A Unified Framework for Comprehensive Cardiac CT Segmentation and Phenotyping: Human-in-the-Loop Data Annotation, Vision Foundation Model Development, Multicenter Evaluation and Clinical Validation

2026-07-13

AI Summary

Conventional cardiac CT scans suffer from limited scalability in comprehensive structural quantification. This work proposes a unified framework that, for the first time, integrates human-in-the-loop annotation, self-supervised pretraining on 60,000 unlabeled CT scans, and advanced data augmentation to construct the largest expert-annotated cardiac CT dataset to date. The study systematically evaluates convolutional, Transformer, and state-space architectures within this framework. The resulting model significantly outperforms existing open-source tools across five external datasets, markedly improves annotation efficiency under low-data regimes, and enables robust population-level phenotypic analysis with strong generalization. Notably, the segmentation outputs demonstrate clinically meaningful associations with ventricular function and disease severity.

Abstract

Comprehensive quantification of cardiac structures from computed tomography (CT) remains limited not by data availability but by the scalability of measurements, which makes routine use impractical. Here we present a unified framework for comprehensive cardiac CT segmentation and phenotyping that combines a human-in-the-loop annotation pipeline, a cardiac CT augmentation technique, and a self-supervised foundation model pre-trained on 60,000 unlabeled cardiac CT scans. Using this approach, we assembled the largest and most comprehensive expert-annotated cardiac CT segmentation dataset to date, comprising 1598 cases and 14 distinct cardiac structures (1000 for training, 598 for the external test set). Across five external datasets, the framework segmented all structures more accurately and comprehensively than existing open-source tools. Self-supervised pre-training improved labeling efficiency, with the most significant gains observed during external evaluation in the low-data regime. Benchmarking across convolutional, transformer, and state-space architectures showed comparable performance, indicating that data quality and pre-training, rather than architecture, drove accuracy. The framework was scaled to population-level phenotyping, with segmented anatomy that carries functionally relevant information about ventricular function and disease severity beyond demographic variables. By openly releasing the largest dataset with human labels, code, model weights, a CT augmentation library, and software, this work provides a reproducible foundation for opportunistic cardiac phenotyping from routinely acquired CT scans.

Problem

• cardiac CT segmentation
• scalability of measurements
• comprehensive phenotyping
• clinical routine application
• quantification of cardiac structures

Innovation

• human-in-the-loop annotation
• self-supervised foundation model
• cardiac CT segmentation
• multicenter validation
• opportunistic phenotyping