Expert-Agnostic Ultrasound Image Quality Assessment using Deep Variational Clustering

TL;DR

This paper introduces US2QNet, an unsupervised deep learning framework that assesses ultrasound image quality without manual annotations, using variational autoencoders to cluster and visualize quality features.

Contribution

It presents a novel unsupervised approach for ultrasound image quality assessment that eliminates the need for labor-intensive manual annotations and noisy labels.

Findings

Achieved 78% accuracy in quality assessment of urinary bladder ultrasound images.

Outperformed state-of-the-art clustering methods in experiments.

Effectively visualized quality feature clusters in 2D space.

Abstract

Ultrasound imaging is a commonly used modality for several diagnostic and therapeutic procedures. However, the diagnosis by ultrasound relies heavily on the quality of images assessed manually by sonographers, which diminishes the objectivity of the diagnosis and makes it operator-dependent. The supervised learning-based methods for automated quality assessment require manually annotated datasets, which are highly labour-intensive to acquire. These ultrasound images are low in quality and suffer from noisy annotations caused by inter-observer perceptual variations, which hampers learning efficiency. We propose an UnSupervised UltraSound image Quality assessment Network, US2QNet, that eliminates the burden and uncertainty of manual annotations. US2QNet uses the variational autoencoder embedded with the three modules, pre-processing, clustering and post-processing, to jointly enhance, extract, cluster and visualize the quality feature representation of ultrasound images. The pre-processing module uses filtering of images to point the network's attention towards salient quality features, rather than getting distracted by noise. Post-processing is proposed for visualizing the clusters of feature representations in 2D space. We validated the proposed framework for quality assessment of the urinary bladder ultrasound images. The proposed framework achieved 78% accuracy and superior performance to state-of-the-art clustering methods.