Scientists have found a way to improve early breast cancer diagnosis through the use of artificial intelligence

Scientists have found a way to improve early breast cancer diagnosis. Picture: Pexels

Scientists have found a way to improve early breast cancer diagnosis. Picture: Pexels

Published Sep 7, 2023


The World Health Organisation states that breast cancer is the most common cancer worldwide, and is the leading cause of death from cancer.

It has an impact on people of all sexes, races, and social classes. According to the 2019 National Cancer Registry (NCR), one in every 27 women is predicted to develop breast cancer throughout their lifetime.

As concerning as this is, Statistics South Africa reports that breast cancer therapy can be quite successful, with a survival probability of 40% in South Africa, especially when the disease is detected early.

The same artificial intelligence that recently made news for ChatGPT and image-generator AI apps like Remini, may be a key tool for determining a person’s risk of developing breast cancer.

An observational study titled “Comparison of Mammography AI Algorithms with a Clinical Risk Model for 5-year Breast Cancer Risk Prediction: An Observational Study” and published today in Radiology, a journal of the Radiological Society of North America (RSNA), reports that AI programmes outperformed conventional models at predicting five-year breast cancer risk.

Data from negative 2D mammograms done in 2016 at Kaiser Permanente Northern California were used by the researchers.

According to “Medical News Today”, the researchers examined 324 009 women and choose 13 628 for investigation. Additionally, 4584 people in the qualifying pool who were given a breast cancer diagnosis within five years of their initial mammogram in 2016 were still included in the study.

Until 2021, the researchers followed the subjects, according to the “New York Post”.

According to the National Institutes of Health, this shows that using AI alone or in conjunction with existing risk prediction models opens up new possibilities for predicting future risks.

According to Healthline, breast cancer specialists who were not engaged in the study praised the findings as being encouraging for healthcare professionals and their patients.

According to Liva, Dr. Andrejeva-Wright, a Yale Medicine breast imager (radiologist) and associate professor at Yale School of Medicine, the study itself, demonstrates that AI holds promise in assisting radiologists detect subtle breast cancer and potentially flagging patients who may be at increased risk of breast cancer within the next ten years.

The study also offers a fresh application for AI. It offers a fresh perspective on artificial intelligence. AI has traditionally been viewed by researchers as a tool for making discoveries.

The goal of this study is not to discover cancer there right away. It involves determining who is more likely to acquire cancer in the future. This is a pretty intriguing and significant way that AI may contribute.

With information from the study on AI and the risk of breast cancer, BioMed Central explains that an AI programme analysed the mammograms and categorise the findings into three groups.

1. Incident cancers diagnosed between 0 and 1 years pose an interval cancer risk.

2. Incident malignancies discovered within the previous one to five years pose a future cancer risk.

3. All malignancies with a risk of developing between 0 and 5 years.

Five AI algorithms were used by the researchers, including two that were developed by other researchers and three that are available commercially.

The researchers contrasted their individual risk scores with those of the Breast Cancer Surveillance Consortium (BCSC).

Radiologist and medical director of breast imaging at Memorial Care Breast Centre at Orange Coast Medical Centre in California, Dr. Richard Reitherman, provided “Medical News Today”, with an explanation of the variables used to determine the risk of breast cancer.

Reitherman observed that the BCSC, which primarily considers five factors including a woman’s age, family history of breast cancer in a first-degree relative (mother, sister, or daughter), race/ethnicity, mammographic breast density, and history of benign breast biopsies, is widely used to determine risk.

The researchers found that in forecasting breast cancer risk at 0 to 5 years, all 5 AI algorithms outperformed the BCSC.

Some algorithms identified patients who were likely to develop interval cancer, which is frequently aggressive and may call for further imaging and screening, including a second mammogram.

When the mammography revealed no malignancy, other algorithms could forecast future cancer risk up to five years in the future.

The Massachusetts Institute of Technology reports that when forecasting cancer risk in the top 10% risk group, researchers found that AI could predict up to 28% of malignancies whereas the BCSC technique could only predict 21%.

According to Dr. Laura Heacock of “Medical News Today”, this study is particularly intriguing because all but one of the AI models looked at were created to determine whether breast cancer was present or absent in a single mammography rather than to forecast a woman’s future risk of getting cancer.

Heacock, a breast radiologist at the New York University Langone Perlmutter Cancer Centre, did not participate in this study but has written prior articles on the use of AI in healthcare.

This is significant since traditional risk models for breast cancer used by doctors and other healthcare professionals often call for a lot of data, including family history, ethnicity, prior breast biopsies, pregnancy and hormone use.

Healthline reports that these AI models surpass the BCSC model in predicting women who are more likely to get cancer in the future, despite just using a single mammography test. A strong strategy that makes use of AI for personal gain is the use of AI to forecast both current and future breast cancer risk.

These kinds of AI studies demonstrate that not all dense breasts are created equal, some complex breast tissue patterns can predict a higher risk of breast cancer, even if they may be invisible to the human eye or can only be seen after training on a large number of mammograms.

Researchers think AI can spot malignancies that have gone unnoticed and breast cancer characteristics that can be used to forecast cancer development in the future.