Abstract:
Swallowing is a complex physiological process transporting food or saliva from the mouth to the stomach, aided by peristaltic waves in the esophagus, generated by Central Pattern Generators (CPGs). Esophageal cancer can cause esophageal stricture, which distresses lumen patency, leading to dysphagia. Esophageal strictures can be addressed by implanting an endoprosthetic stent in the esophagus, which can hold open the esophagus and provide relief to patients. However, stent migration due to its interaction with the continuous peristaltic waves, is a significant shortcoming of concern to the patients. Stent Radial Force (RF), force applied by the stent on the esophageal lumen, is a crucial design parameter of a stent for maintaining the in situ lumen patency, which is, unfortunately, still an unknown parameter due to the poorly understood association of the RF and clinical outcomes. In robotics, biological and natural phenomena continue to be a vital source of inspiration for researchers in developing proficient robots. With the demand for studying various biological processes, designing robots capable of soft and continuous interaction with the environment has led to soft robotics. In this research, a bio-inspired Robotic Soft Esophagus (RoSE) with embedded sensing capability has been developed as an alternative platform for conducting stent testing under various bolus swallow conditions. Like its biological counterpart, RoSE can generate peristalsis waves of different speeds and wavelengths to push the bolus forward. The research is divided into two parts, where the first part focuses on proving the capability of RoSE in providing an alternative platform for stent testing in the presence of food bolus. The second part centers on embedded sensing and Machine Learning (ML) based modeling and control of RoSE to enhance the stent testing capabilities. The compliant nature of RoSE makes it a suitable platform for deploying stents of various shapes and designs. RoSE has a stack of regular and repeating layers of pneumatic hollow chambers, arranged axis-symmetrically throughout its length, controlled by air pressure to achieve various actuation patterns. This unique feature of RoSE makes it a perfect choice for conducting stent RF testing under food bolus swallow conditions. Food boluses of varying consistencies were prepared by using an artificial starch thickener. Experimental validation for two candidate stents, stent A and B (mean radial stiffness of 1.55 0.24, 3.13 0.53 Nmm⁻¹), respectively), were performed, and their v respective change in RF with conduit diameter and the impact of RF on their migration was recorded with and without food bolus. In the initial state of RoSE, stent A has reported a low RF of 0.33 0.01 N compared to stent B (18.22 0.60 N), which results in a higher stent A migration relative to stent B. The results proved the capability of RoSE in performing stent RF and migration testing under food bolus swallow. For studying the impact of stenting on bolus swallow efficacy in terms of Intra- Bolus Presssure Signature (IBPS), endoscopic manometry tests were performed on the RoSE with different peristalsis trajectories and artificial food boluses. Additionally, experiments were also conducted to analyze the influence of stent dysfunctionality on swallow efficacy by measuring IBPS and its gradient. The presence of stiffer stent B has significantly reduced the IBPS from 4.15 1.00 to 1.3 0.43 kPa. The experimental results have shown the successful usability of RoSE in performing food bolus swallow efficacy tests in the presence of stents. In the absence of any embedded sensor in RoSE and limited conduit visibility, Quarter- Robotic Soft Esophagus (QRoSE) was developed, and motion capture experiments by placing retroreflective markers were performed on it. The x, y, and z-axis movement of the markers were recorded in datasets. A data-driven model, explicitly defining the underlying Differential Equations (DEs) of RoSE dynamics, was identified by applying several ML techniques. The DEs were finally validated with the results of articulography and capacitive sensor array experiments on RoSE. The applied methodology can be extended to other soft robotics systems, and unlike other ML models, the identified model is not a blackbox model. To enhance the capability of RoSE further, Robotic Soft Esophagus version 2.0 (RoSEv2.0) was developed with an embedded Time of Flight (TOF) sensing feature to measure its conduit deformation. With the integration of TOF sensors, a closed-loop control system in the form of Model Predictive Control (MPC) was established in RoSEv2.0 to govern the peristalsis profile autonomously. Since the MPC design methodology can be extended to any other soft-robotic application; thus, MPC was also implemented to control RoSEv2.0 air chamber pressure by following the same procedure. Peristalsis waves of speed 20 mm.s⁻¹, wavelength 75 mm, and amplitudes 5, 7.5, and 10 mm were successfully generated by the MPC. Additionally, the newly designed RoSEv2.0 with MPC was employed to perform stent migration testing with food boluses of varying consistencies to prove the enhancement in the capability of RoSEv2.0.