First setup the input to the network which has the dimensions of the patches contained within PatchCAMELYON
input_img = Input(shape=(96,96,3))
# Now we define the layers of the convolutional network: three blocks of two convolutional layers and a max-pool layer.
x = Conv2D(16, (3, 3), padding='valid', activation='relu')(input_img)
x = Conv2D(16, (3, 3), padding='valid', activation='relu')(x)
x = MaxPool2D(pool_size=(2,2), strides=(2,2))(x)
x = Conv2D(32, (3, 3), padding='valid', activation='relu')(x)
x = Conv2D(32, (3, 3), padding='valid', activation='relu')(x)
x = MaxPool2D(pool_size=(2,2), strides=(2,2))(x)
x = Conv2D(64, (3, 3), padding='valid', activation='relu')(x)
x = Conv2D(64, (3, 3), padding='valid', activation='relu')(x)
x = MaxPool2D(pool_size=(2,2), strides=(2,2))(x)
# Now we flatten the output from a 4D to a 2D tensor to be able to use fully-connected (dense) layers for the final
# classification part. Here we also use a bit of dropout for regularization. The last layer uses a softmax to obtain class
# likelihoods (i.e. metastasis vs. non-metastasis)
x = Flatten()(x)
x = Dense(256, activation='relu')(x)
x = Dropout(rate=0.2)(x)
x = Dense(128, activation='relu')(x)
x = Dropout(rate=0.2)(x)
predictions = Dense(2, activation='softmax')(x)
# Now we define the inputs/outputs of the model and setup the optimizer. In this case we use regular stochastic gradient
# descent with Nesterov momentum. The loss we use is cross-entropy and we would like to output accuracy as an additional metric.
model = Model(inputs=input_img, outputs=predictions)
sgd_opt = SGD(lr=0.01, momentum=0.9, decay=0.0, nesterov=True)
model.compile(optimizer=sgd_opt,
loss='categorical_crossentropy',
metrics=['accuracy'])
model.summary()
&lt;/code>&lt;/pre>
&lt;blockquote>
&lt;pre>&lt;code class="language-text">&lt;/code>&lt;/pre>
&lt;/blockquote>
&lt;p>Model: &amp;ldquo;model&amp;rdquo;&lt;/p>
&lt;hr>
&lt;h1 id="layer-type-----------------output-shape--------------param-">Layer (type) Output Shape Param #&lt;/h1>
&lt;p>input_1 (InputLayer) [(None, 96, 96, 3)] 0&lt;/p>
&lt;hr>
&lt;p>conv2d (Conv2D) (None, 94, 94, 16) 448&lt;/p>
&lt;hr>
&lt;p>conv2d_1 (Conv2D) (None, 92, 92, 16) 2320&lt;/p>
&lt;hr>
&lt;p>max_pooling2d (MaxPooling2D) (None, 46, 46, 16) 0&lt;/p>
&lt;hr>
&lt;p>conv2d_2 (Conv2D) (None, 44, 44, 32) 4640&lt;/p>
&lt;hr>
&lt;p>conv2d_3 (Conv2D) (None, 42, 42, 32) 9248&lt;/p>
&lt;hr>
&lt;p>max_pooling2d_1 (MaxPooling2 (None, 21, 21, 32) 0&lt;/p>
&lt;hr>
&lt;p>conv2d_4 (Conv2D) (None, 19, 19, 64) 18496&lt;/p>
&lt;hr>
&lt;p>conv2d_5 (Conv2D) (None, 17, 17, 64) 36928&lt;/p>
&lt;hr>
&lt;p>max_pooling2d_2 (MaxPooling2 (None, 8, 8, 64) 0&lt;/p>
&lt;hr>
&lt;p>flatten (Flatten) (None, 4096) 0&lt;/p>
&lt;hr>
&lt;p>dense (Dense) (None, 256) 1048832&lt;/p>
&lt;hr>
&lt;p>dropout (Dropout) (None, 256) 0&lt;/p>
&lt;hr>
&lt;p>dense_1 (Dense) (None, 128) 32896&lt;/p>
&lt;hr>
&lt;p>dropout_1 (Dropout) (None, 128) 0&lt;/p>
&lt;hr>
&lt;h1 id="dense_2-dense--------------none-2-----------------258">dense_2 (Dense) (None, 2) 258&lt;/h1>
&lt;p>Total params: 1,154,066
Trainable params: 1,154,066
Non-trainable params: 0&lt;/p>
&lt;hr>
&lt;pre>&lt;code>
To keep the dataset size small PatchCAMELYON is stored as int8 patches. For network training we need float32 and we want to normalize between 0 and 1. The function below performs this task.
```python
def convert_sample(sample):
image, label = sample['image'], sample['label']
image = tf.image.convert_image_dtype(image, tf.float32)
label = tf.one_hot(label, 2, dtype=tf.float32)
return image, label
&lt;/code>&lt;/pre>
&lt;p>Now we use the &lt;code>tf.data&lt;/code> pipeline to apply this function to the dataset in a parallel fashion. We also shuffle the training data with a shuffle buffer (which is randomly filled with samples from the dataset) of 1024. Next we define batches of 64 patches for training and 128 for validation. Last, we prefetch 2 batches such that we can get batches during training on the GPU.&lt;/p>
&lt;pre>&lt;code class="language-python">train_pipeline = pcam['train'].map(convert_sample,
num_parallel_calls=8).shuffle(1024).repeat().batch(64).prefetch(2)
valid_pipeline = pcam['validation'].map(convert_sample,
num_parallel_calls=8).repeat().batch(128).prefetch(2)
&lt;/code>&lt;/pre>
&lt;p>Now we just apply train and evaluate the model using our dataset pipeline. We pick the steps per epoch such that the entire training and validation set are covered each epoch. We keep the History object that &lt;code>fit&lt;/code> returns to plot the loss progression later on. We now just do 5 epochs for illustration purposes. Feel free to experiment with the number of epochs. If you want to keep the best model during training you can use the Keras &lt;code>ModelCheckpoint&lt;/code> callback to write each improvement to disk.&lt;/p>
&lt;pre>&lt;code class="language-python">hist = model.fit(train_pipeline,
validation_data=valid_pipeline,
verbose=2, epochs=5, steps_per_epoch=4096, validation_steps=256)
&lt;/code>&lt;/pre>
&lt;blockquote>
&lt;pre>&lt;code class="language-text">&lt;/code>&lt;/pre>
&lt;/blockquote>
&lt;p>Epoch 1/5
4096/4096 - 101s - loss: 0.6756 - accuracy: 0.5501 - val_loss: 0.5228 - val_accuracy: 0.7527
Epoch 2/5
4096/4096 - 98s - loss: 0.4292 - accuracy: 0.8071 - val_loss: 0.3946 - val_accuracy: 0.8249
Epoch 3/5
4096/4096 - 99s - loss: 0.3165 - accuracy: 0.8675 - val_loss: 0.3634 - val_accuracy: 0.8390
Epoch 4/5
4096/4096 - 100s - loss: 0.2586 - accuracy: 0.8958 - val_loss: 0.3707 - val_accuracy: 0.8340
Epoch 5/5
4096/4096 - 99s - loss: 0.2260 - accuracy: 0.9118 - val_loss: 0.3353 - val_accuracy: 0.8568&lt;/p>
&lt;pre>&lt;code>
If we are happy with our performance on the validation set we can check whether it generalized to the test set. Note that it is bad practice to look at your test set performance too often; you will start making modification to your network/training procedure to optimize test set performance which results in optimistically biased performance estimates.
