Evaluating the BioTac's Ability to Detect and Characterize Lumps in Simulated Tissue

Surgeons can detect and characterize tumors in open surgery by palpating tissue with their fingertips, but palpation is not currently possible in minimally invasive surgery (MIS). Motivated by the goal of creating an automatic palpation tool for MIS, we evaluated the SynTouch BioTac sensor's ability to detect and characterize lumps in simulated tissue. Models were constructed from silicone rubber with rigid spheres of three sizes embedded at three depths, plus models without embedded lumps. Electrode impedance and DC pressure were recorded as each model was indented into the BioTac at sixteen indentation depths up to 4.0 mm. Support vector machine classifiers were trained on subsets of the data and tested on trials from withheld models for three tasks: lump detection, lump size characterization, and lump depth characterization. The lump detection and lump size classifiers achieved relatively high accuracies, especially at the deepest indentation depths, but the lump depth classifier performed no better than chance.