PNIPU: Neural network for detecting posture disorders

The main articles are:

• Scoliosis
• Artificial Intelligence in Medicine
• Artificial intelligence in Russian medicine
• Video analytics (terms, applications, technologies)

2024: Russian scientists have created a neural network to help diagnose scoliosis

Perm Polytechnic scientists have created and trained a neural network to find key back points in the diagnosis of scoliosis. The application of computer vision makes the definition of disease more accurate and accessible to the patient. This was announced on January 29, 2024 by representatives of the Perm Polytechnic.

According to the company, an article with the results of the study was published in the journal Vestnik PNIPU. Applied Mathematics and Management Issues, "No. 4. The work was carried out with the financial support of the Perm World-Class Scientific and Educational Center "Rational Subsoil Use."

source = PNIPU

Scoliosis is especially characteristic of children, often it forms during the period of active growth, starting from the age of 5. A healthy spine is a curve with physiological bends in the cervical, thoracic and lumbar regions. In a child, it is quite plastic, and improper load distribution and other factors provoke the deviation of individual vertebrae from the main curve, thereby forming scoliosis.

source = PNIPU

Timely detection of the disease will avoid lameness, flat-footedness circulatory disorders, breathing, infringement nerves and other complications in the child in the future. Diagnosing scoliosis at the initial stage is difficult. As of January 2024, it is determined by physical observation with a doctor and by a radiation method (or X-ray), MRI which has a number of restrictions with frequent repetition.

As of January 2024, biometric technologies are popular in medicine. They use the physical and behavioral characteristics of a person and through computer vision contactless recognize the disease. Scientists of the Perm Polytechnic University have developed a project that determines key points on its surface using a created neural network algorithm from a photograph of a person's back.

Polytechnics have been researching scoliosis detection technologies for several years. They previously developed a mathematical algorithm that diagnoses curvature from a three-dimensional model of the spine. The interface applications for phone and its web version are already ready. As of January 2024, PNIPU scientists have introduced the technology. artificial intelligence Together, this allows a comprehensive assessment of posture disorders and deformation of the musculoskeletal system.

To teach and test the neural network, the researchers used 3,000 photos of the backs of adults (18-40 years old) and elementary school students. Key points in all photographs were determined using optical technologies that analyze the image of a person's body surface. So you can remotely and contactlessly determine the shape of the body of a patient with disorders of the musculoskeletal system.

We have developed a neural network algorithm that identifies 16 special points from a photograph of the back. The location of the points relative to each other allows us to conclude that there are various posture disorders. We compared the neural network model with a previously created spatial three-dimensional model based on the photogrammetry method. With its help, by filming the back with a smartphone camera from different angles, you can restore the volumetric model. shared Vladislav Nikitin, Candidate of Physical and Mathematical Sciences, Associate Professor of the Department of Computational Mathematics, Mechanics and Biomechanics, PNIPU

source = PNIPU

The doctor or the person himself will be able to open the installed program (application) and choose the diagnostic option. Express analysis will determine violations using artificial intelligence in just one photo, and an expanded version - by a video file of the back surface taken from different angles. As a result, a person will receive a decoding of the values ​ ​ and recommendations for preventive exercises. explained Ivan Shitoev, Assistant of the Department of Computational Mathematics, Mechanics and Biomechanics, PNIPU

The researchers note that after conducting clinical trials and finalizing the program, the application will be ready to run on computers and phones. It can be used by both a doctor and an ordinary person to determine scoliosis.

The development of PNIPU scientists reaches 85% accuracy. A trained neural network can be used in clinical medicine, whose specialists are interested in the emergence of valid tools for diagnosing spinal deformity.