```python
test_pipeline = pcam['test'].map(convert_sample, num_parallel_calls=8).batch(128).prefetch(2)
print(&amp;quot;Test set accuracy is {0:.4f}&amp;quot;.format(model.evaluate(test_pipeline, steps=128, verbose=0)[1]))
&lt;/code>&lt;/pre>
&lt;blockquote>
&lt;pre>&lt;code class="language-text">&lt;/code>&lt;/pre>
&lt;/blockquote>
&lt;p>Test set accuracy is 0.8563&lt;/p>
&lt;pre>&lt;code>
If we are happy with the performance we can write the model to disk.
```python
model.save(&amp;quot;./patchcamelyon.hf5&amp;quot;)
&lt;/code>&lt;/pre>
&lt;p>If you followed along with all the steps you should get a test set performance between 0.8 and 0.87. Differences occur due to different weight initialization of the network for example. To make network training more reproducible you can specify the random seed for the weight initializers manually. Note that the training process cannot be fully deterministic due to backpropogation step in the CUDNN library not being deterministic. I hope you enjoyed this brief tutorial, the next post will be about how to use our saved model to classify a whole-slide image.&lt;/p></description></item><item><title>CAMELYON for Elementary School</title><link>https://geertlitjens.nl/post/ml-for-elementary-school/</link><pubDate>Fri, 19 Apr 2019 13:41:14 +0200</pubDate><guid>https://geertlitjens.nl/post/ml-for-elementary-school/</guid><description>&lt;p>In 2018 &lt;a href="https://www.computationalpathologygroup.eu/members/jeroen-van-der-laak/" target="_blank" rel="noopener">Jeroen van der Laak&lt;/a> and I were nominated and eventually won a Radboud Science Award for our work with the CAMELYON challenge. One part of the prize was the oppertunity to turn our research into an educational program for elementary schools, specifically ages 9 till 12. This post is a summary of my experience going through this process and showcases the things we developed. Luckily, we were not in this alone and we got great support from the &lt;a href="https://www.ru.nl/wetenschapsknooppunt/" target="_blank" rel="noopener">Radboud Wetenschapsknooppunt&lt;/a> and two of our PhD students: &lt;a href="https://www.computationalpathologygroup.eu/members/meyke-hermsen/" target="_blank" rel="noopener">Meyke Hermsen&lt;/a> and &lt;a href="https://www.computationalpathologygroup.eu/members/maschenka-balkenhol/" target="_blank" rel="noopener">Maschenka Balkenhol&lt;/a>. In addition, the kids from the elemantary schools also made a great introductory for us:&lt;/p>
&lt;div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;">
&lt;iframe src="https://www.youtube.com/embed/UHsXJe_fXgY" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" allowfullscreen title="YouTube Video">&lt;/iframe>
&lt;/div>
&lt;p>An &amp;lsquo;educational program&amp;rsquo; might sound a bit vague, so I&amp;rsquo;ll dive into some specifics. The idea was to develop in total six activities which the children could do under supervision of their teachers (and partly by themselves) which would guide them through the background and the different steps of our research. These activities would take place over the course of several weeks in which there would be lessons to prepare, execute, present, and provide feedback on the activities. In the end the children need to define their own research questions in relation to the activities and execute it. The biggest challenge was to build these activities in such a way that they stay true to our research results but are also understandable for children of varying ages.&lt;/p>
&lt;p>We quickly decided to split our research into two separate components with separate activities which would come together in the final activity. The two components were: histopathological diagnostics and artificial intelligence. In the end we came up with these six activities:&lt;/p>
&lt;ol>
&lt;li>When are computers intelligent?&lt;/li>
&lt;li>What are applications of artifical intelligence you encounter?&lt;/li>
&lt;li>How do you train a smart computer?&lt;/li>
&lt;li>What is a diagnosis and how do you perform one?&lt;/li>
&lt;li>How does a pathologist do diagnoses?&lt;/li>
&lt;li>Diagnosing cancer with artificially intelligent computers.&lt;/li>
&lt;/ol>
&lt;p>The fist two activities were aimed at getting a discussion going in the classroom on what the definition of intelligence is. Think along the lines of: Is a calculator intelligent? Or a navigation system? Secondly, the children were asked to discuss this at home and find examples of devices in their own house which they think were intelligent. In the end we would provide them with what we, within this project, mean with artificial intelligence: a computer program which automatically learns by example and can generalize what is learned this to unknown situations; similarly to the way they learn new skills.&lt;/p>
&lt;p>The third activity was the first real hands-on activity for which I adapted the excellent webapp &lt;a href="https://teachablemachine.withgoogle.com/" target="_blank" rel="noopener">Teachable Machine&lt;/a> by Google. I translated the app to Dutch and made some fixes for tablets and phones so it would be easier to use in the classroom. My adaptation can be found &lt;a href="https://geertlitjens.github.io/teachable-machine/" target="_blank" rel="noopener">here&lt;/a>, with the source code &lt;a href="https://github.com/GeertLitjens/teachable-machine" target="_blank" rel="noopener">here&lt;/a>. This webapp allows you to train a three-class classifier with your webcam. One fun application is to turn hand-gestures into instruments, as depicted in the video below.&lt;/p>
&lt;div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;">
&lt;iframe src="https://www.youtube.com/embed/oP8-_0ZyY3U" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" allowfullscreen title="YouTube Video">&lt;/iframe>
&lt;/div>
&lt;p>The idea of this activity was that children can figure out what a computer can easily learn and what is more difficult. For example, if you teach it to react to your face, will it also react to your friend&amp;rsquo;s face? Can it discriminate between your left and right hand? And if it doesn&amp;rsquo;t, what do you need to do to make it learn the difference. Many challenges that actually appear in machine learning, like bias, can be explored this way in a playful manner.&lt;/p>
&lt;p>The next two activities completely moved away from artifical intelligence and focused on a key job of a doctor: obtaining a diagnosis. In activity 4 we developed several disease scenarios and the children had to identify what questions you need to ask to figure out what disease the person has. Specifically, we had a scenario where a child was either suffering from food poisoning or the flu. This activity intends to show how the process of a diagnosis works and how, by asking the right questions, you can narrow down your options and eventually figure out what ails the patient.&lt;/p>
&lt;p>Activity 5 move to the domain of histopathology. Here we introduced the microscope and concept of &amp;lsquo;good&amp;rsquo; cells and &amp;lsquo;bad&amp;rsquo; cells (cancer). We prepared a lot of images (one example below) to show the children the differences in appearance and how a pathologist, with a microscope, can see what is wrong with a patient. We also gave them a couple of images were they had to figure out themselves what were the good and bad cells.&lt;/p>
&lt;p>The last activity combined artificial intelligence and histopathological diagnostics. Here again we used a &lt;a href="https://geertlitjens.github.io/metastaticcellclassifier/index.html" target="_blank" rel="noopener">webapp&lt;/a> (source &lt;a href="https://github.com/GeertLitjens/metastaticcellclassifier" target="_blank" rel="noopener">here&lt;/a>) which I made based of the &lt;a href="https://github.com/brendansudol/faces" target="_blank" rel="noopener">face classifier&lt;/a> by Brendan Sudol. Here the students could upload images with cells and the computer would tell them whether they were good or bad. Initially they had to classify these images themselves and then see if the computer agreed. Additionally, they could try to find cases were they think the computer was wrong and see if a pattern could be established. This way they could interactively explore the limitations of this simple &amp;lsquo;smart&amp;rsquo; computer system.&lt;/p>
&lt;figure id="figure-example-of-the-webapp-for-cell-classification">
&lt;div class="d-flex justify-content-center">
&lt;div class="w-100" >&lt;img alt="Example of the webapp for cell classification." srcset="
/post/ml-for-elementary-school/featured_hue74fcaf7eec50e399212e8fcf24aaef6_643294_5ea18d715725ca992d5d11c9f0b761b1.webp 400w,
/post/ml-for-elementary-school/featured_hue74fcaf7eec50e399212e8fcf24aaef6_643294_20a25ea0ac7af6ab8826b5f5aed3093b.webp 760w,
/post/ml-for-elementary-school/featured_hue74fcaf7eec50e399212e8fcf24aaef6_643294_1200x1200_fit_q75_h2_lanczos_3.webp 1200w"
src="https://geertlitjens.nl/post/ml-for-elementary-school/featured_hue74fcaf7eec50e399212e8fcf24aaef6_643294_5ea18d715725ca992d5d11c9f0b761b1.webp"
width="50%"
height="645"
loading="lazy" data-zoomable />&lt;/div>
&lt;/div>&lt;figcaption>
Example of the webapp for cell classification.
