New prostate cancer prediction tool has unmatched accuracy

A novel machine-learning framework that distinguishes between low- and high-risk prostate cancer with more precision than ever before has been developed by researchers, including one of Indian origin.

The framework, described in the journal Scientific Reports, is intended to help physicians -- in particular, radiologists -- more accurately identify treatment options for prostate cancer patients, lessening the chance of unnecessary clinical intervention.

"By rigorously and systematically combining machine learning with radiomics, our goal is to provide radiologists and clinical personnel with a sound prediction tool that can eventually translate to more effective and personalised patient care," said Gaurav Pandey, an assistant professor at the Icahn School of Medicine in the US.

"The pathway to predicting prostate cancer progression with high accuracy is ever improving, and we believe our objective framework is a much-needed advancement," Pandey said.

Prostate cancer is one of the leading causes of cancer death in American men, second only to lung cancer, said researchers, including those from the University of Southern California (USC) in the US.

While recent advances in prostate cancer research have saved many lives, objective prediction tools have, until now, remained an unmet need, they said.

Presently, the standard methods used to assess prostate cancer risk are multiparametric magnetic resonance imaging (mpMRI), which detects prostate lesions, and the Prostate Imaging Reporting and Data System, version 2 (PI-RADS v2), a five-point scoring system that classifies lesions found on the mpMRI.

Together, these tools are intended to soundly predict the likelihood of clinically significant prostate cancer.

However, PI-RADS v2 scoring is subjective and does not distinguish clearly between intermediate and malignant cancer levels (scores 3, 4, and 5), often leading to differing interpretations among clinicians.

Combining machine learning with radiomics -- a branch of medicine that uses algorithms to extract large amounts of quantitative characteristics from medical images -- has been proposed as an approach to remedy this drawback.

However, other studies have only tested a limited number of machine learning methods to address this limitation.

In contrast, the researchers developed a predictive framework that rigorously and systematically assessed many such methods to identify the best-performing one.

The framework also leverages larger training and validation data sets than previous studies did.

As a result, researchers were able to classify patients' prostate cancer with high sensitivity and an even higher predictive value.