Table 2 Definitions of parameters provided by the artificial intelligence (AI) model.

From: Artificial intelligence-integrated video analysis of vessel area changes and instrument motion for microsurgical skill assessment

Algorithm

Abbreviation

Definition/equation

Semantic segmentation algorithm for vessel area (VA)

CV-VA

The coefficient of variation (CV) of all measured VA values

VA = Total number of pixels in the vessel area

CV-VA = σ/μ

σ: standard deviation (SD) of VA, μ: mean value of VA

ΔVA

The relative change in VA over time

ΔVA = dVA/dt

t: time

Max-ΔVA

Maximum change in vessel area: The highest absolute value of ΔVA during the procedure

TDE

Tissue deformation error: Any ΔVA values that exceed the defined threshold

Threshold: Mean ± 1.96 × SD of ΔVA, calculated across all trials by all surgeons (± 1.13)

Trajectory tracking algorithm for instrument-tip

PD

Path distance: The total distance traveled by the forceps tip during the procedure

NJI

Normalized jerk index: A measure of motion smoothness, calculated based on acceleration and procedural time

\({\text{NJI}} = \frac{1}{2}*\frac{{t^{5} }}{{D^{2} }}*\smallint \left( {\frac{da}{{dt}}} \right)^{2}\)

a: acceleration, t: procedural time, D: path distance