Sidaty El Hadramy, PhD.
Postdoctoral Researcher
MIRACLE II Technical coordinator
University of Basel
About
I am a Postdoctoral Researcher in the Center for Image Analysis and Navigation (CIAN) at the University of Basel, where I work under the supervision of Prof. Philippe C. Cattin. My research lies at the intersection of artificial intelligence, medical image analysis, and soft-tissue biomechanics, with a focus on developing solutions for surgical guidance systems. I am particularly interested in building medical image registration frameworks that can be seamlessly integrated into real-time augmented reality systems while respecting the underlying physical properties and behavior of biological tissues.
I received my PhD in Computer Science from the University of Strasbourg, where I worked in two leading research groups: the MIMESIS team at Inria, led by Prof. Stéphane Cotin, and CAMMA at IHU Strasbourg, led by Prof. Nicolas Padoy. My doctoral research focused on developing the first use of IntraVascular Ultrasound (IVUS) imaging in an augmented reality system to assist liver surgery. This work required tackling several technical challenges at the intersection of multiple disciplines. I developed computational methods that integrate AI with the finite element method to achieve real-time and accurate simulation of soft-tissue deformation. I also developed AI-based techniques for registration, reconstruction and segmentation of ultrasound data, with the goal of enabling accurate and efficient intraoperative workflows.
Before embarking on my doctoral studies, I obtained my engineering degree from École des Mines de Saint-Étienne in France, where I developed a strong foundation in applied mathematics, computer science, and engineering principles that continue to inform my interdisciplinary research approach.
Current Research Focus
Implicit Neural Representations (INRs) can provide continuous and highly differentiable parameterizations of signals that are inherently discrete or discontinuous, such as images. In our previous work, we introduced cIDIR, a conditioned INR framework for deformable image registration. This approach offers the key advantage of producing smooth and fully differentiable displacement fields, enabling the use of advanced regularization schemes (such as bending energy or hyperelastic regularizations) that rely on accurate higher-order gradient computation. However, the framework is fundamentally limited by its patient-specific design (requires a training per pair of images), which restricts its generalization capabilities. To overcome this limitation, I am currently exploring neural-operator–based approaches that learn mappings between function spaces. The objective is to generalize INR-based deformable image registration by predicting an INR describing a displacement field directly from the INRs of the moving and fixed images.
I am always open to collaboration. Feel free to drop me an e-mail.
News
Selected Publications
Publications
*: Shared first authorship.2025
Optimization in latent space for real-time intraoperative characterization of digital twins
Towards MR-Based Trochleoplasty Planning
Soft Tissue Simulation and Force Estimation from Heterogeneous Structures using Equivariant Graph Neural Networks
2024
2023
