Research On Automatic Of Extraction Region Of Interest(ROI) In Small Animal Chemical Exchange Saturation Transfer (CEST) MRI

Chemical exchange saturation transfer(CEST)magnetic resonance imaging(MRI)is a new type of molecular imaging technology,The technology can perform in-vivo imaging of molecules containing exchangeable protons,including proteins and sugars.At present,in CEST MRI image analysis,the researcher or doctor usually draws the region of interest(ROI)on the CEST MRI image based on experience,and then analyzes the CEST Z spectrum characteristics and other quantification of the region parameters to extract changes in biochemical metabolism of the lesion.However,the efficiency of manual delineation is low,and it may cause a large human error,which cannot accurately reflect the true CEST signal in the lesion area.Therefore,the main purpose of this paper is to find a fast and accurate method of automatic or semi-automatic extraction of ROI.The specific research work is as follows:(1)Automatic extraction of CEST MRI image ROI based on region growing algorithm.By improving the region growing algorithm,the initial seed point is automatically located based on the image intensity information,and through the CEST image experiment based on mouse brain tumors,a reasonable growth criterion and optimal stop growth threshold are determined.Combining the closed operation in the morphological transformation with the region growing algorithm improves the robustness.The experimental results of the multi-saturated intensity CEST MRI image show that the improved region growing algorithm can automatically locate the focus area of different contrast images according to the CEST signal at different frequency offsets.Our method improve the accuracy of selecting ROI,the areas with different CEST signal intensities inside the tumor can be further subdivided.(2)Use active shape model(ASM)method to select ROI based on anatomical structure in CEST MRI image.The implementation steps of the ASM algorithm include: building a shape model,building a local gray model and ASM search.We use CEST MRI images of mouse brain for experiments.The experimental results show that the ASM algorithm can semi-automatically select six brain structures required for CEST analysis.Correlation Coefficient between the selected ROI and the gold standard is between 0.933 and 0.976,and the segmentation effect is ideal.(3)Aiming at the problems of tedious post-processing of CEST MRI data,a set of MATLAB-based ROI selection and Z-spectrum display system was established.The system includes the entire post-processing process of CEST MRI,and integrates the ROI selection module based on the region growing algorithm and ASM algorithm,so that researchers can use CEST MRI for clinical research.