With every year, AI is finding its way more and more into the radiology field. Not only does the technology optimize processes and efficiency, but it will help with cases where radiologists haven’t found a satisfactory solution yet. In a recent case of a young patient, radiologists were able to interpret blurry MRI images with the help of AI, avoiding unnecessary pain and cost for the patient.
Just one false move
One afternoon, a woman in her twenties got up from the couch to get something to drink. All she felt was one short and silent click in her lower back, and from one second to the next she couldn’t move anymore. The pain shot through her spine, making it impossible to sit or stand without any form of support. Her family helped her into the car, and they rushed to the doctor, who administered an injection against inflammation and acute pain. The doctor prescribed medication and scheduled a check-up a few days later. The situation didn’t improve, and with her pre-existing condition of congenital scoliosis in mind, the doctor ordered an MRI scan of the lumbar spine for the following week. At that time, neither the patient nor the doctor expected a serious problem. Since she could still walk with a straight back, albeit slowly, the appointment was scheduled without sedation or anesthesia. But with each passing day, she had more difficulty walking upright. The pinched sciatic nerve shot waves of pain down one of her legs and forced her to hunch her back more and more.
The day the MRI scan was set, she took the maximum dose of pain killers. That way, she was sure, she would be able to lie straight and still in the machine. The radiology technician briefed her that she needed to lie still for 15 minutes. Then she could take a break for a couple of minutes before enduring another 15-minute scan. The patient was certain that she could handle this. 15 minutes sounded like nothing. How hard can it be? Turns out – very.
After only a couple of minutes in the MRI machine, her leg started to twitch involuntarily. The stabbing pain was much worse when lying down with her legs extended than with the legs up. Shortly before the first sequence was done, the pain was too much, and she had to shift the position of her leg. Hopefully no one noticed. Through the speakers, the radiology technician said the scans turned out well and they could proceed with the second scan. During the short break, she unsuccessfully tried to find a less painful position.
The last 15 minutes were hell for both the patient and the radiology team. She couldn’t help the twitching leg, and from time to time she moved her back a few millimeters. After the last sequence, the technician spoke out what she already expected: she moved too much. The scans might be too blurry to diagnose correctly, and she might need a new scan.
What would that entail? Another appointment needed to be scheduled, this time with sedation or anesthesia that worked with her medication. The patient and the healthcare team would lose valuable time. The symptoms could get worse and might affect a successful treatment.
AI to the rescue
Fortunately, the treating radiologist has already been working with radiology AI solutions for a while. He used the applications on the case to enhance MRI image quality and automatically detect abnormalities in the lumbar spine. Within a couple of minutes, the AI software improved the quality of the blurry images. With the enhanced resolution, the algorithm immediately detected the herniated disc between L5 and S1, calculated the volume of the injured disc and spinal canal, and indicated the severity of the injury. After the radiologist confirmed the diagnosis presented by the AI solutions, he sent the report back to the doctor who scheduled a surgery for the next day. It turned out that the sciatic nerve was already severely compressed and could have suffered long-term damage if the surgery was delayed. With the MRI images in hand, the surgery team was well-informed and prepared for the operation.
Better outcomes for everyone involved
Patients may misjudge how long they can lie completely still, especially when it’s their first MRI scan. They simply want to get through with it and may not realize the consequences if the images turn out blurry. Additionally, the radiology staff often can’t react flexibly as they may have other patients sitting in the waiting room.
AI assisted in several aspects of this case:
- The patient was quickly diagnosed and treated, preventing unnecessary pain.
- Another MRI scan was avoided, which saved time and money for the radiology center and insurance company, respectively.
- Since the surgery was scheduled promptly, the post-treatment was speeded up and the overall healthcare costs could be reduced.
- Lastly, the AI-automated report from the enhanced MRI images offered quantitative information for comparison with the follow-up scan a year later.
The case clearly illustrates how AI won’t replace radiology teams, but rather support them to do their job in the best possible way.