Architecture of neural network model for 3D image processing for digital automated design of receptor sleeve for shin prosthesis: prospective study

INTRODUCTION. The demand for lower limb prostheses is growing every year, and improving the technological processes involved in their manufacture is an important technical challenge that could solve the medical and social problem of providing prosthetic and orthopedic devices to those in need.

OBJECTIVE. Develop the architecture of a neural network model for processing 3D scans of limb stumps for the digital automated design of individual prosthetic socket shells for the lower leg using artificial intelligence (AI) technologies.

MATERIALS AND METHODS. 3D scans of lower leg stumps and the inner surfaces of prosthetic sockets. Software for analyzing and processing data using neural networks. Python programming language (Netherlands, Opensource) and MeshLab modeling environment (Italy, Opensource). Identification of key features and creation of a dataset for the design of individual lower leg prosthesis modules. Development of a neural network model architecture for the digital automated design of individual lower leg prosthesis sockets based on a scan of the patient’s stump and prediction of load and unload areas when using the prosthesis.

RESULTS. The architecture of a neural network model for processing 3D scans of stumps has been developed for the digital automated design of prosthetic lower leg socket receptacles. Data sets have been formed, including 3D scans of stumps and characteristics of the inner surfaces of sleeves. The prototype system allows predicting areas of load and unload inside the sleeve, taking into account the individual characteristics of the patient, based on a three-dimensional scan of their amputated limb.

DISCUSSION. The use of artificial intelligence (AI) technologies for processing 3D scans of the stump offers significant advantages: reduction in the manufacturing time of individual modules (receiving sleeves) for prosthetic limbs without compromising quality; automation of manual labor, reduction of human error, and increased accuracy in the design of prosthetic sockets; increased efficiency in providing prosthetic limbs to people with disabilities and ensuring comfort when using the product. However, the implementation of such systems requires further research, including model validation on a large volume of data and integration with existing technological processes.

CONCLUSION. The developed architecture of the neural network model for processing 3D scans of stumps significantly reduces the duration of the design process for individual prosthetic socket designs without compromising quality. The use of artificial neural networks accelerates manufacturing and reduces human error. The implementation of the developed model contributes to improving the quality of life of people with amputations, as well as the effectiveness of rehabilitation and prosthetic-orthopedic measures and services.

1. Ponomarenko G. N. Physical and rehabilitation medicine: fundamental principles and clinical practice. Physiotherapy, balneology and rehabilitation, 2016, Vol. 15, No. 6, pp. 284–289 (In Russ.)

2. Ponomarenko G. N. Restorative medicine: fundamental principles and prospects for development Physical and rehabilitation medicine, 2022, Vol. 4, No. 1, pp. 8–20 (In Russ.)

3. Ponomarenko G. N., et al. Medical rehabilitation: the state of the domestic flow of scientific publications. The health care manager, 2020, No. 7, pp. 53–59 (In Russ.)

4. Sufelfa A. R., Kaplun D. I., Chernikova M. V. A study of a set of initial data for the development of a decision support system for the selection of an individual replacement sleeve for a prosthetic leg. All-Russian Scientific Conference on management problems in Technical Systems. V. I. Ulyanov (Lenin) St. Petersburg State Pedagogical University. St. Petersburg, September 21–23, 2021. 2021, Vol. 1, pp. 89–91 (In Russ.)

5. Varrecchia T., et al. Common and specific gait patterns in people with varying anatomical levels of lower limb amputation and different prosthetic components. Human movement science, 2019, Т. 66, С. 9–21.

6. Alrasheedi N. H., Ben Makhlouf A., Louhichi B., Tlija M., Hajlaoui K. Customized Orthosis Design Based on Surface Reconstruction from 3D-Scanned Points. Prosthesis, 2024, 6(1), 93–106.