CHUM researchers refine the diagnosis of prostate cancer

the demolition of the North Hatley bridge requires long detours by car

MONTREAL – Prostate cancer is the most frequently diagnosed form in men in North America and yet its screening method remains rudimentary. However, a team from the CHUM Research Center (CRCHUM) and Polytechnique Montréal believe they have developed an improved method to detect the most aggressive forms of the disease.

According to the press release published on Tuesday by the CRCHUM, the method combining Raman microspectroscopy imaging and artificial intelligence would make it possible to obtain precise diagnostic results of prostate cancer in nearly 9 out of 10 cases, and a rate even higher to identify the more aggressive form called ‘intracanal carcinoma of the prostate’.

As part of this study published last Friday in PLoS Medicine, researchers Dre Dominique Trudel and Frédéric Leblond as well as postdoctoral researcher Andrée-Anne Grosset wanted to facilitate the identification of patients at risk and improve the diagnostic capacity of doctors.

According to data provided by the CRCHUM, approximately 20% of the approximately 4,200 Canadians who die of prostate cancer each year have this intracanal carcinoma. Despite these high figures, doctors can only detect this variant by “a visual observation of the tissues removed”, one explains.

Researchers believe they have finally solved this problem and developed a way to detect this dangerous form of the disease.


To complete their study, the researchers analyzed tissue samples from 483 patients with prostate cancer. These samples were collected at the CHUM, the Center hospitalier universitaire de Québec (CHUQ) and the University Health Network in Toronto.

The procedure consisted of identifying the molecular signature of each sample using Raman microspectroscopy imaging, a specialty of Frédéric Leblond, professor in the Department of Engineering Physics at Polytechnique Montreal.

This involves using light rays to bombard the molecules of a sample, then analyzing their movements to identify the chemical bonds that compose it.

This data was then used to train algorithms to enable artificial intelligence to recognize and classify molecular signatures of healthy tissue and different forms of prostate cancer.

Eventually, the researchers hope that their technique will improve diagnostic accuracy and prevent patients from undergoing unnecessary treatment in the event of the development of prostate cancer deemed harmless.

Their method would also be shorter and less expensive than the current means available to health professionals.

Ugo Giguère, The Canadian Press


Please enter your comment!
Please enter your name here