NEUROMORPHIC TACTILE SENSING FOR SOFT TISSUE PALPATION USING A ROBOTIC FINGER

Abstract

Palpation is an examination procedure to identify tissue abnormality. Robot-assisted tactile sensing provides robust capabilities for differentiating hard inclusions. In this work, an integrated approach for robotic palpation with a bio-inspired tactile sensing approach is presented to discriminate hard inclusions inside soft tissue. The overall goal is to demonstrate technology feasibility and model application of palpating tissue, such as a nodule embedded in soft tissue. The neuromorphic tactile sensing framework and computational analysis with finite element modelling are developed, and palpation testing with phantom and ex-vivo tissue are conducted. Supervised learning methods were implemented to classify the hard inclusions from spike-based features extracted from the tactile responses. An ex-vivo tissue palpation test successfully differentiated the hard inclusions with a classification accuracy of 77.50%. The ability to discriminate hard inclusions inside soft tissue using neuromorphic tactile encoding could help advance the soft tissue palpation for medical diagnosis using robotic fingers.