I'm an MSc student in the Integrated Program in Neuroscience at McGill, and I just completed my first semester. I finished my undergrad at McGill last year, and my major was cognitive science with concentrations in neuroscience and computer science. Throughout undergrad I dabbled in a variety of research methods. My cognitive science undergrad project was in a neuropsychology lab at the Douglas where I tried relating personality and creativity measures to spatial and stimulus-response navigation. My wet lab experience in two other labs was brief, and I learned microtome methods for sectioning finch brains and a bit of MATLAB for analyzing bird song. As I took more computer science classes I became interested in the intersection of computer science, statistics, and medical research; thus, I joined breast cancer informatics lab where I worked on a machine learning project to subtype breast cancer patients in a generalizable way based on microarray data. Upon graduating, I joined the McGill Center for Integrative Neuroscience as a software developer for LORIS.
The main inspiration for my MSc project is Itturia-Medina's Epidemic Spreading Model (ESM). In a nutshell, the ESM is a mechanistic model that predicts amyloid beta deposition based on a clearance rate, production rate, and a noise factor. At the moment, I have several goals for my project:
- apply the ESM to ADNI grey matter cortical volume and thickness data
- develop a comparable deep learning model to predict Alzheimer's disease progression
- apply the ESM to the DIAN (Dominantly Inherited Alzheimer's Network) data and further optimize it
In addition to my MSc project, I have been working with @crocodoyle on a separate project to create a REST-API for a deep learning model to predict QC probabilities and to make it interoperable with LORIS. I'm interested in potentially turning the Flask app into a Docker container during the Brainhack school.
I also am working on the first of my project goals. I'm new to neuroimaging, so I'd like to learn more about Nipype and Keras/Pytorch.
If time permits, I'd love to also learn more about ndmg for connectome estimation.