&lt;/figcaption>&lt;/figure>
&lt;p>After the activities were finished a group of teacher at elementary school &lt;a href="https://www.po.deltascholen.org/BS-de-Gazelle/" target="_blank" rel="noopener">&amp;lsquo;de Gazelle&amp;rsquo;&lt;/a> in Arnhem tested the lessons at their school. We visited roughly halfway and gave a presentation after which the kids had the oppertunity to ask questions. I had a lot of fun and got great feedback, both from the teachers and the children.&lt;/p>
&lt;p>Currently, we are in the phase of wrapping up the project. Our package of lessions and activities will be bundled in a book such that other schools can also use it. This book will be available through the &lt;a href="https://www.ru.nl/wetenschapsknooppunt/" target="_blank" rel="noopener">Radboud Wetenschapsknooppunt&lt;/a>. Last, we will give a presentation about this project at the Winterschool organized by the Wetenschapsknooppunt for teachers to share our experience.&lt;/p>
&lt;p>All in all, it was great to see the enthusiasm for our research in both the teachers and the children and for me it was a very rewarding experience. If you are interested in any of the materials or the project, don&amp;rsquo;t hesitate to reach out to me.&lt;/p></description></item><item><title>Robust and accurate quantification of biomarkers of immune cells in lung cancer micro-environment using deep convolutional neural networks.</title><link>https://geertlitjens.nl/publication/apru-19/</link><pubDate>Mon, 08 Apr 2019 13:58:34 +0200</pubDate><guid>https://geertlitjens.nl/publication/apru-19/</guid><description/></item><item><title>Automated Gleason Grading of Prostate Biopsies Using Deep Learning</title><link>https://geertlitjens.nl/publication/bult-19-a/</link><pubDate>Mon, 18 Mar 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/bult-19-a/</guid><description/></item><item><title>A large annotated medical image dataset for the development and evaluation of segmentation algorithms</title><link>https://geertlitjens.nl/publication/simp-19/</link><pubDate>Tue, 01 Jan 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/simp-19/</guid><description/></item><item><title>Computer aided quantification of intratumoral stroma yields an independent prognosticator in rectal cancer</title><link>https://geertlitjens.nl/publication/gees-19/</link><pubDate>Tue, 01 Jan 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/gees-19/</guid><description/></item><item><title>CV</title><link>https://geertlitjens.nl/cv/</link><pubDate>Tue, 01 Jan 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/cv/</guid><description/></item><item><title>Epithelium segmentation using deep learning in H&amp;E-stained prostate specimens with immunohistochemistry as reference standard</title><link>https://geertlitjens.nl/publication/bult-19/</link><pubDate>Tue, 01 Jan 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/bult-19/</guid><description/></item><item><title>High resolution whole prostate biopsy classification using streaming stochastic gradient descent</title><link>https://geertlitjens.nl/publication/pinc-19/</link><pubDate>Tue, 01 Jan 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/pinc-19/</guid><description/></item><item><title>Resolution-agnostic tissue segmentation in whole-slide histopathology images with convolutional neural networks</title><link>https://geertlitjens.nl/publication/band-19-a/</link><pubDate>Tue, 01 Jan 2019 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/band-19-a/</guid><description/></item><item><title>Introduction to Deep Learning in Medical Imaging</title><link>https://geertlitjens.nl/talk/intro-deep-learning/</link><pubDate>Sun, 16 Sep 2018 09:30:00 +0200</pubDate><guid>https://geertlitjens.nl/talk/intro-deep-learning/</guid><description>
&lt;div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;">
&lt;iframe src="https://www.youtube.com/embed/UYe6fn-P6_s" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" allowfullscreen title="YouTube Video">&lt;/iframe>
&lt;/div>
&lt;p>This presentation was the first part of a half-a-day workshop on deep learning in medical imaging. It introduces the basic deep learning concepts, contrasts them to more traditional pattern recognition approaches, and shows some examples from the field. If you are interested in a more thorough overview of different applications, I can recommend &lt;a href="https://geertlitjens.nl/publication/litj-17/">this&lt;/a> publication.&lt;/p></description></item><item><title>From detection of individual metastases to classification of lymph node status at the patient level: the CAMELYON17 challenge</title><link>https://geertlitjens.nl/publication/band-18/</link><pubDate>Wed, 01 Aug 2018 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/band-18/</guid><description/></item><item><title>BNR Wetenschap Vandaag - Zomercollege</title><link>https://geertlitjens.nl/media/wetenschapvandaag-zomercollege/</link><pubDate>Wed, 18 Jul 2018 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/media/wetenschapvandaag-zomercollege/</guid><description>&lt;p>For a special series of Summer Mini-Lectures Jeroen van der Laak and I were interviewed by Karlijn Meinders on AI in diagnostic pathology.
&lt;div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;">
&lt;iframe src="https://www.youtube.com/embed/sKz4mGxbSxw" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" allowfullscreen title="YouTube Video">&lt;/iframe>
&lt;/div>
&lt;/p></description></item><item><title>Bessensap</title><link>https://geertlitjens.nl/talk/bessensap/</link><pubDate>Fri, 15 Jun 2018 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/talk/bessensap/</guid><description>&lt;p>Should doctors fear for their jobs due to the rise of AI? In this presentation I explained which aspects of their job will change and why they should not fear, but welcome the introduction of AI in healthcare.&lt;/p></description></item><item><title>BNR Beter - Interview</title><link>https://geertlitjens.nl/media/beter-interview/</link><pubDate>Mon, 04 Jun 2018 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/media/beter-interview/</guid><description>&lt;p>Interview with BNR Beter on the digitization of pathology and the role of artificial intelligence in the workflow of the pathologist. Together with prof. Paul van Diest.
