Incorporating evidential learning to enhance efficiency and generalizability of vision-language model adaptation. Check out our recently accepted MICCAI paper [early acceptance], Evi-steer by Taha Koleilat! arxiv.org/abs/2605.26292
How to train AI models for generalized blood vessel segmentation without expert annotation? Check out our newly accepted MICCAI paper [early acceptance] by Erin Rainville! arxiv.org/abs/2605.26277
Very excited to announced the 8th MICCAI workshop on Machine Learning in Clinical Neuroimaging (MLCN)! We are accepting paper submissions in both ML and clinical neuroimaging tracks (deadline on July 1). mlcnworkshop.github.io
Attention @arxiv authors: Our Code of Conduct states that by signing your name as an author of a paper, each author takes full responsibility for all its contents, irrespective of how the contents were generated. 1/
I am going to give a talk for the IEEE Engineering In Medicine & Biology Society (EMBS) Montreal Chapter on Jun 18. If you are interested, come join us :) events.vtools.ieee.org/m/558…
Talking at the Future of Math Symposium in 10 minutes (livestream link: youtube.com/live/tN4hsT5t0nw…). I decided to make the talk "personal" and explain the five moments that updated me on how fast AI will change mathematics. (The ChatGPT's illustration of these moments is so good!!)
Check out our TMLR paper "CLIP-SVD: Efficient and Interpretable Vision–Language Adaptation via Singular Values" by Taha Koleilat! CLIP-SVD offers highly efficient and robust CLIP model adaptation, with text-based explanation for the adaptation dynamics🔍. openreview.net/forum?id=XYy8…
Some thoughts being a reviewer, AC, and committee member this year so far: medical AI should always prioritize the design towards the end application than exclusively chasing novelty and fancy concepts.
This is a great tool for surgical plan review. However, most of such tools still overlook effective user interaction methods and real-time quality feedback in planning and monitoring. Reviewers also tend to overlook this fact as well🤷‍♂️ youtu.be/C60QCvdq6cs?si=2CUz…
Combining mobile AR and cue cards in a structured pedagogical framework for clinical stroke education. Check our work led by Bahar Jahani "NeuroVase: A Tangible Mobile Augmented Reality Learning System for Neurovascular Anatomy and Stroke Education"! arxiv.org/abs/2604.00296
Excited to share our new work: "EEG-ChiMamba: Towards a Robust Mamba-Based Architecture for Dementia Detection from Resting State Electroencephalography" by Christopher Neves! biorxiv.org/content/10.64898…. We explore a new EEG channel-indep. Mamba model & XAI for dementia diagnosis.
I am not against using coding/writing agents for research works, but there seems to be a big drive to use them to speed up publication counts, potentially at the expenses of depriving the learning experience/thinking of trainees. Is it what university research is going to now?