University of Twente Student Theses


Confidence driven control for autonomous robotic 3D ultrasound acquisition of the breast

Groot, A.G. de (2019) Confidence driven control for autonomous robotic 3D ultrasound acquisition of the breast.

[img] PDF
Abstract:The MURAB-project combines pre-operative high quality MR images which provide excellent soft tissue imaging with robotic guided intra-operative ultrasound images to do a biopsy of the possible cancer lesions. The precision and therefore the quality of the biopsy is dependent on the quality of the ultrasound imaging. The robotic guided ultrasound imaging follows a scan path that is generated from the MR images. Unfortunately, after the MR images are acquired the patient is moved, which results in a possible error between the shape of the breast during the MR imaging and the robotic guided ultrasound imaging. The aim of this project is to optimise the ultrasound acquisition during the scanning of the virtual scan path. Where optimal ultrasound images satisfy two conditions: firstly, they are acquired under low contact force between the transducer and the breast to ensure a minimum in deformation of the breast. Secondly, the transducer should have a sufficient contact area with the breast to scan the breast within reasonable time. This report presents a confidence driven controller that optimises two visual parameters (z and θ) that translate the ultrasound transducer from the scan path to the visual servoing path where high quality ultrasound images are acquired. The workings of the confidence driven controller are based on the confidence maps which present a per-pixel representation of the quality of the US echo. Only the first ten depth millimetres are used of the confidence map, because this region present a binary contact-non contact situation with scan lines that have high quality ultrasound echoes when tissue is imaged or no ultrasound echoes when air is imaged (no contact). This binary situation presents an excellent representation of the amount of transducer contact and location along the transducer surface. With the contact area information the visual parameters are calculated in order to centre the contact area around the centre scan line (θ) and increase or decrease the amount of contact area (z). The complete movement control of the robotic manipulator is done in an energy-based impedance controller that incorporates three safety layers (safety, motion and passivity layer) that establish safe human-robot interaction that guarantee overall passivity of the system. Four experiments with different quality scan paths are used to experimentally validated the ca- pabilities of the confidence driven controller. These experiments show that during scanning of low quality paths the confidence controller increases or maintains the transducer contact with the phantom. During scanning of a high quality path the controller shows first indications of also lowering the contact force during the scan, this cannot be tested with a force sensor be- cause there is no force sensor available in robotic manipulator, but can is tested by comparing the deformation within two US images acquired at the same location. The confidence driven controller shows improvements of the contact area along the scan path with a maximum im- provement of 397 percent if the amount of contact area is compared between the scan path with (81.8%) and without (20.3%) confidence driven controller. The controller also opens up a possibility to scan from a virtual path that is not constructed with the use of a MR images.
Item Type:Essay (Master)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:53 electrotechnology, 54 computer science
Programme:Systems and Control MSc (60359)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page