Electronic Thesis and Dissertation Repository

Thesis Format



Master of Engineering Science


Electrical and Computer Engineering


Patel, Rajni V.


Needle-based interventions such as brachytherapy are among the most common minimally invasive procedures performed. Despite the numerous advantages of such procedures, Surgeons are met with an array of challenges, most significantly, determining a strategy in real time to compensate for needle deflection as the needle passes through various layers of tissue, all having different mechanical properties. This thesis focuses on exploring new state estimation and control strategies to enhance the quality of needle-based interventions. These strategies include the use of machine learning for path planning and state estimation, while congruently exploring how the shape of a deflected needle can be used to explore reachable needle trajectories. Results and limitations are presented for the proposed strategies. A particular focus is made so that the strategies find a needle manipulation strategy that requires as few manipulations as possible.

Summary for Lay Audience

The work presented in this thesis is focused on exploring ways in which the quality of care for patients undergoing a needle-based intervention can be improved. Common examples of these procedures include brachytherapy and muscle biopsies. Despite all the training a surgeon must do, needle-based interventions are still difficult for surgeons to perform. Human tissue is highly elastic, making it incredibly difficult to predict how a needle will pass through the tissue. Needles will deflect while travelling through tissue, sometimes resulting in the needle missing its target. In such an event, the surgeon may need to retract the needle and perform the insertion again until he/she is able to reach the target location. Excessive needle insertions may cause patient discomfort and tissue damage leading to longer recovery times. This thesis explores iii methods of predicting how a needle will travel through tissue. The work presented is intended to be carried out using robotics to perform needle manipulations. Much of the work done for needle-based interventions focuses on finding analytic models to describe the needle-tissue interaction during a procedure. Parameters in these models include friction, cutting force, clamping force, expected curvature and many more. An issue arises when using analytical models because there are always going to be unmodelled dynamics that are not captured by the proposed model. This thesis focuses on adaptive strategies that use the shape of the needle and modern machine learning techniques to predict the trajectory of a needle in response to manipulations of the needle performed using a robotic arm.

Creative Commons License

Creative Commons Attribution 4.0 License
This work is licensed under a Creative Commons Attribution 4.0 License.

Available for download on Thursday, August 01, 2024