# An X-Ray Is Worth 15 Features: Sparse Autoencoders for Interpretable Radiology Report Generation

arXiv: https://arxiv.org/abs/2410.03334 · HTML: https://mlsanity.com/paper/2410.03334/

## Key facts

- Cited by 25 papers in the map (data through 14 Oct 2026).
- Placed in Interpreting biomedical and science models, one of 13 directions in Mechanistic interpretability.
- Builds on 53 of its 81 references that are in the map.
- First posted on arXiv 4 Oct 2024.

## Placed in

- Mechanistic interpretability / [Interpreting biomedical and science models](https://mlsanity.com/mechinterp/interpreting-biomedical-and-science-models/)

## Builds on (most cited first, 30 of 53)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [Improved Baselines with Visual Instruction Tuning](https://mlsanity.com/paper/2310.03744/) (2023)
- [Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answering](https://mlsanity.com/paper/2209.09513/) (2022)
- [CheXpert: A Large Chest Radiograph Dataset with Uncertainty Labels and Expert Comparison](https://mlsanity.com/paper/1901.07031/) (2019)
- [Sparse Autoencoders Find Highly Interpretable Features in Language Models](https://mlsanity.com/paper/2309.08600/) (2023)
- [Video-LLaVA: Learning United Visual Representation by Alignment Before Projection](https://mlsanity.com/paper/2311.10122/) (2023)
- [LLaVA-Med: Training a Large Language-and-Vision Assistant for Biomedicine in One Day](https://mlsanity.com/paper/2306.00890/) (2023)
- [Multimodal Chain-of-Thought Reasoning in Language Models](https://mlsanity.com/paper/2302.00923/) (2023)
- [The Linear Representation Hypothesis and the Geometry of Large Language Models](https://mlsanity.com/paper/2311.03658/) (2023)
- [MIMIC-CXR-JPG, a large publicly available database of labeled chest radiographs](https://mlsanity.com/paper/1901.07042/) (2019)
- [Med-Flamingo: a Multimodal Medical Few-shot Learner](https://mlsanity.com/paper/2307.15189/) (2023)
- [Generating Radiology Reports via Memory-driven Transformer](https://mlsanity.com/paper/2010.16056/) (2020)
- [CheXbert: Combining Automatic Labelers and Expert Annotations for Accurate Radiology Report Labeling Using BERT](https://mlsanity.com/paper/2004.09167/) (2020)
- [Finding Neurons in a Haystack: Case Studies with Sparse Probing](https://mlsanity.com/paper/2305.01610/) (2023)
- [Hallucination is Inevitable: An Innate Limitation of Large Language Models](https://mlsanity.com/paper/2401.11817/) (2024)
- [RadGraph: Extracting Clinical Entities and Relations from Radiology Reports](https://mlsanity.com/paper/2106.14463/) (2021)
- [Towards Generalist Biomedical AI](https://mlsanity.com/paper/2307.14334/) (2023)
- [Learning to Exploit Temporal Structure for Biomedical Vision-Language Processing](https://mlsanity.com/paper/2301.04558/) (2023)
- [LLaVA-Plus: Learning to Use Tools for Creating Multimodal Agents](https://mlsanity.com/paper/2311.05437/) (2023)
- [Improving Dictionary Learning with Gated Sparse Autoencoders](https://mlsanity.com/paper/2404.16014/) (2024)
- [A is for Absorption: Studying Feature Splitting and Absorption in Sparse Autoencoders](https://mlsanity.com/paper/2409.14507/) (2024)
- [MAIRA-2: Grounded Radiology Report Generation](https://mlsanity.com/paper/2406.04449/) (2024)
- [Improving Factual Completeness and Consistency of Image-to-Text Radiology Report Generation](https://mlsanity.com/paper/2010.10042/) (2020)
- [Interactive and Explainable Region-guided Radiology Report Generation](https://mlsanity.com/paper/2304.08295/) (2023)
- [Exploring scalable medical image encoders beyond text supervision](https://mlsanity.com/paper/2401.10815/) (2024)

