Master's Thesis opportunity about "Multi-Plane Cardiac View Synthesis for Cardiac MRI"

31.10.2025, Abschlussarbeiten, Bachelor- und Masterarbeiten

This work leverages generative AI to synthesize standard cardiac MRI views (e.g., 4-chamber, short-axis) from routine localizers. Although low in quality, localizers provide comprehensive 3D heart coverage. The goal is to enable view synthesis and plane interpolation to aid scan planning, diagnosis, and efficient cardiac imaging.

In cardiac MRI protocols, non-diagnostic localizer scans are routinely acquired to visualize the heart prior to detailed, time-resolved diagnostic sequences. Although these localizers are static and of lower image quality, they provide a better understanding of the cardiac volume, offering a comprehensive 3D overview of the heart and representing a valuable yet underutilized data source for advanced image generation.

Multiple views are essential in cardiac MRI because the heart is a complex, three-dimensional, and dynamically moving organ. Different views capture complementary anatomical and functional information that cannot be obtained from a single plane. Conventional methods require acquisition of multiple views sequentially, increasing the total scan time.

This work aims to develop methods for synthesizing arbitrary or anatomically standard cardiac MRI planes (e.g., 4-chamber, short-axis) from localizers, leveraging generative AI models to enable view synthesis and plane interpolation that could facilitate scan planning, support diagnosis, or serve as a surrogate for software-based acquisitions in time-constrained settings.

Your tasks will include: - Development of generative or neural-hybrid models for novel plane synthesis - Exploitation of localizer data for 3D representation learning - Evaluation of spatial and anatomical accuracy of generated planes

The student will first explore the landscape of multi-view synthesis and generative AI models. You'll have a chance to: - Access to a rich cardiac MRI datasets. - Guidance from both clinical and AI experts in a collaborative, interdisciplinary research setting. - Opportunity to contribute to ongoing research and potential publication in medical imaging journals or conferences. - Exposure to cutting-edge AI tools and frameworks for 3D medical image generation.

References: 1- Bourigault, Emmanuelle, Abdullah Hamdi, and Amir Jamaludin. "X-diffusion: Generating detailed 3d mri volumes from a single image using cross-sectional diffusion models." (2024). 2- Liu, Ruoshi, et al. "Zero-1-to-3: Zero-shot one image to 3d object." Proceedings of the IEEE/CVF international conference on computer vision. 2023.

Kontakt: s.kafali@tum.de, niklas.bubeck@tum.de

Mehr Information

https://thesis.aet.cit.tum.de/topics/2a74e736-83d0-4996-be2a-501080e5bd4c