Study On Segmentation Method And Quantitative Analysis Of Lesion Area In Leukoaraiosis’s MR Image

Leukoaraiosis (LA) is a kind of white matter lesion caused by a variety of factors andpathogenesis, its main clinical features include cognitive impairment and mental dysfunction.Leukoaraiosis is the old people’s common diseases. The methods of Leukoaraiosis clinicaldiagnosis mainly include behavior test score, CT and MR imaging technique. Compared with CT,the MR image technique has better detection effect for the lesion area, especially for the shape andlocation of a single lesion on the T2flair sequence of MR image. It can greatlyimprove Leukoaraiosis’s accuracy rate of diagnosis, therefore, it is the preferred clinical method tothe detection of Leukoaraiosis.Currently, clinicians diagnose the extent of Leukoaraiosis lesion mainly by clinician’s observingand analyzing the patients’MR image. The extent of Leukoaraiosis lesion is related to the positionand number of lesions, proportion of lesion and white matter, etc. Obviously, this diagnosis methodof Leukoaraiosis has strong subjective, and is easy to make misdiagnosis. Also, quantitative analysisand visualization research is unable to realize by this diagnosis method.In this paper, the lesion is been segmentation and quantitative analyzed on the basis of medicalimage segmentation algorithm research and Leukoaraiosis lesions characteristics analysis in MRimages. The method has some value for assistant diagnosis of Leukoaraiosis’s patient, can realizethe quantitative analysis and visualization of the lesion.The T1flair and T2flair sequence of Leukoaraiosis patients’MR images have the characteristicof complementary: the T1flair sequence can clearly show white matter area, but white matter areahas weak difference with gray matter area and effusion surrounding on the T2flair sequence; whitematter lesion area only has slight dark on the T1flair, but it presents hyperintense signal on theT2flair sequence. Based on the complementary of Leukoaraiosis’s MR images, in this paper, amethod of lesion’s segmentation and quantitative anlysis is been developed. White matter area isbeen segmentation on the T1flair sequence and white matter lesion area is been segmentation on theT2flair sequence respectively. The major work can be divided into the following three parts.(1) White matter area’s segmentation on the T1flair sequence: first, the skull and othertissues was removed with the method of mark watershed; second, white matter area wasextracted with the three different segmentation methods, including the morphologicalmethod, FCM clustering method and multi-threshold wavelet analysis based on thegray-level histogram; finally, the experimental segmentation results of threesegmentation methods were compared and analyzed with the gold standard, and the multi-threshold wavelet analysis based on the gray-level histogram was proved to be themost effective method of white matter area segmentation.(2) White matter lesion area’s segmentation on the T2flair sequence: the combination ofOtsu method and improved C-V model was proposed. The Otsu method was used forimage’s pre-segmentation, and then the image’s pre-segmentation result was directlyused as the initial contour for the improved C-V model, the finally segmentation resultwas obtained by curve evolution. The results show that the proposed segmentationmethod can get better separation effect and has some value for assistant diagnosis ofLeukoaraiosis. The proposed method can realize the fast auto-segmentation of whitematter lesion area.(3) The quantitative analysis of segmentation results: because the T1flair and T2flairsequence of Leukoaraiosis patients’MR images have the characteristic ofcomplementary, therefore we firstly calculated the volume of white matter area andlesion area on the T1flair and T2flair sequence respectively; second, the percentage ofwhite lesion area’s volume with whole white matter area’s volume was calculated, thequantitative result can been as an assist standard to measure the extent of Leukoaraiosislesion; finally, the quantitative results were compared with the physician diagnosisresults. The quantitative method was proved to be effectively and accurately, it can be asan objective standard index for Leukoaraiosis’s diagnosis.