AI Assisted Prostate Cancer Treatment

A common and effective treatment for prostate cancer is radiation therapy. Its success hinges on precisely targeting cancer cells while avoiding damage to nearby organs like the bladder and bowel. To ensure accuracy, clinicians perform daily image guided radiation therapy, comparing a new CT scan against the original treatment plan before every session.

This daily review is labour intensive and time consuming. Any significant differences between the plan and the patient's current state require additional procedures, causing delays that are stressful for the patient and can compromise the accuracy of the treatment. There is a clear need for a system that can streamline this process, helping clinicians make faster, more confident decisions.

In collaboration with QLD Health radiation therapists, we are building an AI assistant to optimise the daily review process. The system uses cutting edge machine learning to automatically segment the prostate and surrounding organs on the daily, lower quality CT scans. It then provides clinicians with easily interpreted, interactive information, highlighting any deviations from the treatment plan.

This is not a fully automated system; the clinician retains the final decision. The AI acts as a powerful support tool, providing the right information at the right time. Explore the interactive dashboard simulation below to see how it works.

The goal of this research is to improve treatment times for patients without compromising the quality of care. By enabling faster decisions, our AI assistant helps increase the accuracy of treatments, as there is less time for the patient's body to change while on the treatment bed.

The system allows radiation therapists to focus more of their attention on patients whose bodies are diverging significantly from their treatment plan, reducing workplace stress and leading to better overall outcomes. By being an "AI assisted" tool rather than a fully automated one, our approach bypasses the current safety and practicality concerns that have slowed the adoption of fully automated solutions in clinical practice.