The Term “Artificial Intelligence (AI)”
Scientists have been working on artificial intelligence since the middle of the last century; however, the term “artificial intelligence” did not come into usage until a bit later. The term “artificial intelligence” was coined in 1955 in in the submitted proposal for a scientist workshop by John McCarthy (Dartmouth College), Marvin Minsky (Harvard University), Nathaniel Rochester (IBM), and Claude Shannon (Bell Telephone Laboratories). The term “artificial intelligence” was officially born a year later in the summer of 1956 when the “2-month, 10 man studies of artificial intelligence” workshop was held at Dartmouth College in New Hampshire.
AI is a HOT Topic to Discuss
In 1972, Edward (Ted) Shortliffe from Standford University developed MYCIN, which was an expert system for identifying bacteria that cause severe infections. It assisted physicians to optimize the antibiotic selection for the patient’s treatments. Since then, AI has found its way into many medical applications such as drug discovery, remote patient monitoring and medical diagnostic imaging. The topic of AI is HOT to discuss. It has also been popping up as a spotlight course at some healthcare conferences and was recently showcased at RSNA and the Arab Health Conference. At the upcoming European Society of Radiology Annual Congress (ECR) in July 2020, which will be held as an online congress due to the current situation with Covid-19, AI will once again play a prominent role. AI stands now at the forefront of innovation in the healthcare industry.
AI development in radiology in general has often raised concerns within radiological societies as to whether AI will at one point replace the need for the trained radiologist. A 2018 article written by the Harvard Business Review indicates that some medical students are hesitant to specialize in radiology because of their fears that their job may not exist anymore in the future. However, AI proves its worth in radiology as described by Bradley J. Erickson, an associate professor at the Mayo Clinic College of Medicine. In a traditional setting, radiologists are under pressure to interpret one image in every 3-4 seconds to meet the workload demand. The use of AI could greatly reduce or remove this strain from radiologists. According to Erickson, “AI can also help to detect lesions that may be subtle, which can be particularly useful when the radiologist is tired or distracted. Finally, we and others are showing that it can find information in images that is not perceived by humans—things like molecular markers in tumors.” As AI proves its value to radiology professionals, it will be most likely transforming the diagnostic imaging world.
AI in Musculoskeletal Imaging
In radiology applications, AI has shown remarkable progress in image diagnostic interpretation and is already being used in various ways. AI in medical imaging has already been widely used in the areas of oncology and cardiovascular medicine. It is less common in musculoskeletal (MSK) applications; however, this does not mean that AI is less useful in MSK radiology than the two other aforementioned areas. For example, research published by the American Society for Bone and Mineral Research (ASMBR) found that MSK radiologists may overlook spine compression fractures if they do not routinely review sagittal midline images on body CT, whereas application of an automated deep learning system identified the fractures with nearly 96% accuracy.
In a blog for the European Society of Musculoskeletal Radiology (ESSR) , Franz Kainberger, a radiologist and professor at the University of Vienna, writes that “MSK imaging is—besides cardiovascular radiology – more prone to AI than other sub disciplines because of the broader spectrum of applications, a longer research tradition and the high complexity of MSK diseases.” He further asserts that the demand for MSK radiology will constantly increase the specific tasks of radiologists. For instance, the increasing number of discrete biomarkers that radiologists will be required to quantify and assess will ultimately result in a costly and time-consuming process.
Using AI as an assistant in such situations could certainly be a great help for the radiologist, as AI is trained to address challenges to high quality image acquisition and also helps to optimize in choosing the protocol. It will generate a standard and automated report for clinical documentation. Furthermore, AI could also automatically prioritize and triage cases during times of high pressure by highlighting the urgent or other relevant findings first; the radiologist will be alerted and therefore not miss any important findings. AI will allow radiologists to gain more efficiency in the workflow radiology, leading to a positive impact on patient care for MSK radiologists.
What would you say to AI for an MSK radiologist: is it a “yay” or a “nay”?
Leave a Reply