Perivascular Spaces Quantification Of ESRD Patients At 3.0T MR

Objectives:It has been found that the expansion of perivascular spaces(PVS)is associated with many diseases and conditions,including senescence,dementia,Alzheimer’s disease,end-stage renal disease(ESRD),and so on.However,the current assessment of PVS is mainly based on the subject observation of the number,size and morphology of PVS in MR images.In addition,there is little research of quantitative analysis of PVS in China.In this study,we used 3.0T magnetic resonance system to obtain brain MR images,and wrote the scripts in MATLAB to segment the PVS automatically to obtain the quantitative result of PVS in the white matter and its volume density.Methods: Part Ⅰ:Two healthy volunteers and two ESRD patients with middle age,male or female.Exclusion criteria:(1)MRI check contraindications(metal implant retention,claustrophobia,etc.);(2)cerebral infarction,cerebral softening and other neurological diseases;(3)MR examination found intracranial space-occupying lesions,cerebrovascular malformations,craniocerebral trauma.Data of all subjects were acquired by GE Discovery MR750 3.0T magnetic resonance imaging system with a standard 8-channelled phased array coil.Data stability tests were performed before scanning.All subjects were first scout-scanned.After localization and gradient shimming,routine T1 WI sequence scans were performed to exclude brain parenchymal abnormalities such as brain tumors,brain infarction,softening,and cerebrovascular malformations.T2 WI sequence scans were performed for the acquisition of high-resolution coronal slices covering the entire brain.Custom MATLAB scripts were wrote to segment the PVS by generating the mask to extract the white matter of the image,filtering out cerebrospinal fluid and gray matter to calculate the total amount of PVS pixels in white matter and its total volume by special algorithm.As compared with the total volume of white matter,volume ratio of PVS could be easily calculated to analyze PVS in white matter quantitatively,trying to explore feasibility of quantitative analysis of PVS on 3.0T MR images.Part Ⅱ:Ten ESRD patients subjected to regular hemodialysis treatment three times a week for more than 3 months in in the outpatient,inpatient,and blood purification center of Jinan Military General Hospital were collected,including six males and four females aged from 42 to 65 years old with average age of 54.4±6.67 years old.Ten healthy subjects with similar age were selected as the healthy control group(HC group).Data collection was performed as described above.Computer-aided diagnosis program based on spatial gradient algorithm was performed to compare the difference of PVS density between ESRD and HC group.Results: Part Ⅰ:Compared to visual counting,Computer-aided diagnosis program based on spatial gradient algorithm yielded an average of 88.45±5.28% of sensitivity and 88.66±4.29% of specificity.Part Ⅱ:The PVS showed hyperintensive points with sparse distribution in the white matter in HC group,and hyperintensive linear or tubular structure in ESRD group.Average PVS density was 8.70±2.07v/v% in ESRD group,significantly higher than 4.55±1.36v/v% in HC group.Conclusions:1.Computer-aided diagnosis program based on spatial gradient algorithm is feasible on 3.0T magnetic resonance system.2.PVS density of ESRD patients is significantly increased compared to healthy controls.