Use of software improves sensitivity of radiologists, reduces number of false-positive findings
THURSDAY, Nov. 21, 2019 (HealthDay News) — Using deep convolutional neural network (DCNN) software can improve detection of malignant pulmonary nodules on chest radiographs, according to a study published online Nov. 12 in Radiology.
Yongsik Sim, M.D., from Yonsei University in Seoul, South Korea, and colleagues examined the performance of radiologists in detecting malignant pulmonary nodules on chest radiographs when assisted by deep learning-based DCNN software. A total of 600 lung cancer-containing chest radiographs and 200 normal chest radiographs were identified. The chest radiographs were analyzed by 12 radiologists from four medical centers. Deep learning-based computer-aided detection software separately trained, tested, and validated with 19,330 radiographs was used to identify suspicious nodules. The images were then reviewed by the radiologists using the DCNN software.
The researchers found that when the radiologists re-reviewed radiographs with the DCNN software, the average sensitivity of radiologists improved from 65.1 to 70.3 percent and the number of false-positive findings per radiograph decreased from 0.2 to 0.18. Using the DCNN software, 104 of 2,400 radiographs were positively changed, and 56 of 2,400 radiographs were negatively changed for the 12 radiologists in the study.
“Computer-aided detection software to detect lung nodules has not been widely accepted and utilized because of high false-positive rates, even though it provides relatively high sensitivity,” a coauthor said in a statement. “DCNN may be a solution to reduce the number of false positives.”
Several authors disclosed financial ties to Samsung Electronics, which funded the study and provided the computer-aided detection system used in the study.
Copyright © 2019 HealthDay. All rights reserved.
