Radiomics Feature Extraction and Calcium Phenotype Discovery

Description

Coronary artery calcium (CAC) scoring traditionally uses the Agatston score, a single number representing calcium burden, as robust major adverse cardiac event (MACE) predictor. However, recent research shows that calcium morphology patterns provide additional predictive value for MACE beyond total calcium burden alone. PrediCT is working with the Stanford COCA dataset (gated CAC scans with segmentation masks) but currently lacks MACE endpoints. This project will develop a feature extraction and phenotyping framework that extracts radiomics features (shape, texture, spatial distribution) from calcium masks, applies dimensionality reduction to identify key features, and discovers calcium phenotypes via unsupervised clustering. We aim to develop a validation framework for discovered phenotypes without clinical endpoints, a novel approach. This can be done through Agatston correlation, reproducibility testing, and clinical pattern alignment.

Duration

Total project length: 175/350 hours.

Difficulty Level

Task ideas

Contributors have creative freedom to explore calcium features. Suggested approaches:

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