AI Model Better Than MRI at Detecting Tumor Margins

Detecting tumors early is a crucial step to ensure a better outcome. According to a recent study, a model of artificial intelligence (AI) created by scientists can assist medical professionals in estimating prostate cancer's stage.

When cells grow and divide more than they should or do not die when they should, a tumor develops. Identifying tumors early and starting appropriate treatment is essential since they might be malignant.

A chain of experiments conducted by UCLA's Jonsson Comprehensive Cancer Center and Department of Urology showed that the AI model was more reliable than magnetic resonance imaging (MRI) at predicting tumor margins, possibly enhancing the efficacy of targeted therapy, standardizing the definition of treatment margins, and lowering the risk of cancer recurrence.

For individuals with intermediate-risk prostate cancer, focal therapy, a less invasive procedure for localized tumors, is an alternate option. The method uses imaging guidance, such as that provided by an MRI, to precisely localize the tumor and direct the course of treatment.

Real-time imaging throughout the process enables accurate energy delivery to the targeted area while also allowing for the monitoring of therapy progress.

However, current techniques tend to underestimate prostate cancer's severity, challenging the definition of targeted treatment boundaries. These margins might be defined more precisely by AI than by MRI alone, which is essential for accurate diagnosis, precise treatment planning, and successful surgical treatments.

In collaboration with Avenda Health researchers, the team trained the Unfold-AI model to determine margins during focused treatment using biopsy data from several institutions. At the Stanford University School of Medicine, testing was done on 50 individuals who underwent radical prostatectomy for intermediate-risk malignancy.

Sensitivity was assessed using Wilcoxon signed-rank tests, and negative margin rates were evaluated using chi-square testing when comparing AI with traditional margins.

The researchers discovered that the AI model was more precise and efficient in predicting tumor margins compared to traditional approaches.

The development results from research that Leonard Marks, professor and deKernion Endowed Chair in Urology at the David Geffen School of Medicine at UCLA, began in 2009.

The program may aid radiation therapists in boosting energy delivery to the most critical areas, assist surgeons in predicting expansion into the prostate capsule, and enhance the outcomes of focused prostate cancer ablation.

The team concludes: "The AI model was accurate and effective in an independent test set. This approach could improve and standardize treatment margin definition, potentially reducing cancer recurrence rates. Furthermore, an accurate assessment of negative margin probability could facilitate informed decision-making for patients and physicians."