Research to help detect breast cancer in women
Posted: July 13, 2016 6:00 a.m.
(l-r) Drs. Esam Hussein, Wei Peng and Rene V. Mayorga used a computer-aided diagnosis based on Computational Intelligence and Soft Computing techniques to develop an automated confirmatory system for analysis of mammograms. Photo by Rae Graham - U of R Photography
|
A research team from the Faculty of Engineering and Applied Science has devised an “Automated Confirmatory System” to help radiologists detect breast cancer in women. The research team includes Dr. Wei Peng, Dr. Rene V. Mayorga, and Dr. Esam Hussein.
Results of the study have been published in Computer Methods and Programs in Biomedicine, a science journal that encourages the application of computing methods in the fields of biomedical and medical practice.
Breast cancer is the most common cancer among women. However, as the paper notes, “the accuracy and diagnosis depends on both the quality of the mammographic images and the ability of the radiologist to interpret those images.”
The U of R research team uses computer-aided diagnosis based on Computational Intelligence and Soft Computing techniques to develop an automated confirmatory system for analysis of mammograms.
The team worked with two independent data sets from which large numbers of random mammogram images were extracted to train, test, and validate the developed system. The researchers report a diagnostic accuracy of at least 92 per cent.
“Our system can provide a diagnosis that can serve radiologists as a second opinion,” says Dr. Mayorga. “It confirms radiologists’ diagnosis. If the radiologist diagnosis and our system diagnosis are not in agreement, then a second look must be taken. This avoids misdiagnosing by incorrect human judgement. Our system also provides clear filtered-segmented and contrast-enhanced images to radiologists for further physical diagnosis.”
As stated in the research paper, “the development of CAD (computer-aided diagnosis) in mammography is still in an early stage, and most of the suggested approaches of CAD systems follow the two-step scheme of pixel-level detection and region-level classification.”
The U of R team adapted the “two-step scheme” to other technologies by a “hybrid approach.” The team combined “image processing techniques, pattern-recognition methods, and Computational Intelligence and Soft Computing techniques” which produced high-quality images to help with diagnosis.
“Our system uses some cutting edge technologies,” says Dr. Peng. “Our tests were done on two independent databases. More databases could be used and as more data becomes available, the system should enhance its analysis and diagnosis capabilities to increase its accuracy. The system also will need to be tailored for routine daily use”.
An important characteristic of the developed system, is that it can be implemented in a small personal computer (PC) such as a laptop.
There is still more work to be done here. Now researchers are working on other computer aided diagnosis to assist radiologists in detecting other cancers.
“Our system can be adapted to detect other cancers,” says Dr. Hussein. “However, different cancers have different features shown in images. Our next project aims at looking at brain tumours. In more complicated structures of the body that may not be as easy, but it is an area of research worth pursuing.”
The team emphasizes that the aim here is not to replace radiologists and other medical specialists, but to help them.
“We are not aiming at replacing human expert judgement,” says Dr. Hussein. “We are simply providing a confirmatory tool to augment that judgement.”
The research was supported by two grants from the Natural Sciences and Engineering Research Council of Canada.
Research Excellence
Professor receives highest honour from Gerontological Society
Ancient bird wings perfectly preserved in amber
Professor brings to life works of influential political thinker
U of R researchers receive almost $1.5 million in federal funding