Application for diagnosis of scoliosis

Scientists have developed a deep learning model that has been trained to determine the extent and shape of scoliosis from a photograph of the back. It turns out that the practice can assess the shape and rate of progression of scoliosis worse than a surgeon. It can even tell you when to seek help from a doctor.

In addition, using the program on a daily basis instead of spine X-ray for scoliosis follow-up will reduce the radiation load on the adolescent’s body.

However, further studies on a larger patient group are needed for the application to be applied to clinical practice. There are also some difficulties with its use: for example, less accuracy in the assessment of scoliosis in people with a body mass index greater than 30 kg/m2, as well as problems with operation on some devices.

What is known about scoliosis

Scoliosis is a lateral curvature of the spine and in 80% of cases the cause is unknown. Occurs during childhood or adolescence. Less often it can be congenital (due to developmental abnormalities) and neuromuscular (due to neurological disease, spinal cord injuries and other diseases).

In most cases, no treatment is needed. The goal of therapy is to reduce pain and improve spinal function. For example, with a pronounced curvature of 40 degrees, or to stop progression, sometimes operations are performed.