Investigation Of Diffusion Sensitivity Index (B-Value) Applied On Segmentation Of Multispectral MR Images In Cerebral Ischemic Infarction

Objective To analysis the MR signal of cerebral ischemic infarction and the effectiveness of b-value on signal intensity, and to discuss the affection of different b-value on ischemic lesion zones segmentationMethods Three different weighted MR images, including T2-weighted imaging (T2WI),diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) map, formed multispectral MR images on a same slice. To analysis the signal intensity of different infarction stages (hyperacute, acute, subacute and chronic) and signal changes of ischemic tissues relative to unaffected parenchyma. Bilinear interpolation and non-linear anisotropic diffusion filtering applied on diffusion image data (DWI and ADC map) to interpolate and smooth these images. Using maximal mutual information method to register multispectral MR images on space, then by fuzzy c-means clustering algorithm to show the results of different b-value segmentation compared with ischemic lesion delineated by experts.Results On mutispectral MR images, there were correlation between the images of different stages of cerebral ischemic infarction and the time of stroke. In acute and subacute stage, the signal in ischemic lesion areas changed to hyperintensity in T2WI and DWI, normal or hypointensity in T1WI. In hyperacute stage, the signal was hyperintense in the DWI, and in T1WI and T2WI was normal. In the chronic stage, the signal in T1WI and DWI showed hypointensity and hyperintensity in T2WI. Because ADC declined obviously in acute and subacute stage, the ischemic lesion zone was hypointensity in ADC map. With the development of the disease, ADC gradually recovered and then changed to hyperintensity in ADC map in chronic stage. The results of signal intensity of different infarction stages compared with that of normal cerebral tissue showed that relatively signal intensity change between 107% and 219% in T2WI, between 89% and 265% in DWI, and between 51% and 154% in ADC map. Coincidence index (k-value) was the result using kappa test to compare manual segmentation with FCM segmentation. When b-value was 500mm2/s, k-value was relatively small in different infarction stage. However when b-value equaled 1000mm2/s or 1500mm2/s, k-value of hyperacute stage was lower than that of the others. But in hyperacute stage, k-value obtained maximum when b-value was 2000mm2/s.Conclusion In different stage of cerebral ischemic infarction, we chosen the appropriate b-value to get the better segmentation of ischemic lesion areas, which accorded with the diagnostic require of radiologist. The results indicated the identification of ischemic lesion with accuracy and reproducibility. The method was useful for a variety of applications in the field of computer aided detection and computer aided diagnosis and evaluated the effect of medical treatment when ischemic lesion cured by an appropriate thrombolytic or other therapy.