Unraveling Normal Anatomy via Fluid-Driven Anomaly Randomization

2025-01-23

AI Summary

Current medical image analysis models exhibit sensitivity to imaging modality, resolution, and orientation, and generalize poorly to pathological or degraded scans, limiting adaptability to clinical diversity. To address this, we propose the first modality-agnostic framework for normal brain anatomical reconstruction, supporting multi-sequence CT/MRI, anisotropic acquisitions, and direct inference on unpaired clinical images—including those with stroke—without fine-tuning. Our key contributions are: (1) a fluid dynamics–driven online lesion randomization method that generates unlimited, high-fidelity synthetic pathology samples; and (2) unified modeling of healthy and pathological anatomy, overcoming dual limitations of modality and pathology sensitivity. The framework integrates fluid dynamics priors, synthetic augmentation, self-supervised reconstruction, and unsupervised anomaly detection. Evaluated on 3D healthy and stroke-affected CT/MRI data, it achieves high-fidelity anatomical reconstruction while simultaneously localizing abnormalities, substantially enhancing large-scale analysis of unpaired clinical imaging data.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Data-driven machine learning has made significant strides in medical image analysis. However, most existing methods are tailored to specific modalities and assume a particular resolution (often isotropic). This limits their generalizability in clinical settings, where variations in scan appearance arise from differences in sequence parameters, resolution, and orientation. Furthermore, most general-purpose models are designed for healthy subjects and suffer from performance degradation when pathology is present. We introduce UNA (Unraveling Normal Anatomy), the first modality-agnostic learning approach for normal brain anatomy reconstruction that can handle both healthy scans and cases with pathology. We propose a fluid-driven anomaly randomization method that generates an unlimited number of realistic pathology profiles on-the-fly. UNA is trained on a combination of synthetic and real data, and can be applied directly to real images with potential pathology without the need for fine-tuning. We demonstrate UNA's effectiveness in reconstructing healthy brain anatomy and showcase its direct application to anomaly detection, using both simulated and real images from 3D healthy and stroke datasets, including CT and MRI scans. By bridging the gap between healthy and diseased images, UNA enables the use of general-purpose models on diseased images, opening up new opportunities for large-scale analysis of uncurated clinical images in the presence of pathology. Code is available at https://github.com/peirong26/UNA.

Problem

• Machine Learning
• Medical Image Analysis
• Performance Limitations

Innovation

• UNA
• Image Reconstruction
• Abnormal Image Recognition