Ultrasonic Image's Annotation Removal: A Self-supervised Noise2Noise Approach

TL;DR

This paper presents a self-supervised Noise2Noise method to automatically remove annotations from ultrasonic images, improving image quality for medical reports without requiring paired datasets.

Contribution

It introduces a novel self-supervised approach using Noise2Noise training for annotation removal in ultrasonic images, reducing manual effort and dataset requirements.

Findings

Models trained with Noise2Noise outperform those trained with noisy-clean pairs.

The custom U-Net achieved the best results on segmentation and reconstruction.

The approach effectively removes various annotation types from ultrasonic images.

Abstract

Accurately annotated ultrasonic images are vital components of a high-quality medical report. Hospitals often have strict guidelines on the types of annotations that should appear on imaging results. However, manually inspecting these images can be a cumbersome task. While a neural network could potentially automate the process, training such a model typically requires a dataset of paired input and target images, which in turn involves significant human labour. This study introduces an automated approach for detecting annotations in images. This is achieved by treating the annotations as noise, creating a self-supervised pretext task and using a model trained under the Noise2Noise scheme to restore the image to a clean state. We tested a variety of model structures on the denoising task against different types of annotation, including body marker annotation, radial line annotation, etc. Our results demonstrate that most models trained under the Noise2Noise scheme outperformed their counterparts trained with noisy-clean data pairs. The costumed U-Net yielded the most optimal outcome on the body marker annotation dataset, with high scores on segmentation precision and reconstruction similarity. We released our code at https://github.com/GrandArth/UltrasonicImage-N2N-Approach.