&lt;div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;">
&lt;iframe src="https://www.youtube.com/embed/nh5ruSNs37Q" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" allowfullscreen title="YouTube Video">&lt;/iframe>
&lt;/div>
&lt;/p></description></item><item><title>1399 H&amp;E-stained sentinel lymph node sections of breast cancer patients: the CAMELYON dataset</title><link>https://geertlitjens.nl/publication/litj-18/</link><pubDate>Fri, 01 Jun 2018 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-18/</guid><description/></item><item><title>Dag van de Pathologie</title><link>https://geertlitjens.nl/talk/dag-van-de-pathologie/</link><pubDate>Fri, 13 Apr 2018 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/talk/dag-van-de-pathologie/</guid><description>&lt;p>Clinical pathology as at the forefront of a digital revolution. In addition to the digital workflow, pathologists will also come into contact with machine learning algorithms aimed at improving their diagnostic accuracy and efficiency. In this presentation I highlight some applications which will be among the first to see use in clinical practice.&lt;/p>
&lt;p>This presentation showed applications of deep learning for diagnostic histopathology. The slides are shown below. Note that some slides are a bit scrambled due to the conversion from PowerPoint to Google Slides.&lt;/p>
&lt;div class="responsive-wrap">
&lt;iframe src="https://docs.google.com/presentation/d/e/2PACX-1vShDh2ln7_BMyIhnBJzjdLw71yzr9ojssfkyiuddkjpg7sZpN_6dbpw2JTEtJnPEMpWJf1zJUA6eeGR/embed?start=false&amp;amp;loop=false&amp;amp;delayms=3000" frameborder="0" width="960" height="569" allowfullscreen="true" mozallowfullscreen="true" webkitallowfullscreen="true">&lt;/iframe>
&lt;/div></description></item><item><title>Interview with Prostate Cancer Patient Foundation</title><link>https://geertlitjens.nl/media/prostaatkankerstichting/</link><pubDate>Thu, 01 Mar 2018 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/media/prostaatkankerstichting/</guid><description>&lt;p>Interview with the Foundation which supports prostate cancer patients in the Netherlands. They supported my Bas Mulder Award application and were interested in my research. This interview gives a good overview of the goals of the project for a general audience. Note that the interview is in Dutch.&lt;/p></description></item><item><title>Whole-Slide Mitosis Detection in H&amp;E Breast Histology Using PHH3 as a Reference to Train Distilled Stain-Invariant Convolutional Networks</title><link>https://geertlitjens.nl/publication/tell-18/</link><pubDate>Thu, 01 Mar 2018 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/tell-18/</guid><description/></item><item><title>Automated segmentation of epithelial tissue in prostatectomy slides using deep learning</title><link>https://geertlitjens.nl/publication/bult-18/</link><pubDate>Thu, 01 Feb 2018 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/bult-18/</guid><description/></item><item><title>Automatic color unmixing of IHC stained whole slide images</title><link>https://geertlitjens.nl/publication/geij-18/</link><pubDate>Thu, 01 Feb 2018 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/geij-18/</guid><description/></item><item><title>H&amp;E stain augmentation improves generalization of convolutional networks for histopathological mitosis detection</title><link>https://geertlitjens.nl/publication/tell-18-a/</link><pubDate>Thu, 01 Feb 2018 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/tell-18-a/</guid><description/></item><item><title>Convolutional Neural Networks for Lymphocyte detection in Immunohistochemically Stained Whole-Slide Images</title><link>https://geertlitjens.nl/publication/swid-18/</link><pubDate>Mon, 01 Jan 2018 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/swid-18/</guid><description/></item><item><title>Training convolutional neural networks with megapixel images</title><link>https://geertlitjens.nl/publication/pinc-18/</link><pubDate>Mon, 01 Jan 2018 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/pinc-18/</guid><description/></item><item><title>Unsupervised Prostate Cancer Detection on H&amp;E using Convolutional Adversarial Autoencoders</title><link>https://geertlitjens.nl/publication/bult-18-a/</link><pubDate>Mon, 01 Jan 2018 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/bult-18-a/</guid><description/></item><item><title>Report on Nieuwsuur about CAMELYON.</title><link>https://geertlitjens.nl/media/nieuwsuur_camelyon/</link><pubDate>Wed, 13 Dec 2017 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/media/nieuwsuur_camelyon/</guid><description>&lt;p>Report by Nieuwsuur on the results of our &lt;a href="https://geertlitjens.nl/publication/ehte-17/">JAMA publication&lt;/a> which showed that AI could perform at the level of expert pathologists.
&lt;div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;">
&lt;iframe src="https://www.youtube.com/embed/yLlk89EAgUU" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" allowfullscreen title="YouTube Video">&lt;/iframe>
&lt;/div>
&lt;/p></description></item><item><title>Diagnostic Assessment of Deep Learning Algorithms for Detection of Lymph Node Metastases in Women With Breast Cancer</title><link>https://geertlitjens.nl/publication/ehte-17/</link><pubDate>Fri, 01 Dec 2017 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/ehte-17/</guid><description/></item><item><title>Context-aware stacked convolutional neural networks for classification of breast carcinomas in whole-slide histopathology images</title><link>https://geertlitjens.nl/publication/bejn-17-b/</link><pubDate>Sun, 01 Oct 2017 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/bejn-17-b/</guid><description/></item><item><title>The Digital Doctor</title><link>https://geertlitjens.nl/talk/kindergeneeskunde/</link><pubDate>Mon, 12 Jun 2017 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/talk/kindergeneeskunde/</guid><description>&lt;p>How will AI impact medicine, and specifically, pediatric medicine? These questions I will try to answer during this presentation with specific examples. Specifically, I will also address the impact in low-resource countries.&lt;/p></description></item><item><title>Evaluation of tongue squamous cell carcinoma resection margins using ex-vivo MR.