At AKDITAL INNOV, AKDITAL, I develop and validate data-driven methods for clinically relevant problems in a hospital-embedded R&D center. My work includes 3D deep-learning methods for MRI-to-synthetic CT generation and radiotherapy dose prediction, with graphical interfaces for reviewing results.
I also investigate machine-learning approaches for ECG-based prediction of pre-diabetes and diabetes. For clinical NLP, I build data pipelines for LLM/RAG-based discharge summary generation, including anonymization and text extraction from clinical documents.
At KINDI Center, Qatar University, I developed a deep-learning model for classifying kidney histopathological images, improving cancer grading accuracy from 90% to 92% and F1-score by 2%. I benchmarked it against other models and used GradCAM to examine its decisions.
At Oticon Medical, I developed cochlear implant electrode-array insertion simulations using SOFA and implemented robotic path-planning algorithms. My projects have also included multimodal prediction of prostate cancer recurrence and transfer learning for EEG-based brain-computer interfaces.
