Research On Neonatal HIE Image Segmentation Method Based On FCM Anchoring

Brain image segmentation is a hot spot in medical image segmentation,but most of them focus on adult brain MR image segmentation.Because of the lack of reliable anatomical map or atlas to guide neonatal image segmentation,neonatal brain volume is much smaller than that of adults,and there are problems such as serious noise and partial volume effect problems,which lead to the difficulty of segmentation is much greater than that of adult MR images.Neonatal hypoxic-ischemic encephalopathy is one of the common causes of neurological disability in the neonatal period.In severe cases,it will endanger the life of the neonate and cause irreversible brain damage.The segmentation and measurement of periventricular white matter damage in MRI is a crucial but tedious task in the early diagnosis of neonatal hypoxic-ischemic encephalopathy.Currently,manual segmentation methods are mainly used to process neonatal HIE images in the clinical stage.With the development of medical imaging technology,relying on manual segmentation methods to process large-scale neonatal image data has problems such as long time,and manual segmentation methods cannot meet current needs.Based on the current needs,this article selects the magnetic resonance imaging that is currently widely used in neonatal hypoxic-ischemic encephalopathy to study the segmentation method of neonatal HIE images.The specific research content is as follows:(1)An adaptive regularization kernel-based fuzzy C-means algorithm based on membership constraint and bias field correction is proposed(GM-ARKFCM).Under the idea of competitive learning based on penalizing opponents,the membership constraint function penalty term is introduced into the ARKFCM algorithm to obtain the G-ARKFCM algorithm,which speeds up the algorithm’s convergence speed,and makes the value of the algorithm fuzzy index m relatively flexible.The energy minimization method multiplicative intrinsic component optimization(MICO)is introduced into the G-ARKFCM algorithm to obtain the GM-ARKFCM algorithm.The GM-ARKFCM algorithm can estimate and correct the offset field when segmenting neonatal HIE images.The results of segmenting brain MR images show that the GM-ARKFCM algorithm has higher segmentation accuracy.Experimental results show that the GM-ARKFCM algorithm can segment the general shape of the brain ventricles of neonatal HIE images,and can retain more information of the original images.(2)An interactive segmentation method for clinical magnetic resonance imaging based on FCM anchoring is proposed(AE3S).This method uses interactive sample estimation at the subvoxel level for quantitative diagnosis.The coarse-grained voxels in clinical MRI are divided into sub-voxels,and gradient adaptive filtering is used to adjust the parameters of the sub-voxels.Considering that the ventricle is clearer than white matter and gray matter in T1-weighted images,T2-weighted images and FLAIR images,in addition,its overall variance follows a chi-square distribution.Use the parameters of the sampling point data obtained by the GM-ARKFCM algorithm to segment the periventricular area,use the periventricular area as an anchor point,and expand the anchor point to obtain ventricular white matter of various widths.By aligning the segmented periventricular white matter in various ways,the imaging radiomics features of the region of interest are calculated to quantitatively diagnose HIE.The AE3 S algorithm can help doctors accurately assess the stage of HIE and determine which combination of modalities and ROIs is closely related to the stage of HIE.Introduces other neonatal HIE diagnostic methods,the AE3 S algorithm is more comprehensive than other neonatal HIE diagnostic methods.Current clinical practice shows that this method is effective and meets the requirements of clinical diagnosis of HIE.