</title><link>https://geertlitjens.nl/publication/stee-17/</link><pubDate>Mon, 01 May 2017 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/stee-17/</guid><description/></item><item><title>Comparison of Different Methods for Tissue Segmentation In Histopathological Whole-Slide Images</title><link>https://geertlitjens.nl/publication/band-17/</link><pubDate>Sat, 01 Apr 2017 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/band-17/</guid><description/></item><item><title>The importance of stain normalization in colorectal tissue classification with convolutional networks</title><link>https://geertlitjens.nl/publication/ciom-17/</link><pubDate>Sat, 01 Apr 2017 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/ciom-17/</guid><description/></item><item><title>Using deep learning to segment breast and fibroglandular tissue in MRI volumes</title><link>https://geertlitjens.nl/publication/dalm-17/</link><pubDate>Wed, 01 Feb 2017 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/dalm-17/</guid><description/></item><item><title>A Survey on Deep Learning in Medical Image Analysis</title><link>https://geertlitjens.nl/publication/litj-17/</link><pubDate>Sun, 01 Jan 2017 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-17/</guid><description/></item><item><title>Large scale deep learning for computer aided detection of mammographic lesions</title><link>https://geertlitjens.nl/publication/kooi-17/</link><pubDate>Sun, 01 Jan 2017 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/kooi-17/</guid><description/></item><item><title>Location Sensitive Deep Convolutional Neural Networks for Segmentation of White Matter Hyperintensities</title><link>https://geertlitjens.nl/publication/ghaf-17-c/</link><pubDate>Sun, 01 Jan 2017 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/ghaf-17-c/</guid><description/></item><item><title>MAGE expression in head and neck squamous cell carcinoma primary tumors, lymph node metastases and respective recurrences: implications for immunotherapy</title><link>https://geertlitjens.nl/publication/laba-17/</link><pubDate>Sun, 01 Jan 2017 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/laba-17/</guid><description/></item><item><title>Intranodal signal suppression in pelvic MR lymphography of prostate cancer patients: a quantitative comparison of ferumoxtran-10 and ferumoxytol.</title><link>https://geertlitjens.nl/publication/deba-16-a/</link><pubDate>Sat, 01 Oct 2016 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/deba-16-a/</guid><description/></item><item><title>Automated Detection of DCIS in Whole-Slide H&amp;E Stained Breast Histopathology Images</title><link>https://geertlitjens.nl/publication/ehte-16-a/</link><pubDate>Thu, 01 Sep 2016 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/ehte-16-a/</guid><description/></item><item><title>Stain specific standardization of whole-slide histopathological images</title><link>https://geertlitjens.nl/publication/ehte-16/</link><pubDate>Thu, 01 Sep 2016 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/ehte-16/</guid><description/></item><item><title>Automated multistructure atlas-assisted detection of lymph nodes using pelvic MR lymphography in prostate cancer patients</title><link>https://geertlitjens.nl/publication/deba-16/</link><pubDate>Wed, 01 Jun 2016 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/deba-16/</guid><description/></item><item><title>In-depth tissue profiling using multiplexed immunohistochemical consecutive staining on single slide</title><link>https://geertlitjens.nl/publication/rema-16/</link><pubDate>Wed, 01 Jun 2016 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/rema-16/</guid><description/></item><item><title>Interview Regional TV at Alpe Hu'Zes</title><link>https://geertlitjens.nl/media/alpehuzes-interview/</link><pubDate>Wed, 01 Jun 2016 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/media/alpehuzes-interview/</guid><description>&lt;p>During Alpe Hu&amp;rsquo;Zes in 2016 I was given the honor to receive the Bas Mulder Award to continue my research. I was invited to speak a bit about my work during the event during the live registration.
&lt;div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;">
&lt;iframe src="https://www.youtube.com/embed/nFWP2vxO2l4" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" allowfullscreen title="YouTube Video">&lt;/iframe>
&lt;/div>
&lt;/p></description></item><item><title>Automated robust registration of grossly misregistered whole-slide images with varying stains</title><link>https://geertlitjens.nl/publication/litj-16/</link><pubDate>Fri, 01 Jan 2016 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-16/</guid><description/></item><item><title>Computer-extracted Features Can Distinguish Noncancerous Confounding Disease from Prostatic Adenocarcinoma at Multiparametric MR Imaging.</title><link>https://geertlitjens.nl/publication/litj-16-b/</link><pubDate>Fri, 01 Jan 2016 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-16-b/</guid><description/></item><item><title>Deep learning as a tool for increased accuracy and efficiency of histopathological diagnosis</title><link>https://geertlitjens.nl/publication/litj-16-c/</link><pubDate>Fri, 01 Jan 2016 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-16-c/</guid><description/></item><item><title>Pulmonary nodule detection in CT images: false positive reduction using multi-view convolutional networks</title><link>https://geertlitjens.nl/publication/seti-16/</link><pubDate>Fri, 01 Jan 2016 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/seti-16/</guid><description/></item><item><title>Clinical evaluation of a computer-aided diagnosis system for determining cancer aggressiveness in prostate MRI</title><link>https://geertlitjens.nl/publication/litj-15-b/</link><pubDate>Sun, 01 Nov 2015 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-15-b/</guid><description/></item><item><title>Multiparametric Magnetic Resonance Imaging for Discriminating Low-Grade From High-Grade Prostate Cancer</title><link>https://geertlitjens.nl/publication/vos-15/</link><pubDate>Sat, 01 Aug 2015 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/vos-15/</guid><description/></item><item><title>Automated Analysis of Histopathological Clinical Trial Data</title><link>https://geertlitjens.nl/project/aahctd/</link><pubDate>Wed, 01 Apr 2015 12:16:34 +0200</pubDate><guid>https://geertlitjens.nl/project/aahctd/</guid><description>&lt;p>To accuractely determine the outcome of clinical trials careful analysis of the biomarkers is required. In recent years this has become more and more complex due to quantity of biomarkers that has to be assessed in new clinical trials. Furthermore, as we move to more personalized therapy, we need to be able to measure even subtle changes in biomarker expression, putting a larger emphasis on accurate and precise biomarker quantification. These changes have made manual assessment of biomarker expression tedious, time-consuming, and, often, inaccurate.&lt;/p>
&lt;figure id="figure-quantification-of-cd3-positive-cells-inside-yellow-and-outside-green-the-invasive-margin-of-prostate-cancer-orange-specimens">
&lt;div class="d-flex justify-content-center">
&lt;div class="w-100" >&lt;img alt="Quantification of CD3-positive cells inside (yellow) and outside (green) the invasive margin of prostate cancer (orange) specimens." srcset="
/project/aahctd/featured_hu8dea62950b20b086faff196c1d701874_176128_b9d45b7c0672b0f780191e26c3615847.webp 400w,
/project/aahctd/featured_hu8dea62950b20b086faff196c1d701874_176128_9cf9f7d9074c024bb7cc885154553a61.webp 760w,
/project/aahctd/featured_hu8dea62950b20b086faff196c1d701874_176128_1200x1200_fit_q75_h2_lanczos_3.webp 1200w"
src="https://geertlitjens.nl/project/aahctd/featured_hu8dea62950b20b086faff196c1d701874_176128_b9d45b7c0672b0f780191e26c3615847.webp"
width="50%"
height="484"
loading="lazy" data-zoomable />&lt;/div>
&lt;/div>&lt;figcaption>
Quantification of CD3-positive cells inside (yellow) and outside (green) the invasive margin of prostate cancer (orange) specimens.
