| Thousands of endoscopic procedures are performed each year. They are invasive and often uncomfortable for patients. They sometimes have serious side effects such as perforation, infection, and hemorrhage. Virtual endoscopic visualization avoids the risks associated with real endoscopy, and when used prior to performing an actual endoscopic procedure for therapeutics can minimize procedural difficulties and decrease the rate of morbidity. Additionally, there are many body regions inaccessible to or complicated with real endoscopy but can be explored with virtual endoscopy. In this dissertation, novel algorithms are proposed in segmentation and digital cleansing, volume rendering, surface flattening, and computer-aided detection (CAD) to improve and enhance virtual endoscopy applications.;Effective colonoscopic screening for polyps with optical or virtual means requires adequate visualization of the entire colon surface. We have investigated the colon surface visibility coverage using a simulation method to estimate the percentage of the colon surface is missed in the optical colonoscopy (OC) and virtual colonoscopy (VC). Our simulation study reveals that about 23% of the colon surface is missed in the standard OC examination and about 9% of the colon surface is missed in the VC examination when navigating in both directions.;We have adopted a partial volume model in the segmentation and digital cleansing to handle the partial volume effect. Our algorithm is demonstrated with contrast-enhanced CT colon data sets. The topological noise is automatically removed from the segmentation result by a 3D region growing based algorithm using the concept of simple point. Furthermore, the topologically simple colon surface is extracted with a dual contouring method for virtual colon flattening.;Most common methods in virtual endoscopy simulate the behavior of a real endoscope. Simulating a real endoscopy is not the most efficient technique in many endoscopy procedures. A real endoscopy is restricted due to physical limitations that a virtual endoscopy does not have. We present a conformal colon flattening technique which virtually unfolds the colon, allowing physicians to inspect its surface and detect polyps on a single 2D image.;Direct volume rendering (DVR) can provide high-quality virtual endoscopic views for virtual endoscopy applications. However, DVR of contemporary clinical data sets in real-time at a high resolution is still a challenge. We present a GPU-based object-order ray-casting algorithm to render large volumetric data sets on the GPU. We also exploit the cooperation and trade-off between the GPU and the CPU to obtain further acceleration. Although our ray-casting approach is of general applicability, we have specifically applied it to our VC system.;We further present a novel pipeline for CAD of colonic polyps by integrating texture and shape analysis with volume rendering and conformal colon flattening. Using our automatic method, the 3D polyp detection problem is converted to a 2D image segmentation problem. The polyps are detected by a clustering method on the 2D flattened colon image. The false positives (FPs) are further reduced by analyzing the volumetric shape features. Our system detects 100% of the adenomatous polyps, and yields a low FP rate. The results are easily integrated into a VC system, which allows physicians to perform their diagnoses more accurately and efficiently. Since the suspicious areas are clearly identified to the physician, the physician needs only traverse the colon in one direction, without fear of missing a polyp.;All presented techniques have been tested with a number of data sets to show their feasibility. In this dissertation, we focus on CT colon data sets although our techniques could be used with a variety of other human organs, such as blood vessels and bladder. |
