Global Editions

Google Can Help Detect Breast Cancer Using Artificial Intelligence

by Mahrukh Sarwar

Artificial Intelligence, also known as AI, has been making leaps and bounds especially in the healthcare domain. Now Google is using AI to review patients biological tissue samples in order to detect breast cancer.

The results of this technology have been revealed to be cutting edge.

According to a Google Research blog published on March 03, 2017, the company is using a branch of AI called deep learning, which can recognize patterns in a huge number of digital representations of images, to detect signs of cancer in breast cancer biopsies. Google has used deep learning to examine thousands of slides of cancer cells provided by the Radboud University Medical Center in the Netherlands to create an automated detection algorithm.

Pathologists have to go over a large number of pathology slides every day, which requires years of training and experience but different doctors can examine the same patient and still make varying diagnoses, which can result in a misdiagnosis. An investigation by The Journal of the American Medical Association reveals that there is only 48% agreement in diagnosis for certain forms of breast cancer while an article in the Journal of Urology states that there are similar low levels of concordance in the diagnosis of prostate cancer.

This technology will be able to assist doctors in making more accurate diagnoses and better treating patients. The algorithm can help improve the efficiency and consistency of pathologists, including reducing their false negative rate, and enabling them to measure the tumor size more accurately which is important for good prognosis.

Also Read: The Tech Healthcare Revolution Pakistan Needs

A closeup of a lymph node biopsy. The tissue contains a breast cancer metastasis as well as macrophages, which look similar to the tumor but are benign normal tissue. Google’s algorithm successfully identifies the tumor region (bright green) and is not confused by the macrophages. Courtesy – Google Research Blog

The Google Research blog post says that the localization score (FROC) for the algorithm was 89%, which was significantly better than the 73% achieved by a pathologist with no time constraint.

However, this software is not meant to replace doctors completely but complement their work while they go over the large amount of information that is needed to make a diagnosis for a patient. The algorithms can only do the tasks that they have been designed for and will not be able to detect other medical issues such as autoimmune disease, inflammatory process, or other cancers since they lack the knowledge and experience of human medical professionals.

This technology could be of vital use in countries like Pakistan which has the highest incidence of breast cancer in Asia. According to experts, 40,000 women die of breast cancer annually in the country and one in nine women will develop breast cancer at some stage of their life.


Related posts