What's new in OnkoDICOM in 2022?
Create ROI
This year the function to draw a new region of interest has been overhauled. The function now operates in the same window as the main window. For the individual user, drawing an ROI is a time during which nothing else is done, so overlapping screens are unnecessary.
The name selection, cursor size and image size manipulation have all been upgraded and made more intuitive. The drawing functions have been expanded to include three distinct methods - using a cursor to manually draw, filling on the displayed slice using selected pixel values, and 3D fill on all slices above and below that match the selection conditions on the starting window (like using a skewer).
The alpha value which determines the conformality of the contour has also been enhanced so that the last selected value remains for future use.
The button arrangement has been updated and is visually more pleasant.
A common start window
OnkoDICOM has previously provided two modes - one for the curation of individual patient image sets, and a second for curation of whole directories (a.k.a. batch processing). These two modes were separate and required a program restart to switch. This year there is now a single starting window for both single patient and batch processing functions meaning that the program does not need to be re-started.
What's new in OnkoDICOM in 2022?
Re-linking DICOM-RT files
There are four file types making up the DICOM-RT data set. These files are all supposed to be linked to one another. If a plan is generated repeatedly and saved in the same place, it is possible for the file linking to become broken. If all the files are linked to the same CT image set, this function will re-link the connections between RTDOSE, RTPLAN and RTSTRUCT files.
What's new in OnkoDICOM in 2022?
Select Subgroup
In the past OnkoDICOM was able to process a single directory without selecting any of the data contained within. This was quite OK while dealing with small patient groups, but as the number of analyses increases, the task of maintaining multiple directories often with overlapping patients became too resource intensive.
Now OnkoDICOM can select a subgroup of patients according to the Clinical Data stored in the Clinical Data DICOM-SR file. This file can be automatically updated from a spreadsheet containing a patient identifier that matches that of the DICOM files.
This subgroup selection also affects all subsequent batch processing functions used.
FMA_ID to ROI Name
OnkoDICOM is very useful for processing anatomical names given to ROIs. The requirement is that all names are identical and standard according to the list already held in OnkoDICOM. This list is configurable for each user but each name has an associated FMA ID.
The function to change Standard Names into FMA IDs (ROI NAme to FMA ID) already exists, but this year the reverse function has been implemented (FMA ID to ROI Name).
This permits data to be shared and for the Standard Names to ne harmonised over international jurisdictions.
ROI Name Cleaning
The radiation plan contains two different types of names - anatomical names and volume names. Previously the ROI Name Cleaning function checked all of the anatomical names and selected those that don't match the Standard Name list to suggest the proper name. The user could then choose to leave the name unchanged, modify or delete it.
This year the function separates Anatomical Names from Volume Names allowing them to be dealt with separately. The new function is also more accurate at offering correct names.
Machine Learning
This year's OnkoDICOM has the first edition of a machine learning tool. The tool has two parts. The first exists inside the Batch Processing portion of OnkoDICOM and is designed to develop a model to answer a clinical question.
The clinical question will use a single primary field inside the Clinical Data spreadsheet (e.g., Local Recurrence) and a clinical group definition also using the Clinical Data (e.g., Lung Cancer, Stage 1, SABR treatment). This is combined with the DVH for the PTV containing the tumour to indicate treatment quality and with the Pyradiomics of the tumour defined on the CT scan (i.e., the GTV).
These data are used to generate a model, and the model parameters are saved to define a 'high' and 'low' risk classification.
The second portion exists in the Individual Curation portion of OnkoDICOM. The radiation plan of a new patient is imported and processed for DVH and Pyradiomics, and matching clinical data is obtained. The relevant model is selected for use, and the new data is processed to assign the new case to either the 'low' or 'high' risk group.
The oncologist is then free to use this classification in patient management.

Please test and provide feedback. Our development teams are students, they work very hard, and feedback is good for the whole team.
Version One is expected shortly.