Abstract
Cancer is a major cause of mortality worldwide in the modern world. It can be detected by X-ray medical imaging, but the resolution of X-ray images are quite poor(about 2mm) at detecting cancer in early phase. In this thesis, the purpose is to improve resolution by one order of magnitude to 0.2mm in conventional X-ray images, in order to detect cancer earlier for early treatment and cost much less than X-Ray Computed Tomography. The hardware architecture is the high resolution X-Ray detection with In-cell TFT panel array. Because of high resolution, the volume of data is too large. To send out the detector array through USB port or to store into flash memory at the signal reserving time(1.5ms), the compressive techniques is needed. To minimize the compression ratio and execution time, and consider the characteristic of radiograph. We propose the compressive sampling theory, which can recover certain signals from far fewer samples than traditional methods use. The method of reconstruction is mainly leading by Over-Detected Orthogonal Matching Pursuit algorithm, which is a modified algorithm of the well-known Orthogonal Matching Pursuit (OMP) algorithm. Besides, this system consists of Daubechies wavelet transform and thresholding. The results show that this method improves the execution time significantly, successfully recover the radiograph of chest and breast. The the requirement of compression ratio (0.27) is achieved.