Development of a robotic biopsy system compatible with label-free digital pathology and methods for needle-tissue classification

The diagnosis of disease faces fundamental limitations due to the uncertainty surrounding the incremental benefits of tissue sampling and biopsy. In response to this critical challenge, a robotic biopsy system with compatibility with label-free digital pathology and methods for tissue classification was developed in this dissertation. This new system incorporates a steerable needle insertion system and is designed to compatible with digital pathology, granting access to tissues' chemical and molecular composition. The tissue classification method holds promise in enhancing the precision and depth of information obtainable through biopsy procedures. In this research we have designed and developed a 5-degree-of-freedom (DOF) robotic biopsy system, integrated with a 3-DOF insertion module, which has been tested in various conditions. Helical motion functionality with detailed workspace analysis has been conducted to optimize the system reach. Safety control algorithms, including a hybrid force/position robotic platform control and semi-autonomous needle insertion control, have been developed to ensure the precision required for needle insertion procedures. Experiments have been carried out on both synthetic and real porcine tissues to validate the system performance. To address a significant challenge in current biopsy procedures, specifically the difficulty surgeons face in accurately placing the needle tip, a novel classification method rooted in machine learning has been developed. Data has been collected from five different types of porcine tissues, and a transformer-based classification model has been trained after thorough data pre-processing. The results have been compared with other state-of-the-art methods, revealing that this new transformer-based classification method is more precise and effective in tissue type detection. This innovation holds potential as a valuable tool for surgeons, by enhancing surgeon awareness and precision. This advancement contributes to the overall success of the biopsy process by providing a new in-situ process that does not exist today. The combination of the robotic biopsy system and the learning-based classification methodology holds the potential to improve the field of needle biopsy. By providing accurate and precise needle placement, access to tissue composition, and real-time environment recognition, the groundwork is laid for the creation of an accurate computer-aided digital needle biopsy system.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Mechanical Engineering

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2023

Author dc:creator

• Wang, Fanxin

Contributors dc:contributor

• Kesavadas, Thenkurussi
• Ferreira, Placid Mathew
• Hovakimyan, Naira
• Bhargava, Rohit

• Robotics
• Medical Diagnosis
• Label-free Digital Pathology
• Ai Classification

Statement dc:rights

• Copyright 2023 Fanxin Wang

Language dc:language

en, eng

Handle dc:identifier

https://hdl.handle.net/2142/122002

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F122002