Tactile perception: a biomimetic whisker-based method for clinical gastrointestinal diseases screening
Zeyu Wang, Frank P.-W. Lo, Yunran Huang, Junhong Chen, James Calo, Chen Wei, Benny Lo
Abstract
Abstract Early screening for gastrointestinal diseases is of vital importance for reducing mortality through introducing early intervention. In this paper, a biomimetic artificial whisker-based hardware system with artificial intelligence-enabled self-learning capability is proposed for endoluminal diagnosis. The proposed method provides an end-to-end screening strategy based on tactile information to extract the structural and textural details of the tissues in the lumen, enabling objective screening and reducing the inter-endoscopist variability. Benchmark performance analysis of the proposed was conducted to assess the electrical characteristics and core functions. To validate the feasibility of the proposed for endoluminal diagnosis, an ex-vivo study was conducted to detect some common tissue structures and our method shows promising results with the test accuracy up to 94.44% with 0.9167 kappa. This previously unexplored tactile-based method could potentially enhance or complement the current endoluminal diagnosis.