Quantitative Assessment of Microsurgical Skill in Intrascleral Fixation Surgery Using Wearable Strain Sensors: A...

The purpose of this study was to evaluate a novel method using stretchable strain sensors for analyzing hand movements during simulated intrascleral intraocular lens (IOL) fixation, allowing differentiation between expert and novice surgeons.

Six participants (3 experts and 3 novices) performed intrascleral IOL fixation on a simulated eye model while wearing stretchable strain sensors on all fingers to capture real-time finger movements. Two critical procedural phases were defined: leading and trailing haptic insertions (critical process [CP]1 and CP2, respectively). Time-series data extracted from the sensors were processed using "tsfresh" to compute multidimensional features, and they were classified using a light-gradient boosting machine (LightGBM). Performance was assessed using three-fold cross-validation, and the feature space was further explored using principal component analysis (PCA).

In total, 107 procedures were analyzed. Experts completed CP1 and CP2 more quickly than novices. Feature extraction yielded 899 and 472 features for CP1 and CP2, respectively. The LightGBM classifier achieved high accuracy in distinguishing experts from novices, with average accuracies of approximately 87% to 88% for CP1 and 86% to 87% for CP2. PCA showed that thumb and index finger movements of the dominant hand substantially contributed to the discrimination of skill levels.

Our findings showed the feasibility of using wearable strain sensors to quantify microsurgical finger movements, facilitating objective evaluation of surgical skills. This approach represents a promising step toward more refined and data-driven surgical training and evaluation methods.

This novel sensor-based method could help refine surgical training and evaluation, potentially enhancing patient safety and outcomes through data-driven skill assessment.