&lt;/figcaption>&lt;/figure>
&lt;p>Within this project, funded by the &lt;a href="https://www.humboldt-foundation.de/web/home.html" target="_blank" rel="noopener">Humboldt Foundation&lt;/a>, we set out to develop machine learning tools to automate the quantification and discovery of biomarkers in a variety of clinical trials. Specifically, we looked at applications in prostate cancer immunotherapy, lung cancer prognosis, and MAGE-expression in head and neck cancers.&lt;/p></description></item><item><title>A multi-scale superpixel classification approach for region of interest detection in whole slide histopathology images</title><link>https://geertlitjens.nl/publication/bejn-15/</link><pubDate>Sun, 01 Feb 2015 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/bejn-15/</guid><description/></item><item><title>PhD Thesis Defense</title><link>https://geertlitjens.nl/talk/lekenpraatje/</link><pubDate>Fri, 23 Jan 2015 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/talk/lekenpraatje/</guid><description>&lt;p>In Nijmegen it is typical to give a 10-minute layman&amp;rsquo;s presentation on your PhD research before the start of official defense.&lt;/p></description></item><item><title>Automated detection of prostate cancer in digitized whole-slide images of H&amp;E-stained biopsy specimens</title><link>https://geertlitjens.nl/publication/litj-15/</link><pubDate>Thu, 01 Jan 2015 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-15/</guid><description/></item><item><title>Computerized detection of cancer in multi-parametric prostate MRI</title><link>https://geertlitjens.nl/publication/litj-15-a/</link><pubDate>Thu, 01 Jan 2015 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-15-a/</guid><description/></item><item><title>Interview with Vox about FameLab</title><link>https://geertlitjens.nl/media/vox-famelab/</link><pubDate>Thu, 01 Jan 2015 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/media/vox-famelab/</guid><description>&lt;p>Interview about my participation in FameLab in 2015.&lt;/p></description></item><item><title>ASAP</title><link>https://geertlitjens.nl/project/asap/</link><pubDate>Tue, 01 Apr 2014 12:25:34 +0200</pubDate><guid>https://geertlitjens.nl/project/asap/</guid><description>&lt;p>ASAP (Automated Slide Analysis Platform) was developed by the Computation Pathology Group, part of the Diagnostic Image Analysis Group, at the Radboud University Medical Center. It was started after frustration with the current freely available software for annotating multi-resolution digital pathology images. For more details head to the project site.&lt;/p></description></item><item><title>Evaluation of prostate segmentation algorithms for MRI: The PROMISE12 challenge</title><link>https://geertlitjens.nl/publication/litj-14/</link><pubDate>Sat, 01 Feb 2014 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-14/</guid><description/></item><item><title>Computer-aided detection of prostate cancer in MRI</title><link>https://geertlitjens.nl/publication/litj-14-c/</link><pubDate>Wed, 01 Jan 2014 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-14-c/</guid><description/></item><item><title>Computer-aided Detection of Prostate Cancer in Multi-parametric Magnetic Resonance Imaging</title><link>https://geertlitjens.nl/publication/litj-14-e/</link><pubDate>Wed, 01 Jan 2014 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-14-e/</guid><description/></item><item><title>Distinguishing benign confounding treatment changes from residual prostate cancer on MRI following laser ablation</title><link>https://geertlitjens.nl/publication/litj-14-b/</link><pubDate>Wed, 01 Jan 2014 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-14-b/</guid><description/></item><item><title>Distinguishing prostate cancer from benign confounders via a cascaded classifier on multi-parametric MRI</title><link>https://geertlitjens.nl/publication/litj-14-a/</link><pubDate>Wed, 01 Jan 2014 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-14-a/</guid><description/></item><item><title>Multiparametric MR imaging for the assessment of prostate cancer aggressiveness at 3 Tesla</title><link>https://geertlitjens.nl/publication/vos-14-c/</link><pubDate>Wed, 01 Jan 2014 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/vos-14-c/</guid><description/></item><item><title>Quantitative identification of magnetic resonance imaging features of prostate cancer response following laser ablation and radical prostatectomy</title><link>https://geertlitjens.nl/publication/litj-14-d/</link><pubDate>Wed, 01 Jan 2014 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-14-d/</guid><description/></item><item><title>Assessment of Prostate Cancer Aggressiveness Using Dynamic Contrast-enhanced Magnetic Resonance Imaging at 3 T</title><link>https://geertlitjens.nl/publication/vos-13/</link><pubDate>Tue, 01 Jan 2013 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/vos-13/</guid><description/></item><item><title>Differentiation of Prostatitis and Prostate Cancer by Using Diffusion-weighted MR Imaging and MR-guided Biopsy at 3 T</title><link>https://geertlitjens.nl/publication/nage-13/</link><pubDate>Tue, 01 Jan 2013 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/nage-13/</guid><description/></item><item><title>Initial prospective evaluation of the prostate imaging reporting and data standard (PI-RADS): Can it reduce unnecessary MR guided biopsies?</title><link>https://geertlitjens.nl/publication/litj-13/</link><pubDate>Tue, 01 Jan 2013 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-13/</guid><description/></item><item><title>Prostate Cancer localization with a Multiparametric MR Approach (PCaMAP): initial results of a multi-center study</title><link>https://geertlitjens.nl/publication/maas-13-b/</link><pubDate>Tue, 01 Jan 2013 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/maas-13-b/</guid><description/></item><item><title>Interpatient Variation in Normal Peripheral Zone Apparent Diffusion Coefficient: Effect on the Prediction of Prostate Cancer Aggressiveness</title><link>https://geertlitjens.nl/publication/litj-12-b/</link><pubDate>Wed, 01 Aug 2012 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-12-b/</guid><description/></item><item><title>Automated computer-aided detection of prostate cancer in MR images: from a whole-organ to a zone-based approach</title><link>https://geertlitjens.nl/publication/litj-12/</link><pubDate>Wed, 01 Feb 2012 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-12/</guid><description/></item><item><title>A multi-atlas approach for prostate segmentation in MRI</title><link>https://geertlitjens.nl/publication/litj-12-d/</link><pubDate>Sun, 01 Jan 2012 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-12-d/</guid><description/></item><item><title>A pattern recognition approach to zonal segmentation of the prostate on MRI</title><link>https://geertlitjens.nl/publication/litj-12-a/</link><pubDate>Sun, 01 Jan 2012 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-12-a/</guid><description/></item><item><title>Computerized characterization of central gland lesions using texture and relaxation features from T2-weighted prostate MRI</title><link>https://geertlitjens.nl/publication/litj-12-c/</link><pubDate>Sun, 01 Jan 2012 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-12-c/</guid><description/></item><item><title>Dynamic contrast enhanced MR imaging for the assessment of prostate cancer aggressiveness at 3T</title><link>https://geertlitjens.nl/publication/vos-12/</link><pubDate>Sun, 01 Jan 2012 