This AI Software Can Predict Breast Cancer Risk Quickly and Accurately
In order to diagnose breast cancer doctors use a range of tests including mammograms, ultrasound scan, fine needle aspiration and core biopsy of the breast.
Now scientists have developed an artificial intelligence (AI) software that they claim could assist doctors to predict breast cancer risk quickly and accurately without the need of unnecessary breast biopsies.
If the developers of AI software from Houston Methodist Research Institute in the United States are to be believed then their computer software can interpret mammogram results at least 30 times quicker than doctors and that too with 99 percent accuracy.
In the study, the AI software found to intuitively translate patient charts into diagnostic information for human review more rapidly, thereby offering doctors a time-saving support and reliable diagnostic information to assess patient cancer risk quicker and the need for further tests accurately.
“This software intelligently reviews millions of records in a short amount of time, enabling us to determine breast cancer risk more efficiently using a patient’s mammogram. This has the potential to decrease unnecessary biopsies,” said Stephen T Wong, chair of the Department of Systems Medicine and Bioengineering at Houston Methodist Research Institute in the US.
To determine the efficacy and accuracy of the software, Wong and his team from Houston Methodist evaluated mammograms and pathology reports of 500 breast cancer patients using the AI software.
The researchers found that AI successfully assessed patient charts for cancer risk in just a few hours for the entire group. Manual review of just 50 patient, conversely, needs two clinicians between 50 and 70 hours. AI, on the other hand, reviewed 500 charts in a few hours, saving hundreds or thousands of physician hours.
“Accurate review of this many charts would be practically impossible without AI,” says Wong.
Until now, when a woman’s mammogram shows something abnormal, she is often recommended for breast tissue biopsies. Alone in the US, more than 1.6 million breast biopsies are performed every year, and the American Cancer Society estimates that approximately 20 percent of these biopsies are unnecessarily performed due to false-positive mammogram results.
As per the recent statistics from the Centers for Disease Control and Prevention (CDC) and American Cancer Society, up to fifty percent of the 12.1 million mammograms performed annually in the United States result in false positives for cancer free breasts.
Wong and his team hope their AI software will help physicians better assess the cancer risk and help them define percent risk requiring a breast biopsy, as well as help decrease unnecessary breast biopsies.
The study findings are published in Cancer Journal.