Label-Free Segmentation of COVID-19 Lesions in Lung CT

TL;DR

This paper introduces a label-free, anomaly modeling approach for segmenting COVID-19 lesions in lung CT scans, reducing the need for annotated data by synthesizing lesions and training a normalcy-converting network.

Contribution

The novel method synthesizes lesions and trains NormNet to segment COVID-19 lesions without requiring annotated lesion data, outperforming existing unsupervised methods.

Findings

NormNet outperforms various unsupervised anomaly detection methods.

Synthesized lesions effectively train the model without manual annotations.

The approach is validated on three different datasets.

Abstract

Scarcity of annotated images hampers the building of automated solution for reliable COVID-19 diagnosis and evaluation from CT. To alleviate the burden of data annotation, we herein present a label-free approach for segmenting COVID-19 lesions in CT via pixel-level anomaly modeling that mines out the relevant knowledge from normal CT lung scans. Our modeling is inspired by the observation that the parts of tracheae and vessels, which lay in the high-intensity range where lesions belong to, exhibit strong patterns. To facilitate the learning of such patterns at a pixel level, we synthesize `lesions' using a set of surprisingly simple operations and insert the synthesized `lesions' into normal CT lung scans to form training pairs, from which we learn a normalcy-converting network (NormNet) that turns an 'abnormal' image back to normal. Our experiments on three different datasets validate the effectiveness of NormNet, which conspicuously outperforms a variety of unsupervised anomaly detection (UAD) methods.