Data Augmentation Using Physics-Informed Plaque Growth Simulation

Description

A key challenge in medical image segmentation and feature analysis is data availability due to IRB and regulatory hurdles. The recent interest in applying ML to medical image analysis demands large and diverse datasets, making data augmentation strategies critical for medical imaging. Project PrediCT aims to leverage data augmentation to boost segmentation and prediction pipelines. However, synthetic calcium plaque data is difficult to produce given the specific nature of calcium deposition patterns and required anatomical accuracy. This project will develop a physics-informed simulation framework to generate synthetic calcium masks for CAC scoring scans. Our approach is to use atlas-based vessel territories combined with hemodynamic-inspired placement rules (rather than exact centerlines) to generate anatomically plausible synthetic calcium. Contributors can optionally incorporate diffusion models for lesion morphology while using rule-based placement to ensure physical constraints.

Duration

Total project length: 175/350 hours.

Difficulty Level

Task ideas

Expected results

Test

Please use this link to access the test for this project.

Requirements

Mentors

Please DO NOT contact mentors directly by email. Instead, please email ml4-sci@cern.ch with Project Title and include your CV and test results. The mentors will then get in touch with you.

Corresponding Project

Participating Organizations