Multilevel Modelling for Surgical Tool Calibration Using LINEX Loss Function

Quantifying the tool–tissue interaction forces in surgery can be utilized in the training of inexperienced surgeons, assist them better use surgical tools and avoid applying excessive pressures. The voltages read from strain gauges are used to approximate the unknown values of implemented forces. To this objective, the force-voltage connection must be quantified in order to evaluate the interaction forces during surgery. The progress of appropriate statistical learning approaches to describe the link between the genuine force applied on the tissue and numerous outputs obtained from sensors installed on surgical equipment is a key problem. In this study, different probabilistic approaches are used to evaluate the realized force on tissue using voltages read from strain gauges, including bootstrapping, Bayesian regression, weighted least squares regression, and multi-level modelling. Estimates from the proposed models are more precise than the maximum likelihood and restricted maximum likelihood techniques. The suggested methodologies are proficient of assessing tool-tissue interface forces with an adequate level of accuracy.