Efficient data encoder for endoscopic imaging applications
The invention of medical imaging technology revolved the process of diagnosing diseases and opened a new world for better studying inside of the human body. In order to capture images from different human organs, different devices have been developed. Gastro-Endoscopy is an example of a medical imaging device which captures images from human gastrointestinal. With the advancement of technology, the issues regarding such devices started to get rectified. For example, with the invention of swallow-able pill photographer which is called Wireless Capsule Endoscopy (WCE); pain, time, and bleeding risk for patients are radically decreased. The development of such technologies and devices has been increased and the demands for instruments providing better performance are grown along the time. In case ofWCE, the special feature requirements such as a small size (as small as an ordinary pill) and wireless transmission of the captured images dictate restrictions in power consumption and area usage. In this research, the reduction of image encoder hardware cost for endoscopic imaging application has been focused. Several encoding algorithms have been studied and the comparative results are discussed. An efficient data encoder based on Lempel-Ziv-Welch (LZW) algorithm is presented. The encoder is a library-based one where the size of library can be modified by the user, and hence, the output data rate can be controlled according to the bandwidth requirement. The simulation is carried out with several endoscopic images and the results show that a minimum compression ratio of 92.5 % can be achieved with a minimum reconstruction quality of 30 dB. The hardware architecture and implementation result in Field-Programmable Gate Array (FPGA) for the proposed window-based LZW are also presented. A new lossy LZW algorithm is proposed and implemented in FPGA which provides promising results for such an application.
DegreeMaster of Science (M.Sc.)
CommitteeKo, Seok-Bum; Li, Chen; Fotouhi, Reza
Copyright DateSeptember 2010