Date of Award
2008
Degree Type
Thesis
Degree Name
Doctor of Philosophy
Program
Biomedical Engineering
Supervisor
Dr. Terry Peters
Second Advisor
Dr. Roy Eagleson
Third Advisor
Dr. Hanif Ladak
Abstract
The treatment of many life-threatening diseases requires physical interventions into the patient’s body. Image-guided minimally invasive surgery (IGMIS) can effectively increase physicians’ performance, while reducing risk, pain, complications and health care costs for the patient. However, the surgeons lose the direct real-time views in this procedure, which seriously degrade their performance. Traditional IG- MIS systems usually rely on surface representations of organs, losing much of the diagnostic information of the inner data. The specific aim of this thesis is to develop new strategies for real-time 4D (3D dynamic) multimodality medical image visualization and manipulation, incorporating techniques such as image processing and navigation, virtual surgical tools, structure of interest feature enhancement, and therefore permitting the full volumetric patient dataset to be displayed and operated adaptively during the therapeutic and surgical procedure without losing anatomical and diagnostic information. I first review volumetric image visualization pipelines, algorithms and their medical applications, illustrate results of my algorithms, and compare my research work with those from the published literature. The next stage is to design and develop a comprehensive stereoscopic 4D cardiac image visualization and manipulation platform that exploits the power of modern graphics processing units (GPUs), during which a dynamic multiresolution scheme is applied to emphasize the region of interest, while at the same time, implementing a colour and opacity adjustment strategy to improve the visual uniformity of the rendered images. In the application, ECG signals are used to synchronize the volume rendered beating heart to the patient, and a novel algorithm is presented to compensate the rendering latency. Finally, this platform is used to display and regulate the registered pre-operative MR and intra-operative 3D ultrasoιmd cardiac images, providing surgeons with complementary diagnostic information in real time. In many medical imaging scenarios, it is still a challenge to achieve an optimized in balance between real-time artifact-free volume visualization and interactive voxel classification. The next step of my research is to present a new post colour-attenuated classification (PosCAC) algorithm for interactive volume classification, which achieves around 100 times higher efficiency than that of the unaccelerated pre-integrated classification approach, while introducing ~10% lower artifacts. In this framework, an artifact-free volume clipping algorithm is presented and employed to perform dynamic “key-hole” clipping and context feature enhancement. Most clinical applications tend to focus on a particular set of organ structures, and an uniform transfer function (TF) cannot effectively isolate such structures because of the lack of spatial information and the small intensity differences between adjacent tissues. Explicit segmentation is therefore a powerful way to address the difficulty. Traditional approaches for rendering such segmented volumetric datasets usually deliver unsatisfactory results. In this thesis, I introduce a new “colour encoding” technique to solve this problem, which deliveries higher quality images and smoother subvolume boundaries than those presented in the published literature. In addition, I present a new depth texture indexing algorithm that permits virtual surgical tools to be included and interactively manipulated within the volume rendered images. The research results of this thesis will have a comprehensive impact on surgical simulation, training, cardiac intervention planning and guidance, as well as enable new research directions in the image-based minimally invasive surgery.
Recommended Citation
Zhang, Qi, "Real-Time Visualization and Manipulation of Volumetric Medical Images" (2008). Digitized Theses. 4741.
https://ir.lib.uwo.ca/digitizedtheses/4741