Final Report - Virtual Colonography

Final Report - Virtual Colonography - ECE 578 Computer...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
ECE 578 Computer Analysis of Biomedical Images, Spring 2007 1 Virtual Colonography Siddhartha Ghosh, Albert Ho, Baaba Andam Abstract — Virtual colonography is a non-invasive method of analyzing computed tomography (CT) scans of the colon using computer algorithms. It is used to detect polyps, which may cause cancer if present in the colon. Our project was to implement an algorithm to accurately detect polyps from CT scans of the colon. We utilized VisionX tools to segment the colon from CT scans of the abdomen. This process involved applying thresholds to the original images to extract the air and fluid filled portions, morphological filtering and 3D connected component analysis to generate a binary mask. The surface of the colon was then approximated through polygonization of the mask. Principal curvature values were extracted for the surface objects and used to determine its shape index and curvedness. From this information, polyp candidates were generated by applying a ranking methodology. The results of this process were compared to the ground truth in the given image set. I. INTRODUCTION Colon cancer is currently the second leading cause of cancer deaths in the United States. Approximately 60,000 deaths are attributed to it each year. This disease can most often be caused by polyps, which are tissue growths on a mucous membrane. Polyps are formed from an overgrowth of cells on the colon wall, called an adenoma. They may be harmless but sometimes cause cancer. The early detection and removal of these polyps therefore reduces the risk of cancer. Traditionally, polyps are detected and removed by a colonoscopy procedure [1]. Colonoscopy is the endoscopic examination of the colon, which involves inserting a colonoscope into a patient’s large intestine, through the rectum, in order to detect polyps and remove them if they are present. Colonography offers a non-invasive alternative to this. Computed tomography (CT) scans of the colon can be taken within 20-30 seconds, providing axial image slices of a patient. Two sets of images are taken with the patient lying in the prone (face down) and supine (face up) positions. This gives about 300-700 images per patient [2], [3], [4]. After acquiring CT scans of the colon, they can be analyzed either visually by a radiologist or with computer algorithms to detect polyps. The use of computer algorithms for analysis is termed virtual colonography. Visual analysis of CT scans can take between 15 – 40 minutes per scan, even for experts in the field. Given the number of scans to be analyzed per patient, virtual colonography is a preferable approach because of its speed and lack of human intervention [3], [4]. A. Issues in Virtual Colonography One of the primary problems in virtual colonography is caused by the resemblance between certain colonic folds, the colon wall and polyps. Since some colonic folds resemble polyps, they can be falsely detected as such, leading to false positive detections in the implementation of polyp detection
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 2
This is the end of the preview. Sign up to access the rest of the document.

Page1 / 20

Final Report - Virtual Colonography - ECE 578 Computer...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online