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/vos-12/</guid><description/></item><item><title>Automated 3-Dimensional Segmentation of Pelvic Lymph Nodes in Magnetic Resonance Images</title><link>https://geertlitjens.nl/publication/deba-11/</link><pubDate>Tue, 01 Nov 2011 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/deba-11/</guid><description/></item><item><title>Automatic Computer Aided Detection of Abnormalities in Multi-Parametric Prostate MRI</title><link>https://geertlitjens.nl/publication/litj-11/</link><pubDate>Tue, 01 Mar 2011 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-11/</guid><description/></item><item><title>Detection of Lymph Node Metastases with Ferumoxtran-10 vs Ferumoxytol</title><link>https://geertlitjens.nl/publication/deba-11-a/</link><pubDate>Sat, 01 Jan 2011 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/deba-11-a/</guid><description/></item><item><title>Differentiation of Normal Prostate Tissue, Prostatitis, and Prostate Cancer: Correlation between Diffusion-weighted Imaging and MR-guided Biopsy</title><link>https://geertlitjens.nl/publication/scho-11/</link><pubDate>Sat, 01 Jan 2011 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/scho-11/</guid><description/></item><item><title>Required accuracy of MR-US registration for prostate biopsies</title><link>https://geertlitjens.nl/publication/ven-11-a/</link><pubDate>Sat, 01 Jan 2011 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/ven-11-a/</guid><description/></item><item><title>Zone-specific Automatic Computer-aided Detection of Prostate Cancer in MRI</title><link>https://geertlitjens.nl/publication/litj-11-b/</link><pubDate>Sat, 01 Jan 2011 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-11-b/</guid><description/></item><item><title>Computerized Detection of Prostate Cancer in Multi-Parametric MRI</title><link>https://geertlitjens.nl/project/mpmri-pca/</link><pubDate>Thu, 07 Jan 2010 11:55:45 +0200</pubDate><guid>https://geertlitjens.nl/project/mpmri-pca/</guid><description>&lt;p>Prostate cancer is the most commonly diagnosed malignancy and the second leading cause of cancer death among men in the Netherlands. Due to the shortcomings of the current diagnostic pathway for prostate cancer, especially with respect to assessing cancer aggressiveness, alternative strategies are being investigated. Magnetic resonance imaging (MRI) has emerged as an important modality to assist and potentially replace (part of) the current diagnostic pathway. The high complexity of prostate MRI and the lack of sufficient expertise among the radiological community at large has opened the door for (semi-)automated analysis of prostate MRI by computer systems, with or without human intervention.&lt;/p>
&lt;figure id="figure-transversal-slide-through-the-prosate-in-a-t2-weighted-mri-sequence">
&lt;div class="d-flex justify-content-center">
&lt;div class="w-100" >&lt;img alt="Transversal slide through the prosate in a T2-weighted MRI sequence." srcset="
/project/mpmri-pca/featured_hu82c705709ed6cbe2deb511e0b8d9074a_253218_2b4c3d61f6cf40e2e403ae0e13e40869.webp 400w,
/project/mpmri-pca/featured_hu82c705709ed6cbe2deb511e0b8d9074a_253218_d4b4077858c208b27c05cd8b85eabe35.webp 760w,
/project/mpmri-pca/featured_hu82c705709ed6cbe2deb511e0b8d9074a_253218_1200x1200_fit_q75_h2_lanczos_3.webp 1200w"
src="https://geertlitjens.nl/project/mpmri-pca/featured_hu82c705709ed6cbe2deb511e0b8d9074a_253218_2b4c3d61f6cf40e2e403ae0e13e40869.webp"
width="50%"
height="545"
loading="lazy" data-zoomable />&lt;/div>
&lt;/div>&lt;figcaption>
Transversal slide through the prosate in a T2-weighted MRI sequence.
&lt;/figcaption>&lt;/figure>
&lt;p>Within this project such as system was developed and evaluated. It consisted of several key components: segmentation of the prostate gland in MRI, discovering cancer-specific features, system development and evaluation. The results were reported through a number of publications which are listed below and summarized in my &lt;a href="https://geertlitjens.nl/publication/litj-15-a/">PhD Thesis&lt;/a>.&lt;/p></description></item><item><title>Computer aided detection of prostate cancer using T2W, DWI and DCE-MRI: methods and clinical applications</title><link>https://geertlitjens.nl/publication/huis-10/</link><pubDate>Fri, 01 Jan 2010 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/huis-10/</guid><description/></item><item><title>Pharmacokinetic models in clinical practice: what model to use for DCE-MRI of the breast?</title><link>https://geertlitjens.nl/publication/litj-10/</link><pubDate>Fri, 01 Jan 2010 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-10/</guid><description/></item><item><title>Simulation of nodules and diffuse infiltrates in chest radiographs using CT templates</title><link>https://geertlitjens.nl/publication/litj-10-a/</link><pubDate>Fri, 01 Jan 2010 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-10-a/</guid><description/></item><item><title>Training a Computer Aided Detection System with Simulated Lung Nodules in Chest Radiographs</title><link>https://geertlitjens.nl/publication/snoe-10/</link><pubDate>Fri, 01 Jan 2010 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/snoe-10/</guid><description/></item><item><title>Pharmacokinetic modeling in breast cancer MRI</title><link>https://geertlitjens.nl/publication/litj-10-b/</link><pubDate>Thu, 01 Jan 2009 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-10-b/</guid><description/></item><item><title>T1 Quantification: Variable Flip Angle Method vs Use of Reference Phantom</title><link>https://geertlitjens.nl/publication/litj-09/</link><pubDate>Thu, 01 Jan 2009 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/publication/litj-09/</guid><description/></item><item><title>CAMELYON</title><link>https://geertlitjens.nl/project/camelyon/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/project/camelyon/</guid><description>&lt;h2 id="overview">Overview&lt;/h2>
&lt;p>Built on the success of its predecessor, CAMELYON17 is the second grand challenge in pathology organised by the Diagnostic Image Analysis Group (DIAG) and Department of Pathology of the Radboud University Medical Center (Radboudumc) in Nijmegen, The Netherlands.&lt;/p>
&lt;p>The goal of this challenge is to evaluate new and existing algorithms for automated detection and classification of breast cancer metastases in whole-slide images of histological lymph node sections. This task has high clinical relevance and would normally require extensive microscopic assessment by pathologists. The presence of metastases in lymph nodes has therapeutic implications for breast cancer patients. Therefore, an automated solution would hold great promise to reduce the workload of pathologists while at the same time reduce the subjectivity in diagnosis.&lt;/p>
&lt;p>
&lt;figure id="figure-detection-of-breast-cancer-metastases-in-lymph-nodes">
&lt;div class="d-flex justify-content-center">
&lt;div class="w-100" >&lt;img alt="Detection of breast cancer metastases in lymph nodes." srcset="
/project/camelyon/featured_hubada0ba51fa091eb34b957ad6af373e8_629988_7328f63abb61955f723e9608130ba948.webp 400w,
/project/camelyon/featured_hubada0ba51fa091eb34b957ad6af373e8_629988_03ef2794f5f197698bfbb938e54e7a90.webp 760w,
/project/camelyon/featured_hubada0ba51fa091eb34b957ad6af373e8_629988_1200x1200_fit_q75_h2_lanczos_3.webp 1200w"
src="https://geertlitjens.nl/project/camelyon/featured_hubada0ba51fa091eb34b957ad6af373e8_629988_7328f63abb61955f723e9608130ba948.webp"
width="500"
height="500"
loading="lazy" data-zoomable />&lt;/div>
&lt;/div>&lt;figcaption>
Detection of breast cancer metastases in lymph nodes.
