Faithful, Interpretable Chest X-ray Diagnosis with Artifact-free B-cos Networks

Abstract:Faithfulness and interpretability are essential for deploying deep neural networks (DNNs) in safety-critical domains such as medical image analysis. B-cos networks modify the parameterization of convolutional and classification layers to measure class evidence via feature-weight alignment, enabling built-in, class-specific contribution maps without post-hoc explanations. While maintaining diagnostic performance competitive with state-of-the-art DNNs, standard B-cos networks exhibit severe aliasing artifacts in their explanation maps, rendering them unsuitable for clinical use, where clarity is essential. In this work, we address this limitation by introducing anti-aliasing strategies using ASAP and BlurPool (BP) to significantly improve explanation quality. Our experiments on chest X-ray datasets demonstrate that the modified $\text{B-cos}_\mathrm{ASAP}$ and $\text{B-cos}_\mathrm{BP}$ preserve strong predictive performance while providing faithful and artifact-free explanations suitable for clinical application in multi-class and multi-label settings. Code is available at: this https URL.

First two listed authors have equal contributions. Accepted at MICCAI 2026

Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)

arXiv:2507.16761 [cs.CV]

(or arXiv:2507.16761v3 [cs.CV] for this version)

https://doi.org/10.48550/arXiv.2507.16761 Focus to learn more arXiv-issued DOI via DataCite

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.