New Research Finds That Artificial Intelligence (AI) Performs Comparably To Human Doctors In Detecting Breast Cancer
AI could be on the verge of helping doctors accurately diagnose breast cancer. A recent study suggests that machine learning algorithms are just as reliable as medical experts when it comes to reading mammogram results.
In partnership with the National Health Service’s Breast Cancer Screening Program, researchers found that the AI tech was slightly better than human experts in identifying breast cancer from a set of 120 mammogram tests.
So, there’s a growing sense of optimism among scientists that ongoing research could make AI a regular fixture in breast cancer screening, aiding doctors in their assessments.
In a perfect world, two different readers would analyze mammograms to reduce the risk of false positives. However, the shortage of radiologists often makes this dual-review process impractical.
A UK-based study published in the Radiological Society of North America’s journal, Radiology, pitted a readily available AI algorithm against human experts from the NHS to test the interpretation of mammograms.
The screening process entails a mammographer capturing several X-ray images of each breast to spot potential indicators of breast cancer that may be too small for manual detection. In the framework of the NHS Breast Screening Program (NHSBSP), women usually get their first invitation for a mammogram between ages 50 and 53. After that, they are invited for follow-up screenings every three years up to the age of 70.
However, this approach isn’t foolproof in catching every instance of breast cancer. False positives can lead to additional, unnecessary imaging or even biopsies.
A study from the University of Nottingham suggests that double-reading mammogram results can increase cancer detection rates by 6 to 15 percent while keeping recall rates low. But implementing such a practice is tough, given the global shortage of skilled interpreters.
This study, led by Professor Yan Chen, used test samples from the Personal Performance in Mammographic Screening (PERFORMS) quality assurance tool, which the NHSBSP uses. Each PERFORMS test includes 60 intricate mammogram exams from the NHSBSP, featuring a mix of benign, normal, and abnormal results. The team compared these readings to the outcomes generated by AI.
Ruan J/peopleimages.com – stock.adobe.com- illustrative purposes only, not the actual person
The researchers also further assessed the AI’s effectiveness using two consecutive sets of PERFORMS tests, amounting to 120 screening mammograms. They evaluated the AI algorithm’s performance using this same dataset.
When the team measured the AI’s test results against those of 552 human assessors, which included 315 board-certified radiologists, 206 radiographers, and 31 breast clinicians, they found the performance to be almost identical. The human evaluators had an average sensitivity rate of 90 percent and a specificity rate of 76 percent. The AI edged them out slightly, registering a sensitivity of 91 percent and a specificity of 77 percent.
“There is a lot of pressure to deploy AI quickly to solve these problems, but we need to get it right to protect women’s health. The 552 readers in our study represent 68% of readers in the NHSBSP, so this provides a robust performance comparison between human readers and AI,” explained Yan Chen.
“The results of this study provide strong supporting evidence that AI for breast cancer screening can perform as well as human readers.”
Still, Yan Chen claimed that more research is needed before AI can be introduced as a second reader for screenings– pointing out how changes in operating environments can impact algorithms and how performance can drift as time goes on.
“I think it is too early to say precisely how we will ultimately use AI in breast screening. The large prospective clinical trials that are ongoing will tell us more. But no matter how we use AI, the ability to provide ongoing performance monitoring will be crucial to its success,” Yan Chen concluded.
“It’s vital that imaging centers have a process in place to provide ongoing monitoring of AI once it becomes part of clinical practice. There are no other studies to date that have compared such a large number of human reader performance in routine quality assurance test sets to AI, so this study may provide a model for assessing AI performance in a real-world setting.”
To read the study’s complete findings, visit the link here.
If true crime defines your free time, this is for you: join Chip Chick’s True Crime Tribe
She’s Talking About What Dating Was Like For Her Growing Up Amish
Get Boo-zy With These Beautiful Halloween Wine Glasses
How To Grow And Care For Black Huckleberries, Which Are Perfect For Making Baked Goods And Jams
Sign up for Chip Chick’s newsletter and get stories like this delivered to your inbox.
More About:Science