University of Calgary Develops Software for Early Cancer Diagnosis

Posted by Alan Zunec on Friday, April 6th, 2012 at 11:16am.

University of Calgary researchers have come up with a software program that can detect breast cancer even before a tumour forms. Dr. Leo Desautels, a radiologist, and engineers at the Schulich School of Engineering were responsible for the design.

The study included looking at 106 mammogram images that had already been deemed normal, but then later on the women involved all developed breast cancer. The software was able to take another look at the images, focusing on suspicious areas that were previously missed. The images were taken, on average, 15 months before there were noticeable signs or lumps found leading to a diagnosis.

Raj Rangayyan, a lead researcher, called these areas architectural distortions. In a healthy breast, all the tissues, blood vessels, ducts and ligament lead towards the nipple. Cancer causes a breakdown of this order and tissues move in different directions. It is even difficult for a trained radiologist to see these distortions, especially with the naked eye. Computer assisted diagnostics also tends to miss the distortions because they are very subtle. This means that between 12 and 45 percent of potential breast cancer cases are overlooked.

The achievement received praise from Dr. Steven Narod from the University of Toronto’s Women’s College Research Institute. The professor noted that current mammogram screenings are not foolproof. Going to an MRI is possible, but expensive. Canada’s Institute for Cancer Research also awarded one of the researchers, PhD student Shantanu Banik, with a Publication Prize. Banik specializes in biomedical engineering.

Leave a Comment