Year: 2025
Type: article
Abstract: This study aims to enhance the suturing skills of medical practitioners by leveraging artificial intelligence (AI) techniques. Initially, a dataset of sutures was obtained from a hospital setting and underwent preprocessing to align with the model requirements. Subsequently, data augmentation was applied to enhance the dataset for improved performance. Using transfer learning, a classification algorithm was trained on the augmented dataset with %96.59 training and %79.24 validation accuracy. To ensure interpretability, SHAP (SHapley Additive exPlanations) analysis was employed to explain the decisions made by the classification algorithm, revealing the influential pixels in suture success. In the final stage, users were introduced to the project via a mobile application developed with Flutter and Dart. This app allows users to capture images of their sutures, which are then uploaded for analysis. The SHAP analysis results are presented visually to users, indicating which parts of the suture are deemed successful and which are not via heatmaps. By providing this feedback loop, the application aims to assist medical professionals in assessing and improving their suturing skills. This study presents a valuable tool for evaluating and enhancing medical suturing abilities, with potential implications for medical education and practice. In the future this preliminary study will be test with application users which will provide continuous feedback for model refinement.
Source: ADBA computer science.
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4412660145