What Do Trainees Think of AI?

How do Radiology Trainees View the Rise of AI?

Over the past year, dozens of studies have demonstrated the value of artificial intelligence (AI) across the field of radiology. Yet, in a recent survey published the European Journal of Radiology, experienced radiologists, surgeons, and medical students may view those advances in differing light.

Researchers from Switzerland’s University of Bern recruited 170 participants-a mix of 59 radiologists, 56 surgeons, and 55 medical students- to complete an online survey regarding the future of radiology, including how AI may affect future career choices and work environments.

Perhaps unsurprisingly, most participants, regardless of position, agreed that AI had value as a support system for radiologists and would have the power to make their workflows more accurate and efficient. Yet, there were wide concerns about its adoption. For example, survey responses suggested participants saw AI as a potential threat to careers in the field. When questioned about perceived “turf losses” to other specialties and disciplines due to advances in AI, radiologists agreed it was a concern. And more than one-quarter of the students who stated they were looking outside radiology for their careers cited AI as being one of the reasons.

While the researchers admit it is difficult to generalize the findings due to the small, specific sample, they concluded that some medical trainees may choose careers outside of radiology. As such, they recommend that better education regarding AI and how it might affect the field would be in order.

7T MRI Offers A Clearer Look at MS

Tracking the progression of multiple sclerosis, a debilitating autoimmune condition with symptoms including muscle weakness, balance issues, and cognitive problems, is a challenge for clinicians. The disease has traditionally been characterized by small demyelinating lesions in the brain’s white matter. Yet, recent work suggests small lesions in the gray matter precede the demyelination-and the ability to follow the development of those gray matter lesions over time can provide important prognostic information about related neurological disability.

Now, new research from Brigham & Women’s Hospital suggests 7T MRI offers an easier way for clinicians to monitor disease progression for the relapsing-remitting form the disease by focusing on leptomeningeal enhancement (LME). The study was published in the Multiple Sclerosis Journal.

Previous work suggested that 7T MRI, with nearly twice the magnetic field strength of conventional 3T scanners, offers radiologists the ability to better visualize MS lesions and track their evolution over time. Because MS is an autoimmune disorder, some have suggested that inflammation of the meninges may play a role in the development of the disease’s hallmark gray matter lesions. It was then proposed that LME, a radiographic finding which shows an enhanced image of the meninges and spinal cord, might provide a diagnostic biomarker of inflammation that can help determine when a case of relapsing-remitting MS (RRMS) progresses to secondary progressive MS (SPMS).

To test the idea, the researchers used 7T MRI to scan the brains of 30 individuals diagnosed with RRMS, as well as 15 healthy matched controls. They observed:

• 20 of the 30 RRMS patients showed signs of LME, making it significantly more common than the 1 in 15 LME seen in controls;

• RMMS patients with LME typically had four to five times the number of lesions in cortex and in the thalamus than those without;

• LME was linked to higher incidence of gray matter lesions.

Based on those findings, the study authors concluded that cerebral LME is common in RMMS when imaged using 7T MRI-and is linked to increased gray matter injury. This suggests that cerebrospinal fluid-related inflammation may lead to MS’ gray matter lesions-and could provide a marker to help clinicians better track MS disease progression in the future.

PET Imaging Helps Distinguish TBI-Related Ischemia

New research published in JAMA Neurology suggests oxygen positron emission tomography (PET) can detect ischemia within the first 24 hours after a traumatic brain injury (TBI).

After a TBI, the brain undergoes a cytotoxic cascade of events. There is the trauma itself to manage, as well as several secondary insults including swelling, ischemia, and intracranial hypertension. Unfortunately, ischemia, a reduction in blood supply to the brain’s tissues, is difficult to visualize using conventional imaging methods-but its presence can significantly change the prognosis after injury.

Researchers from the University of Cambridge tested the utility of a special type of PET, ¹⁵oxygen PET, to see if it could better characterize cerebral physiology after TBI. They looked at the resulting images from 68 patients with TBI who required intracranial pressure monitoring as well as 27 control participants. They discovered:

• Patients with TBI showed lesser global cerebral brain flow (CBF) and cerebral oxygen metabolism (CMRO₂) measures than controls, indicating classical ischemia;

• Ischemic brain volume was elevated in TBI patients even in the absence of intracranial hypertension and was at its highest within the first 24 hours after injury;

• CBF/CMRO₂ ratios were higher in TBI patients than in controls, suggesting abnormal flow-metabolism and vascular reactivity;

Taken together, the study authors argue that ¹⁵oxygen PET may be a tool to help clinicians better detect ischemia in patients immediately after TBI, given that methods like jugular or brain tissue oximetry offer inconsistent results.

Can AI Help with Radiology Triage?

Recently, there have been a deluge of studies suggesting artificial intelligence (AI) algorithms can help radiologists better detect cancer or other medical issues in a variety of imaging study types. A new analysis published in Academic Radiology suggests it may offer as much value in helping to triage patients, especially in areas where radiologists may be lacking.

The author of the analysis argued that AI tools could help prioritize which images should be placed at the top of a worklist for review-indicating which reads may require more time or a consult. Furthermore, in areas where there may be a shortage of radiologists, she said such algorithms might help determine which cases should be referred to a different hospital or health system for appropriate care. While there would be challenges in designing algorithms that could minimize false-positives or negatives, the author concluded that it would be valuable for AI developers to design and test such algorithms and then pilot such platforms in randomized clinical trials to prove their efficacy.

New Imaging Techniques Offers View into Brain’s “Glymphatic” System

A novel imaging technique allowed scientists to visualize how the brain rides itself of waste during slow-wave sleep. The research was published in Science.

For centuries, clinicians and scientists wondered why the brain was not connected to the lymph nodes, the body’s waste disposal system. Over the past few years, studies in animal models suggested that, during sleep, specialized vessels within the brain used cerebrospinal fluid (CSF) to remove waste and toxins. The discovering scientists named it the “glymphatic” system, as it used glial cells and the brain’s vasculature to move the CSF through the brain. However, despite this exciting finding, conventional imaging techniques did not allow scientists to visualize the same process in humans.

Given that the build-up of excess proteins is associated with neurodegenerative conditions like Alzheimer’s disease, being able to visualize how a healthy brain manages waste is of great interest. And now, researchers from Boston University, Harvard University, Massachusetts General Hospital, and Beth Israel Deaconess Medical Center, have shown that they can do just that.

The researchers recruited 13 individuals to fall asleep inside an MRI scanner. Using accelerated neuroimaging techniques coupled with the brain waves collected from from a conventional electroencephalogram (EEG) cap, the researchers were able to capture a pattern of neural slow waves and hemodynamic oscillations, which, together, allow CSF to flow throughout the brain and clear the brain of debris.

The researchers concluded that CSF dynamics are associated with neural and hemodynamic rhythms and this visualizing method offers a new window for studying how such processes go awry in disease states.