CXR-CML: Improved Zero-Shot Classification of Long-Tailed Multi-label Diseases in Chest X-Rays

Madhipati R, Bhat S, Bueß L, Maier A (2025)

Publication Language: English

Publication Type: Conference contribution, Conference Contribution

Publication year: 2025

Publisher: Springer, Cham

Series: Lecture Notes in Computer Science

Book Volume: 15966

Pages Range: 119–129

Conference Proceedings Title: 28th International Conference of Medical Image Computing and Computer Assisted Intervention – MICCAI 2025

Event location: Daejeon KR

ISBN: 978-3-032-04977-3

URI: https://link.springer.com/chapter/10.1007/978-3-032-04981-0_12#chapter-info

DOI: 10.1007/978-3-032-04978-0_20

Abstract

Chest radiography (CXR) plays a crucial role in the diagnosis of various diseases. However, the inherent class imbalance in the distribution of clinical findings presents a significant challenge for current self-supervised deep learning models. These models often fail to accurately classify long-tailed classes. Current Vision-Language models such as Contrastive Language Image Pre-training (CLIP) models effectively model the manifold distribution of the latent space, enabling high zero-shot classification accuracies. Although CLIP performs well on most of the primary classes in the dataset, our work reveals that its effectiveness decreases significantly for classes with a long-tailed distribution. Our approach employs a class-weighting mechanism that directly aligns with the distribution of classes within the latent space. This method ensures a substantial improvement in overall classification performance, with particular emphasis on enhancing the recognition and accuracy of rarely observed classes. We accomplish this by applying Gaussian Mixture Model (GMM) clustering to the latent space. The subsequent clusters are further refined by Student t-distribution, followed by a metric loss that utilizes the altered embeddings. Our approach facilitates stable and adaptive clustering of the features. This results in a notable average improvement of 7% points in zero-shot AUC scores across 40 classes in the MIMIC-CXR-JPG dataset from previous SOTA models. Our code is publicly available at: CXR-CML

Authors with CRIS profile

Sheethal Bhat Lehrstuhl für Informatik 14 (Bild- und Sprachverarbeitung) (LME) Lukas Bueß Lehrstuhl für Informatik 14 (Bild- und Sprachverarbeitung) (LME) Andreas Maier Lehrstuhl für Informatik 5 (Mustererkennung)

How to cite

APA:

Madhipati, R., Bhat, S., Bueß, L., & Maier, A. (2025). CXR-CML: Improved Zero-Shot Classification of Long-Tailed Multi-label Diseases in Chest X-Rays. In 28th International Conference of Medical Image Computing and Computer Assisted Intervention – MICCAI 2025 (pp. 119–129). Daejeon, KR: Springer, Cham.

MLA:

Madhipati, Rajesh, et al. "CXR-CML: Improved Zero-Shot Classification of Long-Tailed Multi-label Diseases in Chest X-Rays." Proceedings of the MICCAI 2025, Daejeon Springer, Cham, 2025. 119–129.

BibTeX: Download