Google recently revealed on Monday, Nov. 28, via a blog, that it has licensed its artificial intelligence (AI) research model for breast cancer detection to iCAD, a medical technology business.
In the long run, the search engine tech giant expects that better breast cancer diagnosis and risk assessment will result from licensing this technology for the first time.
Google's communications manager Nicole Linton told The Verge via email that both companies' goal is a "2024 release" for the technology to be used in actual clinical settings. However, commercial implementation is still contingent on ongoing research and testing results.
"We will move deliberately and test things as we go," Linton stated in the email.
Partnership with iCAD
The relationship expands upon the previous work that Google has done to enhance breast cancer detection.
According to an article published in the journal Nature by Google researchers in the year 2020, the company's AI system surpassed a panel of radiologists at spotting breast cancer symptoms. When applied to thousands of mammograms, the model could minimize false negatives by up to 9.4% and false positives by up to 5.7%.
iCAD has ambitions to update its current products using the mammography AI research model developed by Google.
Digital breast tomosynthesis (DBT), sometimes known as "3D mammography," is a kind of sophisticated imaging that may be used to examine breast tissue. The software examines DBT pictures for signs of malignant calcification and soft tissue density.
In addition to this, iCAD intends to include Google's model in its risk assessment tool, which the business claims deliver individual-specific estimates of breast cancer risk.
Experts' Concerns on AI
One potential use of AI is to aid radiologists in their work with their patients. However, experts in the medical field are often wary of AI.
Within the scope of Google's 2020 study, there were a few occasions in which radiologists detected malignancy that the model had not previously picked up on. Apart from that, there is not tried-and-true method for identifying cancer.
As per prior reporting of The Verge, it might be challenging to create a decent baseline when training an algorithm. Such AI techniques could be enhanced by providing more than two possibilities to account for the "gray area of diagnosis.
They currently provide either a "cancer" or "no cancer" result.
Overdiagnosis may occur if AI is used extensively without human physicians being brought in to evaluate all of a patient's health details, particularly when attempting to identify early-stage cancer.
Google's Another Venture
Meanwhile, Google has said that it is collaborating with the National Health Service (NHS) of the UK and Imperial College London to investigate whether or not its AI system can function as a "second independent reader" while double-reading mammograms.
Reports indicate this would enable radiologists to concentrate on high-priority cases while enhancing consistency and quality of screening.
This article is owned by Tech Times
Written by Trisha Kae Andrada
