AI for healthcare
Deep learning for stroke diagnosis, breast cancer detection and multimodal imaging, built to be explainable, because a clinician has to be able to see why.
News
MaxVQA accepted in IEEE J-BHI
Our paper "MaxVQA: A Large-Scale Multi-Source Dataset for Explanatory Visual Question Answering in Mammography" has been accepted for publication in the IEEE Journal of Biomedical and Health Informatics. Authors are Mohamed Mabrok, Yalda Zafari, Rehab Elsheikh, Rashad Mohamed and Essam A. Rashed.
TomoMamba accepted in Multimedia Tools and Applications
Our paper "TomoMamba: A Two-Stage Model with State-Space Cross-Slice Propagation for Breast Cancer Diagnosis in Digital Breast Tomosynthesis" has been accepted for publication in Multimedia Tools and Applications. Authors are Shahd Soliman, Yalda Zafari, Essam Rashed and Mohamed Mabrok.
HamVision accepted in Medical Image Analysis
The single-authored paper "HamVision: Hamiltonian Dynamics as Inductive Bias for Medical Image Analysis" has been accepted in Medical Image Analysis (Elsevier), the flagship journal of the MICCAI community, and is among the first publications there from Qatar University. It embeds a damped harmonic oscillator as a bottleneck inside a vision model, reaching state of the art on 5/5 medical segmentation benchmarks and leading 8/9 MedMNIST classification tasks across 9 imaging modalities with far fewer parameters than transformer baselines.
Attending MICCAI
I will be attending MICCAI 2026, Strasbourg, France, September 27 - October 1, 2026.
1 earlier item
Paper published in IEEE TNNLS
Our paper "Controllability Analysis for Vision State Space Models: A Structural Interpretability Framework" has been published in IEEE Transactions on Neural Networks and Learning Systems.
What the lab works on
Agentic AI
Systems that decide and act rather than only predict: agents that plan, use tools, hold a goal across many steps, and can be held to account for what they did.
Robust and optimal control
Negative imaginary systems theory, robust stability, and feedback design for systems that ring, settle and interconnect.
Robotics and autonomy
Perception, navigation and online learning for UAVs and ground robots in environments that do not hold still.
Dynamics and game theory
Passivity of replicator dynamics, stability analysis, and how strategic agents settle into equilibria.
AI in mathematics education
What a course looks like when every answer is machine checkable, the tutor has read the same pages the student has, and the record of a term is a dataset.
Recent work
- Journal2026
Controllability Analysis for Vision State Space Models: A Structural Interpretability Framework
- Journal2026
HamVision: Hamiltonian dynamics as inductive bias for medical image analysis
- Journal2025
Negative Imaginary Systems Theory for Nonlinear Systems: A Dissipativity Approach
- Journal2021
Passivity Analysis of Replicator Dynamics and Its Variations
- Journal2014
Generalizing Negative Imaginary Systems Theory to Include Free Body Dynamics: Control of Highly Resonant Structures With Free Body Motion
- Journal2013
Spectral Conditions for Negative Imaginary Systems With Applications to Nanopositioning
Everything, by year → · The same work in plain words →
Teaching
Seven courses at Qatar University run on a platform built for them rather than bought: worked examples that stay closed until you have tried the problem, practice that explains rather than ticks, a tutor that has read the same pages the student has, and an exam bank where every answer has been re-derived by a computer algebra system before a student ever meets it.
Calculus I
Calculus II
Differential Equations
Mathematics for Engineering
Foundations of Mathematics
Linear Algebra
Mathematics for Electrical Engineering
How the teaching works → · Go to the courses →
Working with the lab
Students interested in AI for healthcare, agentic AI, control theory or robotics are welcome to write. A CV and a short note about what you want to work on is much easier to answer well than a general enquiry.