Template For Foetal Structural Imaging And Improved Brain Fiber Reconstruction

The study towards foetal brain development using magnetic resonance imaging(MRI)performs significant in clinical research,especially for the development of the fine structures which is helpful for prenatal diagnosis.However,neuroimaging for foetus brain is a challenging problem,especially for micro pathological change,because the complex motion artefacts gener-ated by the freely moving foetus during data acquisition decreases the imaging quality.Both the foetal brain structural MR imaging and diffusion tensor imaging have this problem.Therefore,an improved image post-processing algorithm framework becomes essential to ameliorate this phenomenon.The algorithm proposed in this paper will be applied in structural MRI and DTI,in order to study the development of the foetal brain.For the structural MRI,the objective is to achieve a relatively accurate volume of the lateral ventricle.The foetal brain MRI data was acquired using 1.5T balanced turbo field echo sequence with sensitivity encoding(sBTFE)sequence.Firstly,the image denoising step was completed using the convex optimisation method based on the total variation semi norm,which is able to relatively decrease the noise.The slices acquired from two dimensional MRI were re-aligned to decrease the bias between neighboured slices.After image denoising,the adaptive image segmentation based on image intensity and Bayesian field was adopted to achieve the relatively accurate two dimensional lateral ventricle MR images,and the super resolution algorithm was adopted to combine the images of the same gestation week to increase the resolution,thus reconstruct the fine lateral ventricle volume.As for the diffusion tensor imaging,the objective is to observe the development of the foetal brain using DTI.The parameters during data acquisition was initially adjusted,and the proposed image reconstruction algorithm was then applied.The DTI data of different pathological state was classified and analysed concerning different gestation weeks,by which the statistical results of the fiber development of the foetal brain was achieved for different side,region and states.The result can be used to analyse the foetal brain development,and are also helpful towards foetal brain diseases study.The algorithm designed in this paper did several exploration and development for foetal brain MR imaging.The reconstructed foetal brain lateral ventricle was of a relatively improved accuracy,and the optimisation method also helps to improve the fiber tracking results.The images achieved were of relatively stable quality that helps to investigate the micro pathological changes,and also able to provide reference for clinical study.