The new age of technology has proved itself more efficient in the field of healthcare. In the data presented by Sarah Eskreis-Winkler, M.D. it was shown that a deep learning trained algorithm can reliably recognize breast cancer in MR imaging making prevailing radiology more efficient.
Eskreis-Winkler at The Society of Breast Imaging (SBI) and American College of Radiology (ACR) Breast Imaging Symposium asserted that the trained algorithm not only identify tumors but also could save time without compromising precision.
Eskreis-Winkler, radiology resident from Weill Cornell Medicine/New York-Presbyterian Hospital said, “Deep Learning is a new powerful technology that has the potential to help us with a wide range of imaging tasks. DL has shown exceeding human-level performance in some cases.”
New AI Beats Earlier CAD Software
• Eskreis-Winkler confirmed that the great advantage of using AI is not only identifying the issue but also doing such a task without any false positives. Unlike CAD software, AI algorithm deployment in such cases also works substantially faster. Computer Assisted Detection software was developed for FFDM (full field digital mammography).
• CAD software used to lead to a large sum of false positives which also tends to slow down the speed of interpretation rather than accelerating it.
• On the other hand, the AI algorithm tested at HUP was implemented to tomosynthesis.
• Emily Conant, M.D. who also lead the study said, “The concept behind this AI is that you will have it from the ‘get-go.’ It can help you rapidly navigate to slices that have concerning lesions.”
Tech-Efficiency Varies with Dense and Non-Dense Breasts
Unexpectedly, one of the outcomes from HUP research showed that the algorithm reduced the reading times more for dense than the non-dense breast. In case of non-dense breast, average reading times dropped from 62.5 seconds to 32.8 seconds, whereas in the dense breast, while reading images it drops from 65.8 seconds to 28 seconds.
However, according to experts, the extra time utilized for reading non-dense image is due to greater complexity attached to the type of image.
One of the experts said, “When it comes to reading time, sometimes there is not a lot to look at with a dense breast. If you have a breast that is less dense and has a lot of little nodules and calcifications, it can be a more complex texture and, therefore, require a longer reading time.”
The major conclusion derived from the whole study is that – use of AI in screening Breast Cancer can reduce the time required to read DBT while enhancing the cancer detection process.