news

New AI model may enable more targeted prostate cancer treatment

Posted: 30 October 2024 | | No comments yet

Based on MRI scans from 732 prostate cancer patients, the AI model identified the edges of 85 percent of the most radiologically aggressive lesions.

prostate cancer

Researchers at Mass General Brigham have trained and validated an AI model based on MRI scans of hundreds of prostate cancer tumours. It was able to identify the edges of 85 percent of the most radiologically aggressive lesions.

In the UK, prostate cancer is the most common cancer in men, and on average, 52,000 men are diagnosed with the disease each year.1 To ensure clinicians make more accurately informed treatment decisions for patients, a consistent method of estimating prostate cancer size is crucial.

Although MRI has improved the diagnosis of prostate cancer, estimates of tumour size by human clinicians can vary from person-to-person. Therefore, in the new study, the team trained an AI model based on MRI images of prostate cancer tumours from 732 patients undergoing treatment at a single centre. Following this, the AI model’s size estimates were linked to treatment success in the five to 10 years after diagnosis.

 

Access your FREE copy

 


Biomarkers aren’t just supporting drug discovery – they’re driving it

FREE market report

From smarter trials to faster insights, this report unpacks the science, strategy and real-world impact behind the next generation of precision therapies.

What you’ll unlock:

  • How biomarkers are guiding dose selection and early efficacy decisions in complex trials
  • Why multi-omics, liquid biopsy and digital tools are redefining the discovery process
  • What makes lab data regulatory-ready and why alignment matters from day one

Explore how biomarkers are shaping early drug development

Access the full report – it’s free!

 

The researchers demonstrated that the AI model could locate and measure around 85 percent of prostate tumours that had a Prostate Imaging Reporting and Data System (PI-RADS) score of 5 within the patient cohort. This score indicates a very high risk of clinically significant prostate cancer.

Also, the model’s size estimates showed potential as a prognostic marker. For patients treated surgically or with radiation therapy, larger tumours were associated with higher risk that prostate cancer would come back, as measured by blood levels of prostate-specific antigen (PSA).   

Martin King, of the Department of Radiation Oncology at the Brigham, said: “The AI measurement itself can tell us something additional in terms of patient outcomes…For patients, this can really tell them something about what are the chances of cure, and the likelihood that their cancer will reoccur or metastasise in the future.” 

Furthermore, the AI model could guide radiation oncologists by pinpointing the tumour’s focal region, enabling more targeted treatment. One of its advantages includes that it is a quicker test compared to existing methods of predicting prostate cancer aggressiveness which normally take around two weeks to yield results. Therefore, this means that patients could begin treatment sooner. 

Moving forward, the team plan to assess their model with a larger, multi-institutional dataset. David Yang, of the Department of Radiation Oncology at Brigham and Women’s Hospital, a founding member of the Mass General Brigham healthcare system, concluded: “We want to validate our findings, using other institutions and patient cohorts with different disease characteristics, to make sure that this approach is generalisable to all patients.”   

This study was published in Radiology.

References

1 Facts and figures. Prostate Cancer UK [cited 2024 October 29]. Available from: https://prostatecanceruk.org/prostate-information-and-support/risk-and-symptoms/about-prostate-cancer#:~:text=Prostate%20cancer%20is%20the%20most,prostate%20cancer%20in%20their%20lifetime.