AI assistance drops prostate MRI read times by 56%
Artificial intelligence software can reduce false positive reports of clinically significant prostate cancer (csPCa) while also reducing radiologist read times.
Using a proprietary deep learning-based AI software, radiologists were able to cut their read times in half when interpreting multiparametric MRI scans of patients with suspected prostate cancer, decreasing from 423 to 185 seconds per exam. What’s more, use of the AI software also resulted in improved sensitivity and specificity, according to new work on the research published in Insights into Imaging.
Corresponding author of the new paper Xiaoying Wang, with the Department of Radiology at Peking University First Hospital in China, and colleagues noted that their findings are significant due to the way the team tested the software; a total of 11 different MRI systems from three different institutions were used for the study, demonstrating the software’s reproducibility and real-world utility—something that has been lacking in prior studies on CAD systems.
“CAD systems can enhance radiologists' diagnostic performance and reduce interpretation inconsistencies. Many studies suggested that AI-based CAD systems have potential clinical utility in csPCa detection. However, the performances of CAD systems reported in these studies may be data set-specific, and their generalization, that is, performance on outside data sets, has not been well studied,” the authors explained.
For the study, the team used 480 multiparametric MRI images with a total of 349 csPCa lesions in 180 cases. Sixteen radiologists with varying experience levels from four hospitals participated, interpreting scans both with and without use of the software and then reading the same scans again four weeks later in switched mode.
Use of the software improved sensitivity from 40.1% to 59.0% and specificity from 57.7% to 71.7%. Its use also reduced reading times by 56.3% and improved diagnostic confidence scores from 3.9 to 4.3.
“The strength of our study is that the external data were collected from three different medical institutions,” the team noted. “The mpMRI images were acquired using a total of 11 different MR devices with some variation in scan parameters. Thus, the data are very heterogeneous, which is a challenging task for AI algorithms.”
While the study did have its limitations, the authors maintained that their results are indicative of how the AI software could benefit patients and providers in real-world clinical scenarios.
The study abstract is available here.
In addition to her background in journalism, Hannah also has patient-facing experience in clinical settings, having spent more than 12 years working as a registered rad tech. She joined Innovate Healthcare in 2021 and has since put her unique expertise to use in her editorial role with Health Imaging.