CT-IDP: Segmentation-Derived Quantitative Phenotypes for Interpretable Abdominal CT Disease Classification

2026-05-09

AI Summary

Abdominal CT–based disease classification often lacks interpretability and standardized quantitative phenotypic support. This work proposes CT-IDP, the first organ segmentation–driven high-dimensional quantitative phenotyping framework, which leverages TotalSegmentator to extract over 900 morphological, density, and contextual features from multi-institutional CT scans. An interpretable classification model is constructed using elastic net–regularized logistic regression and benchmarked against a DINOv3 vision transformer baseline. Evaluated on three external datasets—MERLIN, Duke-Abdomen, and AMOS—the proposed model achieves macro-AUCs of 0.897, 0.877, and 0.780, respectively, significantly outperforming the baseline. The approach demonstrates strong interpretability, reproducibility, and cross-institutional generalizability in abdominal CT disease classification.

Abstract

In this retrospective multi-institutional study, a quantitative phenotyping framework, CT-IDP (CT Image-Derived Phenotypes) was developed on the MERLIN abdominal CT benchmark (training, validation, and test sets- 15,175, 5,018, and 5,082 studies, respectively) and externally evaluated on two independent dataset: Duke-Abdomen (2,000) and AMOS (1,107). Multi-organ segmentations were generated with TotalSegmentator and used to derive over 900 organ and compartment-level descriptors spanning morphometry, attenuation, and contextual/burden findings. Sparse disease-specific logistic regression with elastic-net regularization was trained on MERLIN and externally validated under a frozen specification. Performance was compared against a DINOv3-based vision-transformer baseline using AUC and average precision (AP), supported by phenotype-stratified audits and coefficient-level inspection. Macro-AUC for CT-IDP versus the baseline was 0.897 versus 0.880 on MERLIN, 0.877 versus 0.857 on the Duke-Abdomen dataset, and 0.780 versus 0.756 on AMOS.

Problem

• abdominal CT
• disease classification
• quantitative phenotyping
• interpretable AI
• medical image analysis

Innovation

• quantitative phenotyping
• interpretable AI
• multi-organ segmentation
• elastic-net regularization
• external validation