Date of Award

5-2017

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Legacy Department

Bioengineering

Committee Member

Dr. David Kwartowitz, Committee Chair

Committee Member

Dr. Delphine Dean

Committee Member

Dr. J. Barr von Oehsen

Committee Member

Dr. S. Duke Herrell

Abstract

Laparoscopic surgery is a minimally invasive technique where surgeons insert a small video camera into the patient's body to visualize internal organs and use small tools to perform these procedures. However, the benefit of small incisions has a disadvantage of limited visualization of subsurface tissues. Image-guided surgery (IGS) uses pre-operative and intra-operative images to map subsurface structures and can reduce the limitations of laparoscopic surgery. One particular laparoscopic system is the daVinci-si robotic surgical vision system. The video streams generate approximately 360 megabytes of data per second, demonstrating a trend toward increased data sizes in medicine, primarily due to higher-resolution video cameras and imaging equipment. Real-time processing this large stream of data on a bedside PC, single or dual node setup, may be challenging and a high-performance computing (HPC) environment is not typically available at the point of care. To process this data on remote HPC clusters at the typical 30 frames per second rate (fps), it is required that each 11.9 MB (1080p) video frame be processed by a server and returned within the time this frame is displayed or 1/30th of a second. The ability to acquire, process, and visualize data in real time is essential for the performance of complex tasks as well as minimizing risk to the patient. We have implemented and compared performance of compression, segmentation and registration algorithms on Clemson's Palmetto supercomputer using dual Nvidia graphics processing units (GPUs) per node and compute unified device architecture (CUDA) programming model. We developed three separate applications that run simultaneously: video acquisition, image processing, and video display. The image processing application allows several algorithms to run simultaneously on different cluster nodes and transfer images through message passing interface (MPI). Our segmentation and registration algorithms resulted in an acceleration factor of around 2 and 8 times respectively. To achieve a higher frame rate, we also resized images and reduced the overall processing time. As a result, using high-speed network to access computing clusters with GPUs to implement these algorithms in parallel will improve surgical procedures by providing real-time medical image processing and laparoscopic data.

Share

COinS