Batoul Dahman

Designed and implemented advanced modules in Machine Learning, Deep Learning, Computer Vision, and Control Systems.

Working towards a deep learning project in medical image segmentation to benchmark convolutional technologies (like Unet) against transformer-based methods. The focus is on optimizing architecture for reduced energy consumption and minimal CO2 emissions. During my teaching tenure in L3 Informatics, focusing on image and signal processing, I led several projects involving image processing, segmentation, detection, and classifcation.

Image-based guidance of a transesophageal HIFU probe for the treatment of cardiac arrhythmias. My thesis in medical image processing, part of an ANR project, focused on developing methods to register preoperative 3D (CT/MRI) images with real-time 2D ultrasound images for image-guided therapy in cardiac arrhythmia treatment. I innovated using deep learning and classical iterative techniques to enhance the accuracy, efficiency, and speed of both rigid and non-rigid registration.

Transesophageal 2D Ultrasound to 3D Computed Tomography registration for the guidance of a cardiac arrhythmia therapy. Initiated my thesis research by developing and testing a classical method for 3D/2D CT image registration, laying the foundational groundwork for my doctoral studies.

- Detection of premature babies’ movements - Skills: Signal processing, Video processing, logistic regression Matlab

- Classification of SSVEP EEG - Skills: Medical signal processing, Linear classification LDA, Brain Computer Interface (BCI), Machin learning, Matlab