&lt;/figcaption>&lt;/figure>
&lt;/p>
&lt;h2 id="task">Task&lt;/h2>
&lt;p>The TNM system is an internationally accepted means to classify the extent of cancer spread in patients with a solid tumour. It is one of the most important tools for clinicians to help them select a suitable treatment option and to obtain an indication of prognosis. Since the histological assessment of lymph node metastases is an essential part of TNM classification, CAMELYON17 will focus on the pathologic N-stage, in short: pN-stage.&lt;/p>
&lt;p>In clinical practice several lymph nodes are surgically removed after which these nodes are processed in the pathology laboratory. In this challenge we forged &lt;strong>artificial patients&lt;/strong>, with 5 slides provided for each patient where each slide corresponds to exactly one lymph node.&lt;/p>
&lt;p>The task in this challenge is to &lt;strong>determine a pN-stage for every patient in the test dataset&lt;/strong>. To compose a pN-stage, the number of positive lymph nodes (i.e. nodes with a metastasis) are counted. For the evaluation of the results we use five class quadratic weighted kappa where the classes are the pN-stages.&lt;/p>
&lt;h2 id="website">Website&lt;/h2>
&lt;p>Further information, registration, and the results are available on the &lt;a href="https://camelyon17.grand-challenge.org" target="_blank" rel="noopener">challenge website&lt;/a>.&lt;/p></description></item><item><title>Deep PCa</title><link>https://geertlitjens.nl/project/deeppca/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/project/deeppca/</guid><description>&lt;p>Most men die with, not because of prostate cancer. This high incidence-to-mortality ratio sounds like a positive trait, but comes with its own share of problems: high risk of overdiagnosis and overtreatment with significant patient morbidity. To combat overtreatment, several models have been developed to assign patients to risk categories with differing treatment regimes. Although these models show good correlation with patient outcome on the group level, their benefit for the individual patient remains limited.&lt;/p>
&lt;figure id="figure-prostate-cancer-segmentation-using-convolutional-neural-networks">
&lt;div class="d-flex justify-content-center">
&lt;div class="w-100" >&lt;img alt="Prostate cancer segmentation using convolutional neural networks." srcset="
/project/deeppca/featured_hu3f41f061ac44526ef00b786c7b376de1_58615_c39d617c39701127e2c5b6f0463b1764.webp 400w,
/project/deeppca/featured_hu3f41f061ac44526ef00b786c7b376de1_58615_4d30aa8e8b7e730f03fb991419b2c68b.webp 760w,
/project/deeppca/featured_hu3f41f061ac44526ef00b786c7b376de1_58615_1200x1200_fit_q75_h2_lanczos_3.webp 1200w"
src="https://geertlitjens.nl/project/deeppca/featured_hu3f41f061ac44526ef00b786c7b376de1_58615_c39d617c39701127e2c5b6f0463b1764.webp"
width="50%"
height="500"
loading="lazy" data-zoomable />&lt;/div>
&lt;/div>&lt;figcaption>
Prostate cancer segmentation using convolutional neural networks.
&lt;/figcaption>&lt;/figure>
&lt;p>Several groups have shown that quantifying the tumour and its micro-environment at the cellular level can result in biomarkers with strong prognostic potential, for example tumour/stroma ratio, the presence and composition of immune infiltrates or the amount of proliferating (Ki67-positive) cells. However, these findings have not translated to clinical practice due to the cumbersome and subjective manual extraction of these biomarkers from tissue slides.&lt;/p>
&lt;p>Recent years have seen the more widespread introduction of whole-slide imaging systems, which for the first time allow computerized processing of tissue slides. Automated extraction of aforementioned quantitative biomarkers through image analysis can achieve the required accuracy and robustness to impact clinical practice. In tandem with these developments, computer vision has seen a machine learning revolution over the past decade due to the advent of deep learning.&lt;/p>
&lt;p>In this project, we will combine deep learning and digitized whole-slide imaging of prostate cancer for reproducible extraction of quantitative biomarkers. Furthermore, due to the ability of deep learning systems to learn relevant features without human intervention, we expect to identify novel biomarkers which allow us to further improve the current risk models.&lt;/p>
&lt;p>The aim of this project is to prevent unnecessary surgery and adjuvant therapy for individual patients by improving currently established risk models. Risk models will be enhanced through the addition of pre- and post-operative quantitative biomarkers obtained via image analysis and deep learning. We will focus both on the accurate and objective quantification of biomarkers already identified in literature and the discovery of novel biomarkers.&lt;/p></description></item><item><title>DeepDerma</title><link>https://geertlitjens.nl/project/deepderma/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://geertlitjens.nl/project/deepderma/</guid><description>&lt;p>Due to the tripling of skin cancer incidence over the past two decades, more skin biopsies and resections are performed than ever before. This has led to an enormous increase in workload for pathologists, who perform the microscopic diagnostics of skin samples.
Most of microscopic skin analysis is not professionally challenging, but it is time consuming and can lead to reduced time for more complex diagnostics and increased wait time for patients. Machine learning and specifically deep learning offers a path to automating the diagnoses of skin samples, which would reduce the pressure on pathologists and the cost of diagnosis, both in time and money.&lt;/p>
&lt;figure id="figure-annotation-of-a-basal-cell-carcinoma-a-skin-cancer-with-typically-good-prognosis">
&lt;div class="d-flex justify-content-center">
&lt;div class="w-100" >&lt;img alt="Annotation of a Basal Cell Carcinoma, a skin cancer with, typically, good prognosis." srcset="
/project/deepderma/featured_huab03e89e8b7ea695843dd8e7fc56edc8_571530_1d607c63ae43719b7dba94a2e3b8b364.webp 400w,
/project/deepderma/featured_huab03e89e8b7ea695843dd8e7fc56edc8_571530_ae20c5f4969135c3fd13782c51c3ffe7.webp 760w,
/project/deepderma/featured_huab03e89e8b7ea695843dd8e7fc56edc8_571530_1200x1200_fit_q75_h2_lanczos_3.webp 1200w"
src="https://geertlitjens.nl/project/deepderma/featured_huab03e89e8b7ea695843dd8e7fc56edc8_571530_1d607c63ae43719b7dba94a2e3b8b364.webp"
width="50%"
height="500"
loading="lazy" data-zoomable />&lt;/div>
&lt;/div>&lt;figcaption>
Annotation of a Basal Cell Carcinoma, a skin cancer with, typically, good prognosis.
&lt;/figcaption>&lt;/figure>
&lt;p>We propose not just to develop an algorithm which can perform skin diagnostics at the level of an expert pathologist, but also explicitly identify the most fruitful way of integrating these algorithms into the routine workflow. This project is exceedingly timely, as by the end of 2019 all histopathological diagnostics in the Radboud University Medical Center will be done digitally.&lt;/p>
&lt;p>The project consists of four work packages. In work package 1, we will collect a large retrospective cohort of annotated and labeled skin biopsies and resections from multiple centers. The focus of work package 2 is on development of algorithms for segmentation of different skin tissue classes, subtyping of basal cell carcinoma, and identification of rare incidental findings. Work package 3 and 4 cover the development and prospective evaluation of the optimal algorithm-integrated workflow in a real world clinical setting.&lt;/p>
&lt;p>After completion, we will have the world’s first prospectively evaluated algorithm-supported workflow for digital pathology, and a valuable, expert labeled, retrospective dataset of skin specimens; both excellent targets for valorization. Last, it will increase the time of pathologists for complex diagnostics and reduce the wait time for patients.&lt;/p></description></item></channel></rss>