Scale-Space Autoencoders for Unsupervised Anomaly Segmentation in Brain MRI

TL;DR

This paper introduces a scale-space autoencoder that uses a Laplacian pyramid to improve unsupervised anomaly segmentation in brain MRI, especially for small lesions, by enhancing reconstruction fidelity across multiple scales.

Contribution

The novel approach learns to compress and reconstruct different frequency bands of brain MRI with a Laplacian pyramid, improving anomaly detection and segmentation at multiple scales.

Findings

Enhanced segmentation accuracy over state-of-the-art methods.

More precise and crisp reconstructions at native resolution.

Effective detection of various lesion sizes using a single model.

Abstract

Brain pathologies can vary greatly in size and shape, ranging from few pixels (i.e. MS lesions) to large, space-occupying tumors. Recently proposed Autoencoder-based methods for unsupervised anomaly segmentation in brain MRI have shown promising performance, but face difficulties in modeling distributions with high fidelity, which is crucial for accurate delineation of particularly small lesions. Here, similar to these previous works, we model the distribution of healthy brain MRI to localize pathologies from erroneous reconstructions. However, to achieve improved reconstruction fidelity at higher resolutions, we learn to compress and reconstruct different frequency bands of healthy brain MRI using the laplacian pyramid. In a range of experiments comparing our method to different State-of-the-Art approaches on three different brain MR datasets with MS lesions and tumors, we show improved anomaly segmentation performance and the general capability to obtain much more crisp reconstructions of input data at native resolution. The modeling of the laplacian pyramid further enables the delineation and aggregation of lesions at multiple scales, which allows to effectively cope with different pathologies and lesion sizes using a single model.