Researchers at UCLA Health discovered that the risk of prostate cancer can be greatly decreased by mapping the boundaries of cancerous prostate tissue using Artificial Intelligence. This advancement can ensure an accurate diagnosis, precise treatment planning, and successful surgical procedures. These findings were published in the Journal of Urology.
Scientists found that when AI was used to help with cancer contouring, the accuracy, and consistency of cancer size prediction was 45 times higher than when doctors relied solely on blood tests and traditional clinical imaging to determine the extent of cancer.
Accurately determining the extent of prostate cancer is crucial for treatment planning, as different stages may require different approaches such as active surveillance, surgery, focal therapy, radiation therapy, hormone therapy, chemotherapy, or a combination of these treatments.
Dr. Wayne Brisbane, Assistant Professor and Study Author, Department of Urology, David Geffen School of Medicine, UCLA
A surgeon must often consider Prostate-Specific Antigen (PSA) blood testing, imaging tests (MRI, CT scans), and other clinical features simultaneously to assess the extent of prostate cancer and ascertain the aggressiveness of the cancer cells.
Doctors frequently depend on an MRI's appearance to determine the size of a tumor, but according to Brisbane, the true extent of prostate cancer can be “MRI-invisible,” leading doctors to underestimate the tumor's size. AI may be able to assist in solving this difficult issue.
Focal therapy is a relatively new approach to treating prostate cancer that aims to eliminate the cancer cells while minimizing damage to surrounding healthy tissue. Researchers at UCLA and Avenda Health developed this new AI system, which has already been shown to better define the margins of prostate cancer than MRI.
This shows the potential of AI to improve minimally invasive treatment approaches. But before this study, it had not been examined how well the AI system performed while in doctors' hands.
The researchers compared cognitive and hemi-gland contouring approaches to AI-assisted contours in a multi-reader, multi-case study to assess cancer contouring and clinical decision-making of doctors with and without AI software.
Fifty patients who had received a prostatectomy but may have been candidates for focused therapy were assessed by seven urologists and three radiologists from various hospitals. Their experience ranged from 2 to 23 years.
Each case contained a report from a biopsy, outlines of the prostate gland, and locations where cancer was suspected. The exact type of MRI scan used was called T2-weighted MRI.
The doctors first examined the photos to encompass any serious disease and manually sketched borders around the potential malignant regions. They re-examined the same instances after at least four weeks, and this time, they used artificial intelligence software to help them detect the malignant spots.
The accuracy and negative margin rate of the cancer outlines created by each approach were then compared. This analysis shows whether or not all malignant tissue was detected.
The researchers discovered that physicians rarely obtained a negative margin when employing only 1.6% of the time in traditional methods. With AI support, the percentage rose to 72.8%.
We saw that doctors who used AI assistance were both more accurate and more consistent, meaning they tended to agree more when using AI assistance.
Shyam Natarajan, Assistant Adjunct Professor and Study Senior Author, Department of Urology, Surgery, and Bioengineering, UCLA
Additionally, the team discovered that the application of AI decreased variation in accurate tumor encapsulation and increased clinician recommendations for focal therapy among patients with unilateral cancer. These findings may help lower the risk of side effects frequently connected to more aggressive treatments like radiation therapy or surgery.
Overall, the use of AI in cancer treatment could lead to more effective and personalized care for patients, with treatments that are better tailored to their individual needs and more successful in fighting the disease.
Dr. Wayne Brisbane, Assistant Professor and Study Author, Department of Urology, David Geffen School of Medicine, UCLA
The National Cancer Institute at the National Institutes of Health partly supported the research.
Sakina Mohammed Mota and Alan Priester, from Avenda Health, are the study's co-first authors. James Sayre from UCLA, Joshua Shubert, Jeremey Bong, and Brittany Berry-Pusey from Avenda Health also contributed to the study.
Journal Reference:
Mota, M S., et al. (2024) Artificial Intelligence Improves the Ability of Physicians to Identify Prostate Cancer Extent. Journal of Urology. doi.org/10.1097/JU.0000000000003960