Evaluation of Transfer Learning Efficacy for Surgical Suture Quality Classification on Limited Datasets
Authors:
- Maksim SolopovMaksim Solopov
- Andrey PopandopuloAPAndrey Popandopulo
- Elizaveta ChechekhinaECElizaveta Chechekhina
- Viktor TurchinVTViktor Turchin
- Fedor PopivnenkoFPFedor Popivnenko
- Aleksandr ErmakAEAleksandr Ermak
- Konstantyn LadykKonstantyn Ladyk
- Anton KonyashinAKAnton Konyashin
- Kirill GolubitskiyKGKirill Golubitskiy
- Aleksei BurtsevABAleksei Burtsev
- Dmitry FilimonovDmitry Filimonov
Open Access
- 8 August 2025
- journal article
- research article
- Published by MDPI AG in Journal of Imaging
- Vol. 11 (8) , 266
Abstract
This study evaluates the effectiveness of transfer learning with pre-trained convolutional neural networks (CNNs) for the automated binary classification of surgical suture quality (high-quality/low-quality) using photographs of three suture types: interrupted open vascular sutures (IOVS), continuous over-and-over open sutures (COOS), and interrupted laparoscopic sutures (ILS). To address the challenge of limited medical data, eight state-of-the-art CNN architectures—EfficientNetB0, ResNet50V2, MobileNetV3Large, VGG16, VGG19, InceptionV3, Xception, and DenseNet121—were trained and validated on small datasets (100–190 images per type) using 5-fold cross-validation. Performance was assessed using the F1-score, AUC-ROC, and a custom weighted stability-aware score (Scoreadj). The results demonstrate that transfer learning achieves robust classification (F1 > 0.90 for IOVS/ILS, 0.79 for COOS) despite data scarcity. ResNet50V2, DenseNet121, and Xception were more stable by Scoreadj, with ResNet50V2 achieving the highest AUC-ROC (0.959 ± 0.008) for IOVS internal view classification. GradCAM visualizations confirmed model focus on clinically relevant features (e.g., stitch uniformity, tissue apposition). These findings validate transfer learning as a powerful approach for developing objective, automated surgical skill assessment tools, reducing reliance on subjective expert evaluations while maintaining accuracy in resource-constrained settings.Keywords
This publication has 35 references indexed in Scilit:
- Comparison study of two anastomosis techniques in right hemicolectomy: a systematic review and pooling up analysisInternational Journal of Colorectal Disease, 2025
- Comparison of different surgical techniques and anastomosis methods in short-term outcomes of right colon cancer: a network meta-analysis of open surgery, laparoscopic, and robot-assisted techniques with extracorporeal and intracorporeal anastomosisUpdates in Surgery, 2025
- Optimal oesophagogastric anastomosis techniques for oesophageal cancer surgery – A systematic review and network meta-analysis of randomised clinical trialsEuropean Journal of Surgical Oncology, 2025
- AIxSuture: vision-based assessment of open suturing skillsInternational Journal of Computer Assisted Radiology and Surgery, 2024
- Surgical approaches to colonic and rectal anastomosis: systematic review and meta-analysisInternational Journal of Colorectal Disease, 2023
- Evaluation of suturing performance in general surgery and ocular microsurgery by combining computer vision-based software and distributed fiber optic strain sensors: a proof-of-conceptInternational Journal of Computer Assisted Radiology and Surgery, 2020
- Video-based surgical skill assessment using 3D convolutional neural networksInternational Journal of Computer Assisted Radiology and Surgery, 2019
- Deep learning with convolutional neural network for objective skill evaluation in robot-assisted surgeryInternational Journal of Computer Assisted Radiology and Surgery, 2018
- Closure methods for laparotomy incisions for preventing incisional hernias and other wound complicationsCochrane Database of Systematic Reviews, 2017
- ImageNet Large Scale Visual Recognition ChallengeInternational Journal of Computer Vision, 2015