Myoblast transplantation is a key approach in regenerative medicine, requiring effective tracking of live cells for successful integration and muscle repair. Manual tracking is impractical, and machine learning offers a viable alternative despite challenges such as unstained, irregularly shaped cells. This study develops a generalized model for shape-invariant segmentation, minimizing dependence on rigorously trained models. Using unsupervised and semi-supervised learning methods, the model integrates clustering, statistical techniques, and neural networks. An autonomous tool with a user-friendly GUI analyzes cell behavior and generates reports. Results show that unsupervised learning with calibration enables reliable segmentation and prediction, advancing regenerative medicine.

Дас С. (науч. рук. Зун П.С.) Ai to optimize myoblast transplantation for muscle repair // Сборник тезисов докладов конгресса молодых ученых. Электронное издание. – СПб: Университет ИТМО, [2025]. URL: https://kmu.itmo.ru/digests/article/15466