What does the chest have to do with the brain? Aside from the role played by the brain in regulating life-sustaining involuntary breathing and heart rate, it doesn’t look like there’s much of a connection.
However, French physician Philippe Grenier, a “respected expert in chest imaging and respiratory disease,” is enthusiastic about integrating Artificial Intelligence (AI) into his own field. As past President of the European Congress of Radiology, he has kept a close eye on the ways AI can benefit chest imaging in particular, and radiologic imaging in general. The “brain” he is connecting to the chest is manmade, yet modeled on how the human brain processes information, learns it, and anticipates probabilities.
Saving interpretation time
Most of us don’t think about how the demand for medical imaging has mushroomed. Imaging technologies to noninvasively probe for internal disease have all but replaced exploratory surgery. Such tools now include not only x-rays and fluoroscopy, but also scans performed by ultrasound, CT, PET/CT, and MRI. The devices themselves are increasingly developed in ways that make them more accessible and less expensive, e.g., portable CT scanners. One economic forecast suggests that in-hospital use for early diagnosis will grow at more than 5% through 2026.
This places a heavy workload on the radiologists who read and interpret the exams. Dr. Grenier states that AI is already shortening the time needed for such interpretation, and cites the example from his own specialty of CT chest scans; a particular AI software program can “….not only detect, segment, and measure suspected pulmonary nodules, but can also in the same time (1) detect and quantify emphysema, a marker of increased probability of nodule malignancy, (2) automatically measure the thoracic diameters of aorta at different levels to detect any early aneurysmal dilatation, and (3) segment and quantify calcifications in coronary arteries, a reliable marker of cardiovascular event prediction.”[i]
Image triage based on AI algorithms
We’ve all seen war movies in which mass casualties occur, and medical decisions must be made on the battlefield, even under fire. Historic documentation points to a system of assigning priorities for medical intervention or removal by ambulance from the line of fire as early as the Napoleonic wars. During the Civil War, the astonishing carnage that occurred at the Battle of Gettysburg left medics facing “the paralyzing task of sorting the dead and dying from those whose lives might be saved.”[ii] Often, dealing with cancer is described as a battle, though thankfully survival rates are on the rise. However, Dr. Grenier says that the complex algorithms when AI recognizes abnormalities based on image features in plain chest radiographs it is able to triage, that is, sort the threat to the patient, and thus highlight for radiologists a clear and present danger with unbelievable speed.
Grenier further points out that AI resources have been developed that assist treatment planning based on the probable course of a patient’s disease progression. The use of AI to “predict the outcome, prognosis, probability of malignancy of a detected lesion, or to predict treatment response to this lesion…” means patients’ access to the highest quality clinical decision-making for the likelihood of a successful outcome.
Resistance to implementing AI
Broad acceptance of AI implementation faces a couple of hurdles. The first, of course, is understandable skepticism. How accurate are the results, and will AI put radiologic readers out of business? Grenier acknowledges that ongoing research, development, and studies of AI application results compared with current diagnostic gold standards must continue. Rather than replace human radiologists, AI is seen as a valuable silent partner. However, until AI is broadly demonstrated to be reliable and efficient, many large medical centers may be on the slow side of adoption.
The second challenge is cost. The price tag has so far been manageable because AI is in development by small, independent or academic software engineers. However, as AI proves its worth, it’s inevitable that big business will capitalize on it, and for-profit competitors will drive up pricing. Grenier anticipates that the cumulative benefits of AI, including better productivity and use of human resources will balance out the corporate costs of delivering globally proven software.
In our own Centers, we have already put AI resources into place, and will continue to add them. We recognize the current and future advantages for our patients, and enthusiastically embrace all that AI can do to improve our clinical services to them.
NOTE: This content is solely for purposes of information and does not substitute for diagnostic or medical advice. Talk to your doctor if you are experiencing pelvic pain, or have any other health concerns or questions of a personal medical nature.
[i] “The Impact of AI on Clinical Workflow.” European Society of Radiology Blog, undated. https://ai.myesr.org/healthcare/the-impact-of-ai-on-clinical-workflow/
[ii] Slawson, Robert. “The Development of Triage.” National Museum of Civil War Medicine, Jun. 29, 2017. https://www.civilwarmed.org/surgeons-call/triage/