Artificial intelligence and machine learning systems continue to be adopted in an ever-widening range of healthcare applications, such as assisting physicians in diagnosing medical images. Able to understanding x-rays and quickly generate MRIs – sometimes even capable of specific cases of COVID – these systems have also been shown to be effective in detecting the first signs of breast cancer that radiologists might otherwise miss. Google and IBM, as good as medical centers and university research teams around the world, have all sought to develop such cancer detection algorithms.
They can spot disturbing bumps as well as radiologists can and predict “significantly” the future onset of the disease than the humans who trained them. However, many AI medical imaging systems produce significantly less accurate results for black and brown people – although WOC is 43% more likely to die from breast cancer than its white counterparts.
“African-American women continue to develop breast cancer at a younger age and often in more advanced stages,” Salewai Oseni, breast surgeon at Massachusetts General Hospital, said in a recent press release. “This, coupled with the higher case of triple negative breast cancer in this group, resulted in increased breast cancer mortality.”
Over the past two years, researchers from MIT CSAIL and Abdul Latif Jameel Clinic for Machine Learning in Health worked to develop a new deep learning system, capable of predicting a patient’s cancer risk using only the person’s mammograms, which would work with the same efficiency regardless of race or ethnicity.
Nicknamed “Mirai” (not to be confused with Toyota’s Fuel Cell EV), this algorithm would be able to model “the risk of a patient over several future time points”, while taking into account minor variations, up to the brand of mammography machine used by the clinic, according to a press release released Wednesday by MIT. Its predictions can be further optimized if other clinical risk factors – such as age or family history – are available.
The CSAIL team initially trained Mirai on a dataset of 200,000 exams from Massachusetts General Hospital (MGH) before validating their predictive results on additional sets from the Karolinska Institute in Sweden and Chang Gung Memorial Hospital in Taiwan. . So far, the results are very encouraging with results suggesting that Mirai is “significantly more accurate”, according to the release, at predicting cancer risks in patients in all three data groups and able to correctly identify close to twice as many potential cancer cases among high risk groups as the diagnosis currently in use Tyrer-Cuzick model during the study.
To ensure that Mirai’s recommendations were consistent, the CSAIL team skewed the algorithm by running it through an adversarial network to differentiate which aspects of the mammogram are important and those caused by random environmental variances. minor (such as make / model of mammography machine).
“Improved breast cancer risk models allow targeted screening strategies that allow earlier detection and less screening damage than existing guidelines,” Adam Yala, lead author of the upcoming CSAIL Scientific translational medicine study, said in a statement. “Our goal is to incorporate these advances into the standard of care.”
This could advance the state of oncology science. Modern mammograms still suffer from reliability issues, even now 60 years after the technology’s widespread adoption. Experts still disagree on how often women should be screened, with some favoring more aggressive strategies for detecting cancerous growths as early as possible, while others advocating longer gaps between them. routine testing to minimize false positive rates (and declining medical costs for patients). Mirai will be used to help doctors determine which patients would benefit most (and most equally) from additional imaging and MRI based on both the mammogram image and other factors such as the age, genetics, family medical history, and breast tissue density.
“We know MRI can detect cancers earlier than mammography, and earlier detection improves patient outcomes,” Yala explained. “But for patients at low risk for cancer, the risk of false positives may outweigh the benefits. With improved risk models, we can design more nuanced risk screening guidelines that offer more sensitive screening, like MRI, to patients who will develop cancer to achieve better outcomes while reducing unnecessary screening. and over-processing for the rest. “
Mirai also takes into account risk factors that may not show up in mammography imaging, such as patient age, hormone levels, and menopausal status. These factors are ingrained during the training phase, which allows the model to predict them based on the given mammographic image, even if the clinician has not manually provided this information.
In the future, Mirai may find use in other medical applications for the benefit of the community. While the system may not be able to interpret a patient’s history of existing imaging results and incorporate them into their assessment, it may rely on the additional X-rays / MRIs provided to it. in the future. The team also plans to integrate tomosynthesis techniques to further increase Mirai’s statistical ability. The CSAIL team also partnered with researchers at Emory University to further validate the model.