Deep-learning platform detects malignant nodules on x-ray, beating radiologists
A deep learning-based automatic detection algorithm (DLAD) detected malignant pulmonary nodules on chest x-rays better than radiologists, according to a Sept. 25 study in Radiology.
The DLAD was trained using 43,292 chest x-rays from more than 34,000 patients performed between 2010 and 2015. Its accuracy was compared to 18 physicians, including nine board-certified radiologists.
Attaining a higher area under the receiver operating characteristic curve (0.92-0.99) and jackknife alternative free-response receiver-operating characteristic (0.831-0.924), the DLAD beat 17 of 18 and 15 of 18 physicians, respectively, on both validation benchmarks.
When used as a second reader, physicians reported improved performance in detecting nodules.
“Our study results demonstrated that DLAD could accurately detect malignant pulmonary nodules on chest radiographs with better performance than that of physicians, and that it enhanced performance of physicians when used as a second reader,” wrote co-lead author Ju Gang Nam, Seoul National University Hospital and College of Medicine in the Republic of Korea, and colleagues.
Additionally, the authors found DLAD read 100 percent of high conspicuity nodules, most large nodules and more in overlapped areas compared to the four physician groups involved in the study.
Despite its accuracy, the DLAD could not detect small (<1 cm) and less conspicuous nodules, the authors reported. This is because, according to Nam et al., the algorithm was trained under supervision, and with information provided by radiologists—those limitations are embedded within the DLAD itself.
“With further optimization from focused training by using retrophrenic nodules of both the posterior-anterior and lateral views, we expect the performance of DLAD to improve,” the authors concluded.
Matt joined Chicago’s TriMed team in 2018 covering all areas of health imaging after two years reporting on the hospital field. He holds a bachelor’s in English from UIC, and enjoys a good cup of coffee and an interesting documentary.