Interval changes in medical images (such as formation of new lesions and changes in existing abnormalities) can be enhanced by removing most of the normal structures. For this purpose, the temporal subtraction image is obtained by subtraction of previous image from the current image. If the image registration is incorrect, not only the interval changes but also the normal structures appear as artifacts on the temporal subtraction image. In this study, we developed a temporal subtraction algorithm for MR images of brain and applied it to five clinical cases. Preliminary results indicated that the algorithm considerably enhanced the interval changes in the subtraction images and substantially reduced the number of misregistered artifacts. Therefore, it was apparent that artifacts in the temporal subtraction images were largely reduced by incorporating the image warping technique into image subtraction algorithm.

View full abstract

To obtain information on perfusion-related, free, and restricted diffusion in uterine cervical cancer using diffusion-weighted magnetic resonance imaging (DWI) , we analyzed three diffusion components using triexponential function. Twenty-two subjects were evaluated using the DWI with multiple b-values. We derived perfusion-related diffusion, fast free diffusion, and slow restricted diffusion coefficients (D_p-tri, D_f-tri, D_s-tri) obtained from triexponential fitting parameters to DWI data. Moreover, the triexponential analysis was compared with biexponential analysis (D_f-bi and D_s-bi; fast and slow diffusion coefficients) and monoexponential analysis (D_m; apparent diffusion coefficient) . In triexponential analysis could well separate combined diffusion coefficient into three diffusion components in cervical cancer, because the order of the diffusion coefficients was D_p-tri>D_f-bi>D_f-tri>D_m>D_s-bi>D_s-tri. triexponential analysis makes it possible to noninvasively obtain more detailed diffusion information in patients with uterine cervical cancer.

View full abstract