| Summary: | 碩士 === 國立臺灣科技大學 === 自動化及控制研究所 === 103 === This study plans to develop a reference system which uses computed tomography to calculate the tumor volume change of lung cancer patients after chemotherapy accurately to assist doctors in clinical treatment and evaluation. The study is divided into two parts. Part 1 uses image processing techniques to analyze the computed tomography of lung cancer, locate the tumor area and calculate the tumor volume. First, the speckle noise caused by computed tomography is removed and the image contrast is enhanced by wiener filter and histogram equalization, and the lung extracting process is designed, the weighted fuzzy-C means is used for lung image binarization, combined with morphological approach to obtain the lung area mask, so as to locate the tumor area, which is marked with a green frame on the original image for tumor localizing. Afterwards, the first tumor image is used for region growing, the center is used as the seed of next image, taking the boundary contour of the region growing as the initial contour for the active contour without edges (ACWE) algorithm. The ACWE does not use image gradient to define the boundary, it extracts the edges with seg, gap and blur better, applicable to the detection of lung tumor edge. Finally, the Marching cube algorithm of isosurface extraction is used for 3D reconstruction of tumor image and calculating the volume, and the accuracy is validated.
Part 2 is medical indicator analysis, the relationships between tumor volume change and survival time of 99 patients after chemotherapy in Tri-Service General Hospital are compared, and the 1-year, 1.5-year, 2-year, 2.5-year and 3-year novel survival prediction indicators and the effect of chemotherapy on sex and age distributions are obtained by three statistical methods, which are Kaplan-Meier method, log-rank test and receiver operating characteristic curve, providing the doctors with the reference for prognostic survival time and diagnosis of lung cancer.
|
|---|
