出版社:The Institute of Image Information and Television Engineers
摘要:To assist radiologists in interpreting diffusion-weighted magnetic resonance images (DWI), we have developed a computerized scheme for proper adjustment of these images that is based on density histogram analysis of concurrent images (b0 images). In our computerized scheme, the gray level of a b0 image is first normalized and the brain region is segmented using thresholding and labeling techniques. The density histogram is then calculated from the brain region on the original b0 image and the pixel value of maximum frequency in the density histogram is determined. The display grayscale in the DWI is finally adjusted by using the determined pixel value. With the existing adjustment method using thalamic signal intensity, the variation between 60 cases of evaluation index for image contrast calculated from the adjusted DWI was ±24.2% and with our method, it was ±16.5%. Our computerized scheme is therefore a very promising technique for determining the display grayscale of brain DWI.