REFLECT: Rectified Flows for Efficient Brain Anomaly Correction Transport

TL;DR

REFLECT introduces a rectified flow-based framework for efficient, single-step correction of brain anomalies in MRI images, improving localization accuracy and outperforming existing unsupervised anomaly detection methods.

Contribution

The paper presents REFLECT, a novel rectified flow approach that enables direct, one-step correction of brain anomalies, reducing inference time and enhancing localization precision.

Findings

Outperforms state-of-the-art UAD methods on brain segmentation benchmarks.

Enables single-step anomaly correction, unlike diffusion models.

Provides precise anomaly localization through discrepancy detection.

Abstract

Unsupervised anomaly detection (UAD) in brain imaging is crucial for identifying pathologies without the need for labeled data. However, accurately localizing anomalies remains challenging due to the intricate structure of brain anatomy and the scarcity of abnormal examples. In this work, we introduce REFLECT, a novel framework that leverages rectified flows to establish a direct, linear trajectory for correcting abnormal MR images toward a normal distribution. By learning a straight, one-step correction transport map, our method efficiently corrects brain anomalies and can precisely localize anomalies by detecting discrepancies between anomalous input and corrected counterpart. In contrast to the diffusion-based UAD models, which require iterative stochastic sampling, rectified flows provide a direct transport map, enabling single-step inference. Extensive experiments on popular UAD brain segmentation benchmarks demonstrate that REFLECT significantly outperforms state-of-the-art unsupervised anomaly detection methods. The code is available at https://github.com/farzad-bz/REFLECT.