Computer-aided Diagnosis of Bladder Neoplasm by Using Imaging Features / 医疗卫生装备

ABSTRACT

Objective To provide theoretical evidences for diagnosis of bladder neoplasm by using more information of imaging features. Methods MRI images of bladder neoplasm were collected for this research. The ROI (region of interest) area was selected manually,and noise reduction and sharpening were applied to the ROI area by using LoG (Laplacian of Gaussian) filter. The texture features of bladder neoplasm and normal bladder wall (smooth muscle),such as mean grey-level intensity,entropy,uniformity were calculated. A statistical analysis was made at last. Results The values of texture features were analyzed by t-test. Entropy and uniformity show significant differences between the two groups. But Mean grey-level intensity hasn't indicated this difference. Conclusion This texture features may be applied to decide the invasive depth of bladder neoplasm,it also means that the stage of bladder neoplasm may be fixed by this system.