Engaging in prolonged keyboard and mouse marathons within typing-focused professions is not just a workout for your fingers but can also be a pain in the wrist. Working on computers for long periods pinches and compresses the median nerve (which facilitates the movement of arms, arm muscles, wrists and hands). The result is wrist pain, pins and needle sensations and weakness— symptoms associated with carpal tunnel syndrome.
Diagnosing this nerve disorder often requires an ultrasound. However, its limitation lies in its inability to explore the deeper structures in the wrist region.
To solve this problem and to assist the clinician in accurately diagnosing the disorder, researchers at the Indian Institute of Science (IISc), Bengaluru, along with neurologists at Aster CMI Hospital, Bengaluru, engineered an AI tool that uses the machine learning model like that of ChatGPT. The details of their study were published in the IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control.
“The model helps to understand hand anatomy and provides a precise cross-sectional area of the median nerve that helps to accurately diagnose carpal tunnel syndrome,” says Phanendra Yalavarthy, an author of the study and Professor at the Department of Computational and Data Sciences, IISc.
Wrist woes
Functioning as a vital messenger, the median nerve transmits both motor and sensory signals back and forth between the brain and the upper limb. The median nerve also relays information about touch, pain, and temperature to the brain. Therefore, a damage to the median nerve (like in carpal tunnel syndrome) indicates motor deficits and sensory loss.
Not all answers in the ultrasound
Experts rely on ultrasound to detect carpal tunnel syndrome as it can assess the median nerve size, shape, and potential abnormalities. “Ultrasound stands out [over MRI and X-ray] for its real-time, dynamic, cost-effective, and radiation-free imaging capabilities, making it a preferred choice for diagnosing carpal tunnel syndrome,” says Prof Yalavarthy.
However, the ultrasound cannot fully visualise the boundaries of the nerves. To comprehend potential treatments, offer pain relief and enhance surgical planning, it is crucial to actively track the median nerve to administer local anaesthesia or perform nerve blocks.
AI helping hand
Prof Yalavarthy and his team looked to AI for solutions to these problems. “AI in ultrasound imaging for detecting carpal tunnel syndrome helps to improve diagnostic accuracy, efficiency, standardisation, and early detection,” he explains.
The researchers crafted a machine learning model, employing the vision transformer architecture. Training the model with ultrasound videos from Aster CMI Hospital, they honed its skills to isolate and pinpoint the median nerve within individual frames. The model could also report the cross-section of the nerve. All these features enabled it to diagnose carpal tunnel syndrome with 95 per cent accuracy.
Prof Yalavarthy explains that carpal tunnel syndrome is linked to occupational factors like repetitive hand movements, awkward wrist positions (common for those who use the keyboard), and prolonged use of vibrating tools or equipment. “Understanding how workplace ergonomics and preventive measures can mitigate the risk of carpal tunnel syndrome can help improve occupational health standards and reduce work-related injuries,” he says.
Prof Yalavarthy hopes that this tool will be a valuable adjunct to healthcare professionals and help diagnose carpal tunnel syndrome accurately. “It also allows more patients to be screened for nerve disorders and, hence, reducing the overall clinical investigation time,” he adds.