Use this URL to cite or link to this record in EThOS:
Title: Sensorisation of a novel biologically inspired flexible needle
Author: Virdyawan, Vani
ISNI:       0000 0004 8504 6031
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
Percutaneous interventions are commonly performed during minimally invasive brain surgery, where a straight rigid instrument is inserted through a small incision to access a deep lesion in the brain. Puncturing a vessel during this procedure can be a life-threatening complication. Embedding a forward-looking sensor in a rigid needle has been proposed to tackle this problem; however, using a rigid needle, the procedure needs to be interrupted if a vessel is detected. Steerable needle technology could be used to avoid obstacles, such as blood vessels, due to its ability to follow curvilinear paths, but research to date was lacking in this respect. This thesis aims to investigate the deployment of forward-looking sensors for vessel detection in a steerable needle. The needle itself is based on a bioinspired programmable bevel-tip needle (PBN), a multi-segment design featuring four hollow working channels. In this thesis, laser Doppler flowmetry (LDF) is initially characterised to ensure that the sensor fulfils the minimum requirements for it to be used in conjunction with the needle. Subsequently, vessel reconstruction algorithms are proposed. To determine the axial and off-axis position of the vessel with respect to the probe, successive measurements of the LDF sensor are used. Ideally, full knowledge of the vessel orientation is required to execute an avoidance strategy. Using two LDF probes and a novel signal processing method described in this thesis, the predicted possible vessel orientations can be reduced to four, a setup which is explored here to demonstrate viable obstacle detection with only partial sensor information. Relative measurements from four LDF sensors are also explored to classify possible vessel orientations in full and without ambiguity, but under the assumption that the vessel is perpendicular to the needle insertion axis. Experimental results on a synthetic grey matter phantom are presented, which confirm these findings. To release the perpendicularity assumption, the thesis concludes with the description of a machine learning technique based on a Long Short-term memory network, which enables a vessel's spatial position, cross-sectional diameter and full pose to be predicted with sub-millimetre accuracy. Simulated and in-vitro examinations of vessel detection with this approach are used to demonstrate effective predictive ability. Collectively, these results demonstrate that the proposed steerable needle sensorisation is viable and could lead to improved safety during robotic assisted needle steering interventions.
Supervisor: Rodriguez y Baena, Ferdinando ; Dini, Daniele Sponsor: European Commission ; Indonesian Ministry of Education Directorate General of Higher Education
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral