I am a PhD student in Computer Vision in the department of Engineering, University of Oxford. My work consists of developping computer models for the automated assessment of scoliosis and spinal deformities. My interests lie in image segmentation, computational geometry, domain adaptation and uncertainty.
Pseudo-labelling the segmentation to handle the domain gap between datasets and a piecewise cubic spline to approximate the predicted spinal midpoints and handle the noise
A new network based on an encoder/decoder architecture with attention mechanisms and full inter- and intra-skip connections inspired from UNet3+ to take advantage of low-level and high-level semantics at full scales is proposed.
For post-processing, we used Conditional Random Fields (CRF) to refine the predicted segmentation maps.
A Cox proportional hazard regression with relevant clinical, radiomic, and deep learning features was our best performing model.
Machine Learning, Programming Python, Data Pre-Processing and Analysis (MRI, fMRI, EEG), Neuropsychology, Pharmacology and Neurodegenerative Diseases.
A deep learning approach using graph neural networks for axon tracking in 2P-microscopy images, supervised by Dr. Anil Anthony Bharath, BICI lab.