AI in Radiology II

Initial benefits of AI in this realm include providing earlier detection of a potentially life-threatening event  and ensuring higher accuracy in reading these studies. If a patient presents with a stroke or a collapsed lung, an algorithm that can immediately look for abnormalities upon completion of the scan and alert the radiologist if it finds something, that can lead to faster life-saving interventions. If an algorithm augments a radiologist in finding hard to spot tumor or spinal stenosis, it can lead to a diagnosis that would have been otherwise missed. Deploying deep learning solutions on existing infrastructure delivers the potential to deliver more efficient and effective care, enhance decision-making, and drive greater value for patients and providers

The algorithms can give radiologists a boost in reading scans. Radiomics is the study of  scans for hidden findings and signatures. Algorithms can help better find these findings and signatures and  lead to deeper quantification of what’s on the scans. Deep learning models can be used to identify incidental findings, as well as help radiologists manage their workload, enhance quality of scans, and reduce ‘retakes’, which can cause unnecessary exposure to radiation. Deep learning is also showing promising results in image reconstruction from the imaging modalities.

Artificial intelligence is making fast progress in the field of radiology. Clinical adoption of AI by radiologists has gone from none to 30% from 2015 to 2020, according to a study by the American College of Radiology. There are, of course, certain issues that would keep an AI algorithm to be the only entity that reads a scan. First, these algorithms are trained to perform a narrow task such as finding tumors or hemorrhage or fractures. That means they will not identify other things that may be present on a scan. This not only presents a major clinical issue but also legal. It is the responsibility of the radiologist to report on all visible abnormalities on a radiologic study. That means an algorithm can help a radiologist in performing a narrow task like finding a fracture but will miss everything else. That is why a radiologist needs to read the study and be ultimately responsible for the final report.