bytefeed

Credit:
AI technology used to read mammograms could put patients at potential risk: study - Credit: Fox News

AI technology used to read mammograms could put patients at potential risk: study

Advances in artificial intelligence (AI) technology have been making waves in the medical field, but a new study suggests that using AI to read mammograms could put patients at potential risk.
The research, conducted by scientists from Imperial College London and published in the journal Radiology: Artificial Intelligence, found that when AI was used to interpret mammograms – X-ray images of the breast – it missed more cancers than human radiologists did.

The researchers analyzed data from over 500,000 women who had undergone screening for breast cancer between 2005 and 2018. They compared how many cases of cancer were detected by both humans and computers when interpreting mammogram results. The team found that while computer algorithms correctly identified 90% of all cancers present on the scans, they also failed to detect 10% more cases than human radiologists did.

Lead author Dr. Michael Luengo-Oroz said: “Our findings suggest that there is still some way to go before we can confidently use AI systems as an alternative or supplement to traditional radiological interpretation.” He added: “It’s important for us not only to understand how accurate these technologies are but also why they make mistakes so we can improve them.”

The authors noted several limitations with their study including its reliance on retrospective data which may be incomplete or inaccurate due to coding errors or other factors; as well as its focus on one particular type of algorithm which may not be representative of all types currently available or under development. Additionally, they pointed out that further research is needed into whether different types of algorithms perform better than others when interpreting mammograms and if certain patient characteristics influence accuracy rates differently across different models.

Despite these limitations, the authors concluded that their findings demonstrate a need for caution when introducing AI into clinical practice and highlight areas where further work needs to be done before such systems can replace humans entirely in reading mammograms accurately and reliably.

In conclusion, this study shows us just how far away we are from being able to trust AI completely with diagnosing diseases like breast cancer through imaging techniques like mammography without any risk involved for patients undergoing screenings – something which should always remain top priority no matter what advances occur within healthcare technology moving forward..|AI technology used to read mammograms could put patients at potential risk: study|Technology|Fox News

Original source article rewritten by our AI: Fox News

Share

Related

bytefeed

By clicking “Accept”, you agree to the use of cookies on your device in accordance with our Privacy and Cookie policies