A5095925698- Medical Imaging and Analysis
- Brain Tumor Detection and Classification
- Neuroinflammation and Neurodegeneration Mechanisms
- Optical Imaging and Spectroscopy Techniques
- Photoacoustic and Ultrasonic Imaging
- Advanced Technologies and Applied Computing
- Conducting polymers and applications
- Advanced MRI Techniques and Applications
- Medical Image Segmentation Techniques
Polytechnique Montréal
2024
Mila - Quebec Artificial Intelligence Institute
2024
Deep learning models have achieved remarkable success in segmenting brain white matter lesions multiple sclerosis (MS), becoming integral to both research and clinical workflows. While gained significant attention MS research, the involvement of spinal cord is relatively understudied. This largely owed variability magnetic resonance imaging (MRI) acquisition protocols, high individual anatomical differences, complex morphology size - lastly, scarcity labeled datasets required develop robust...
Abstract Metabolism and bioenergetics in the central nervous system play important roles pathophysiology of Parkinson’s disease (PD). Here, we employed a multimodal imaging approach to assess oxygenation changes spinal cord transgenic M83 murine model PD comparison non-transgenic littermates at 9-12 months-of-age. A lower oxygen saturation (SO 2 ) SVOT was detected vivo with spiral volumetric optoacoustic tomography (SVOT) mice compared littermate mice. Ex-vivo high-field T1-weighted...
Abstract Purpose Metabolism and bioenergetics in the central nervous system play important roles pathophysiology of Parkinson’s disease (PD). Here, we employed a multimodal imaging approach to assess oxygenation changes spinal cord transgenic M83 murine model PD overexpressing mutated A53T alpha-synuclein form comparison with non-transgenic littermates. Methods In vivo spiral volumetric optoacoustic tomography (SVOT) was performed oxygen saturation (sO 2 ) cords mice Ex high-field...
Motivation: 3D visualisation of the spinal cord and vertebrae anatomy is critical for treatment planning assessment atrophy in neurodegenerative traumatic diseases. Goal(s): Develop a fully automatic segmentation whole cord, discs. Approach: The hybrid method combines nnU-Net with an iterative processing algorithm Spinal Cord Toolbox to conveniently generate ground truth labels. We used T1w T2w scans from three different databases. Results: A validation Dice score 0.928 was obtained...
Motivation: Longitudinal analysis of spinal cord multiple sclerosis (MS) lesions is clinically relevant for the early diagnosis and monitoring MS progression. Goal(s): Develop a deep learning tool automatic segmentation on PSIR STIR images from sites. Approach: A nnUNet model was trained tested baseline data applied to follow-up scans create lesion distribution maps. Results: We demonstrated utility map spatio-temporal across phenotypes. The packaged into an open-source software. Impact:...