Correction of Endoscopic Images and Measurement of Lesion Area

碩士 === 大葉大學 === 電機工程研究所 === 89 === The measurement of lesion size using endoscopic images is important in performing clinical trial. Endoscopes are constructed using wide-angle lenses in order to increase the field of view. However, the displayed images are seriously distorted. The distortion gets e...

Full description

Bibliographic Details
Main Authors: Jr-Hung Lin, 林志鴻
Other Authors: Ren-Jean Liu
Format: Others
Language:zh-TW
Published: 2001
Online Access:http://ndltd.ncl.edu.tw/handle/28966065816880134719
Description
Summary:碩士 === 大葉大學 === 電機工程研究所 === 89 === The measurement of lesion size using endoscopic images is important in performing clinical trial. Endoscopes are constructed using wide-angle lenses in order to increase the field of view. However, the displayed images are seriously distorted. The distortion gets even more serious as the objects extend outward from the center of the lens in radial. In consequence, the measurement of lesion size is a difficult task. The traditional measuring-by-eye method will create significant error. This project will discuss the effect of image distortion and correct the effect. A more accurate size-measuring algorithm will be developed and compared to those of traditional approaches. We will prove that it is easier to analyze the effect of radial distortion in polar coordinate. The distorted and corrected images can be mapped by two coordinate systems. They can be transformed to each other by a special orthogonal polynomial. The proposed method is very effective to plane images. The accuracy is high. To measure lesion size, graphic user interface (GUI) is used. The number of pixels of lesion is first calculated using the method of region growing. The area is then computed based upon the ratio to the entire image. In conducting simulations, we will first analyze the lenses on the endoscope. The approach is then verified on images with grids. Finally, the system will be applied to images of real human organs.