# Spatially Grounded Concept Bottleneck Models for Trustworthy Breast Ultrasound Diagnosis

arXiv: https://arxiv.org/abs/2607.20691 · HTML: https://mlsanity.com/paper/2607.20691/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 8 of its 19 references that are in the map.
- First posted on arXiv 22 Jul 2026.

## Builds on (most cited first, 8 of 8)

- [Concept Bottleneck Models](https://mlsanity.com/paper/2007.04612/) (2020)
- [SAM 3: Segment Anything with Concepts](https://mlsanity.com/paper/2511.16719/) (2025)
- [A comprehensive survey on deep active learning in medical image analysis](https://mlsanity.com/paper/2310.14230/) (2023)
- [On the Concept Trustworthiness in Concept Bottleneck Models](https://mlsanity.com/paper/2403.14349/) (2024)
- [BI-RADS-Net: An Explainable Multitask Learning Approach for Cancer Diagnosis in Breast Ultrasound Images](https://mlsanity.com/paper/2110.04069/) (2021)
- [Weakly Supervised Lesion Detection and Diagnosis for Breast Cancers with Partially Annotated Ultrasound Images](https://mlsanity.com/paper/2306.06982/) (2023)
- [Learning a Clinically-Relevant Concept Bottleneck for Lesion Detection in Breast Ultrasound](https://mlsanity.com/paper/2407.00267/) (2024)
- [Unmasking Dementia Detection by Masking Input Gradients: A JSM Approach to Model Interpretability and Precision](https://mlsanity.com/paper/2402.16008/) (2024)

## Built on by (most cited first, 1 of 1)

- [Loss Invariance Determines What Concept Layers Encode: Volume Grounding in Echocardiography](https://mlsanity.com/paper/2607.25748/) (2026)

## Abstract (arXiv)

Concept Bottleneck Models provide interpretable-by-design predictions by mediating diagnosis through human-understandable concepts, but in medical imaging, their trustworthiness is often limited by the quality and granularity of available supervision. In particular, predicted concept activations can be driven by irrelevant regions, leading to spatially unfaithful explanations. We study a data-centric spatially grounded Concept Bottleneck Model (SG-CBM) that leverages coarse lesion delineations as weak supervision to encourage anatomically plausible concept evidence. For breast ultrasound, we derive two clinically motivated zones from each lesion mask: (i) an in-lesion region of interest for morphology-related concepts and (ii) a posterior acoustic band for posterior phenomena. We train concept maps using a grouped spatial grounding objective and preserve semantic faithfulness with a linear bottleneck classifier. Across five-fold stratified group cross-validation, the proposed SG-CBM improves diagnostic AUROC and concept macro-AUROC while markedly increasing spatial alignment of concept evidence. We also perform a Train-corrupt/Test-clean annotation-quality stress test to quantify the impact of supervision quality on diagnosis and spatial faithfulness. Overall, the results underscore the need for data-quality-aware supervision design and systematic trustworthiness validation for deployable healthcare AI systems.