AI Improves Ultrasound Imaging Accuracy and Efficiency

Advances in deep learning (DL) and other artificial intelligence (AI) methodologies are demonstrating their potential to help health systems improve access to high-quality healthcare services. In China, Zhejiang University (ZJU) and Zhejiang DE Image Solutions Co., Ltd. have partnered with Intel to train DL models and deploy them on a DL Inferencing Solution to analyze ultrasound images of thyroids. Depending on the experience level of the radiologist, this solution has the potential to act as a preliminary screening tool to improve both speed and accuracy of diagnosis.

Challenge

The incidence of thyroid cancer has increased dramatically in recent years. Ultrasound imaging is the most common diagnostic method for thyroid nodules and biopsy, a painful and expensive procedure, is often required to determine whether they are cancerous. The number of radiologists, particularly in developing countries like China, available to read these images have not kept up with demand; and these available radiologists are overloaded and overworked which can lead to fatigue and decreased accuracy in analyses.