## Built on by (most cited first, 25 of 25)

- [SAeUron: Interpretable Concept Unlearning in Diffusion Models with Sparse Autoencoders](https://mlsanity.com/paper/2501.18052/) (2025)
- [Interpreting CLIP with Hierarchical Sparse Autoencoders](https://mlsanity.com/paper/2502.20578/) (2025)
- [Steering CLIP's vision transformer with sparse autoencoders](https://mlsanity.com/paper/2504.08729/) (2025)
- [Sparse Autoencoder Features for Classifications and Transferability](https://mlsanity.com/paper/2502.11367/) (2025)
- [TIDE : Temporal-Aware Sparse Autoencoders for Interpretable Diffusion Transformers in Image Generation](https://mlsanity.com/paper/2503.07050/) (2025)
- [SAEmnesia: Erasing Concepts in Diffusion Models with Supervised Sparse Autoencoders](https://mlsanity.com/paper/2509.21379/) (2025)
- [Towards Understanding the Nature of Attention with Low-Rank Sparse Decomposition](https://mlsanity.com/paper/2504.20938/) (2025)
- [CytoSAE: Interpretable Cell Embeddings for Hematology](https://mlsanity.com/paper/2507.12464/) (2025)
- [Understanding Emergent Misalignment via Feature Superposition Geometry](https://mlsanity.com/paper/2605.00842/) (2026)
- [Dimensional Collapse in Transformer Attention Outputs: A Challenge for Sparse Dictionary Learning](https://mlsanity.com/paper/2508.16929/) (2025)
- [Agglomerating Large Vision Encoders via Distillation for VFSS Segmentation](https://mlsanity.com/paper/2504.02351/) (2025)
- [Resurrecting the Salmon: Rethinking Mechanistic Interpretability with Domain-Specific Sparse Autoencoders](https://mlsanity.com/paper/2508.09363/) (2025)
- [A Unified Theory of Sparse Dictionary Learning in Mechanistic Interpretability: Piecewise Biconvexity and Spurious Minima](https://mlsanity.com/paper/2512.05534/) (2025)
- [Representations of Text and Images Align From Layer One](https://mlsanity.com/paper/2601.08017/) (2026)
- [Sparse probes and murky physics: a case study of interpretability challenges in a foundation model for continuum dynamics](https://mlsanity.com/paper/2606.11657/) (2026)
- [Insights into a radiology-specialised multimodal large language model with sparse autoencoders](https://mlsanity.com/paper/2507.12950/) (2025)
- [CXR-LanIC: Language-Grounded Interpretable Classifier for Chest X-Ray Diagnosis](https://mlsanity.com/paper/2510.21464/) (2025)
- [MedSAE: Dissecting MedCLIP Representations with Sparse Autoencoders](https://mlsanity.com/paper/2510.26411/) (2025)
- [GeoSAE: Geometric Prior-Guided Layer-Wise Sparse Autoencoder Annotation of Brain MRI Foundation Models](https://mlsanity.com/paper/2605.01829/) (2026)
- [SoftSAE: Dynamic Top-K Selection for Adaptive Sparse Autoencoders](https://mlsanity.com/paper/2605.06610/) (2026)
- [When Language Overwrites Vision: Over-Alignment and Geometric Debiasing in Vision-Language Models](https://mlsanity.com/paper/2605.08245/) (2026)
- [Correcting Influence: Unboxing LLM Outputs with Orthogonal Latent Spaces](https://mlsanity.com/paper/2605.12809/) (2026)
- [Sparse Autoencoders for Interpretable Medical Image Representation Learning](https://mlsanity.com/paper/2603.23794/) (2026)
- [Medical AI Encodes a "Feeling of Error": Verifying Cancer Segmentation via Internal Concepts](https://mlsanity.com/paper/2609.08879/) (2026)
- [Clinical Concept Centers in LLMs](https://mlsanity.com/paper/2610.02829/) (2026)

## Abstract (arXiv)

Radiological services are experiencing unprecedented demand, leading to increased interest in automating radiology report generation. Existing Vision-Language Models (VLMs) suffer from hallucinations, lack interpretability, and require expensive fine-tuning. We introduce SAE-Rad, which uses sparse autoencoders (SAEs) to decompose latent representations from a pre-trained vision transformer into human-interpretable features. Our hybrid architecture combines state-of-the-art SAE advancements, achieving accurate latent reconstructions while maintaining sparsity. Using an off-the-shelf language model, we distil ground-truth reports into radiological descriptions for each SAE feature, which we then compile into a full report for each image, eliminating the need for fine-tuning large models for this task. To the best of our knowledge, SAE-Rad represents the first instance of using mechanistic interpretability techniques explicitly for a downstream multi-modal reasoning task. On the MIMIC-CXR dataset, SAE-Rad achieves competitive radiology-specific metrics compared to state-of-the-art models while using significantly fewer computational resources for training. Qualitative analysis reveals that SAE-Rad learns meaningful visual concepts and generates reports aligning closely with expert interpretations. Our results suggest that SAEs can enhance multimodal reasoning in healthcare, providing a more interpretable alternative to existing VLMs.