Automated detection and segmentation of pleural effusion on ultrasound images using an Attention U‐net

• Published in: Journal of applied clinical medical physics Vol. 25; no. 1; pp. e14231 - n/a
• Main Authors: Huang, Libing, Lin, Yingying, Cao, Peng, Zou, Xia, Qin, Qian, Lin, Zhanye, Liang, Fengting, Li, Zhengyi
• Format: Journal Article
• Language: English
• Published: United States John Wiley & Sons, Inc 01.01.2024; John Wiley and Sons Inc
• Subjects: Abdomen; an Attention U‐net; Annotations; Area Under Curve; Artificial Intelligence; Ascites; Automation; Datasets; Deep learning; Emergency medical care; Humans; Medical Imaging; Medical prognosis; Pleural effusion; Pleural Effusion - diagnostic imaging; ROC Curve; segmentation; Set design; Trauma; Ultrasonic imaging; Ultrasonography; ultrasound; deep learning; segmentation; pleural effusion; an Attention U-net; ultrasound
• ISSN: 1526-9914; 1526-9914
• DOI: 10.1002/acm2.14231

Background Ultrasonic for detecting and evaluating pleural effusion is an essential part of the Extended Focused Assessment with Sonography in Trauma (E‐FAST) in emergencies. Our study aimed to develop an Artificial Intelligence (AI) diagnostic model that automatically identifies and segments pleural effusion areas on ultrasonography. Methods An Attention U‐net and a U‐net model were used to detect and segment pleural effusion on ultrasound images of 848 subjects through fully supervised learning. Sensitivity, specificity, precision, accuracy, F1 score, the receiver operating characteristic (ROC) curve, and the area under the curve (AUC) were used to assess the model's effectiveness in classifying the data. The dice coefficient was used to evaluate the segmentation performance of the model. Results In 10 random tests, the Attention U‐net and U‐net ’s average sensitivity of 97% demonstrated that the pleural effusion was well detectable. The Attention U‐net performed better at identifying negative images than the U‐net, which had an average specificity of 91% compared to 86% for the U‐net. Additionally, the Attention U‐net was more accurate in predicting the pleural effusion region because its average dice coefficient was 0.86 as opposed to the U‐net's average dice coefficient of 0.82. Conclusions The Attention U‐net showed excellent performance in detecting and segmenting pleural effusion on ultrasonic images, which is expected to enhance the operation and application of E‐FAST in clinical work.