Towards improving the Statscan X-ray image quality through sliding mode control of the C-arm

This dissertation investigates methods for improving the image quality of a digital radiography system. The Lodox Statscan™ X-ray system provides a full-body scanned image for initial diagnosis. The system is driven by a permanent magnet linear motor (PMLM) controlled by a cascaded proportional-proportional integral controller (P-PI). Transient errors in the trapezoidal motion profile of the scanning C-arm may cause mismatches between the detector and the collimated beams from the X-ray source. This results in a partial degradation of image quality. The Statscan™ X-ray system was investigated and the following constraints were identified: The scanning time is limited to 13 seconds and the maximum scan length is limited to 1.8 m. Since it was not possible to obtain the Lodox Statscan™ dynamics model, because of the similarity, a characteristic model was then developed using a DC motor in order to investigate the control dynamics. It is not advisable for a designer to manipulate the controllers on commercial machines except for changing the parameters for tuning. Therefore, a P-PI controller and a proportional-integral sliding mode controller (P-ISMC) as well as a Boundary Layer variant (P-ISMC+BD) were designed for fair comparison purposes. Root locus, Bode diagrams and integral sliding mode control methods, respectively, were used to design the P-PI and P-ISMC controller groups. Each controller consists of an inner loop and an outer position loop. Proportional integral (PI) and integral sliding mode controllers (ISMC) were used for the inner loop. The two inner loop controllers were tuned, and then tested, before cascading them with the outer position loop. The simulation and experiments were conducted to compare each controller’s performance on step set-point tracking, trapezoidal motion profile tracking, the time transient’s specifications and robustness against disturbances. In order to test image quality, 27 distance profiles were generated from P-PI, P-ISMC and P-ISMC+BD. In addition, four images captured by the Statscan™ were also selected. A time delay integrator (TDI) simulator was used on the distance profiles and the four images to generate 108 distorted image profiles.

Grantor dc:publisher.institution

Department of Medicine

Year dc:date.issued

2012

Author dc:creator

• Esmail, Mohammed

Advisors dc:contributor.advisor

• Tsoeu, Mohohlo Samuel
• John, Lester

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/11427/3382

OAI identifier oai:identifier

oai:open.uct.ac.za:11427/3382