AI Software Pinpoints Biopsies to Detect Prostate Cancer

The software is deployed via a system called SmartTarget and enables surgeons to pick up clinically relevant cancers that were missed when using current visual detection methods.

MRI-targeted biopsies, where MRI scans are used to inform surgeons where a tumor lies before they conduct a biopsy, have improved detection rates to nearly 90% from 50% in the last five years. The SmartTarget system further enhances this technique by allowing a 3D model of the prostate and cancer to be created for each patient from their MRI scans using advanced image processing and machine learning algorithms. During a biopsy, this model is fused with ultrasound images to highlight the area of concern, which otherwise does not appear in the ultrasound images, helping to guide the surgeon while conducting the procedure.

In a study, 129 people with suspected prostate cancer underwent two biopsies – one using the SmartTarget system, and one where surgeons could only visually review the MRI scans. The two strategies combined detected 93 clinically significant prostate cancers, with each of them picking up 80 of these cancers; each missed 13 that the other method picked up. According to the researchers, the surgeons’ visual review of MRI scans should be used in tandem with SmartTarget, as using this technique enables surgeons to learn to make subtle adjustments such as adapting to the movement of the patient and the prostate as the needle is inserted. The researchers say that the new methods could reduce the number of biopsies needed, and reduce the unnecessary surgeries caused by over diagnosis of less harmful cancers.

“There has been much discussion and speculation in the media recently on the degree to which computers and artificial intelligence will be integrated into clinical care. Studies such as this one are extremely important as they provide valuable evidence on the performance of a new technology in the clinical setting,” said co-senior author Professor Mark Emberton (Dean, UCL Medical Sciences). “With this study we now have hard data showing that SmartTarget is as good as a group of experts in targeting tumors in the prostate, and have a glimpse of how clinicians and computers will be working together in the future for the good of the patient.”

Related Links:
University College London