Abstract

Minimally invasive surgery has turned into the most commonly used approach to treat cardiovascular diseases during the surgical procedure; it is hypothesized that the absence of haptic (tactile) feedback and force presented to surgeons is a restricting factor. The use of ablation catheters with the integrated sensor at the tip results in high cost and noise complications. In this thesis, two sensor-less methods are proposed to estimate the force at the intracardiac catheter’s tip. Force estimation at the catheter tip is of great importance because insufficient force in ablation treatment may result in incomplete treatment and excessive force leads to damaging the heart chamber. Besides, adding the sensor to intracardiac catheters adds complexity to their structures. This thesis is categorized into two sensor-less approaches: 1- Learning-Based Force Estimation for Intracardiac Ablation Catheters, 2- A Deep-Learning Force Estimator System for Intracardiac Catheters. The first proposed method estimates catheter-tissue contact force by learning the deflected shape of the catheter tip section image. A regression model is developed based on predictor variables of tip curvature coefficients and knob actuation. The learning-based approach achieved force predictions in close agreement with experimental contact force measurements. The second approach proposes a deep learning method to estimate the contact forces directly from the catheter’s image tip. A convolutional neural network extracts the catheter’s deflection through input images and translates them into the corresponding forces. The ResNet graph was implemented as the architecture of the proposed model to perform a regression. The model can estimate catheter-tissue contact force based on the input images without utilizing any feature extraction or pre-processing. Thus, it can estimate the force value regardless of the tip displacement and deflection shape. The evaluation results show that the proposed method can elicit a robust model from the specified data set and approximate the force with appropriate accuracy.