Magnetic Resonance Imaging In Neonatal Bilirubin Encephalophy

Background:Neonatal bilirubin encephalopathy(NBE)is the most serious complication of hyperbilirubinemia.Now there is no gold standard for the diagnosis of NBE.It is necessary to combine the clinical manifestations,laboratory findings and imaging findings for comprehensive diagnosis.Three-Dimensional Magnetization-Prepared 2 Rapid Acquisition Gradient-Echo(3D MP2RAGE)has been recently introduced as a self-bias-corrected sequence to obtain homogeneous T1-weighted contrast with an intrinsic correction of B1-inhomogeneities and reduced residual proton density and T2*weighting,as well as high-resolution T1 relaxation time map in a clinically compatible scan time means.DKI has been described as an extension of DTI,which is based on the diffusion properties of water motion in complex media and has been developed to probe the non-Gaussian diffusion characteristics.Compared with DTI,DKI has shown greater promise to better characterize white and gray matter alterations in the brain.DKI has been used in several studies on humans and animals,including pathological and normal conditions,such as age-related microstructural changes,chronic mild stress,and attention-deficit hyperactivity disorder,among others.In recent years,functional MRI,such as DWI,MRS,DTI,have been used in NBE.To our best knowledge,the T1 relaxation time and DKI studies in NBE brain tissue have not been reported in the literature.Objectives:Our study retrospectively analyzed 3D MP2RAGE and DKI data to determine whether the T1 relaxation time and DKI profiles of newborns with NBE can evaluate microstructural changes and illustrate related mechanisms.This study also verified whether T1 relaxation time and DKI parametrics can serve as an in vivo marker for NBE.Methods:Retrospectively analyzed of the 30 NBE patients which were diagnosed in our neonatal intensive care unit(NICU),the peak levels of serum bilirubin on admission ranged from 451.1 ?mol/L to 679.5 ?mol/L.27 age-matched control neonates were included in this study,who underwent brain MR imaging for non-neurological indications(e.g.,facial hemangiomas,sinonasal abnormalities,and orbital lesions),and excluding hyperbilirubinemia,perinatal hypoxia-ischemia,or any other significant neurological illness.All participants underwent brain MRI at a 3 Tesla MR scanner(Skyra,SIEMENS).Conventional MRI,DWI,MP2RAGE and DKI sequence were performed for all patients and controls.After strictly controlling the subjects' head motion artifact and mechanical noise,22 subjects with NBE and 21 controls were enrolled in the study.The image quality of conventional T1WI and MP2RAGE T1WI was compared using double blind method by two experienced neuroimaging physicians,and the abnormal signal was analyzed.The image quality score were divided into 5 levels:5,excellent,anatomical structure clearly displayed,and the boundary of the internal and external capsule,the capsule basal ganglia nuclei was clear;4,good,the boundary of basal ganglia nuclei was slightly blurred,but did not affect the diagnosis;3,common,most of the basal ganglia structural can clearly displayed,and can be used for diagnosis;2 poor,the anatomy of the basal ganglia was blurred,and influence final diagnosis;1,very poor,can not be used for diagnosis.Regions of interest(ROI)were manually drawn on the T1 maps and ADC maps in the bilateral GP and putamen,and the cross-sectional of monro foramen were chosen as the measurement level.DKI data were processed by using Matlab and Diffusional Kurtosis Estimator software.Subsequently,the diffusion and kurtosis tensors were calculated on a voxel-by-voxel basis to produce parametric maps for DKI parameters.And voxels of interest(VOIs)were manually drawn in the bilateral globus pallidus,putamen,and thalamus by u sing MRIcron software.Data were analyzed using MedCalc 15.6.The consistency of the image quality score of the two judges was checked by Kappa test,and Z test was used to compare the image quality.Bilateral regions that showed a main group effect,but no lateralization effect,were analyzed as a single region,and diffusion metrics of these bilateral regions were averaged.Two sample t test was used to compare between groups.The correlation analysis of parameters was carried out with Spearman 's correlation analysis.The maximum sum of the sensitivity and specificity,the cut-off values of T1 relaxation time and DKI parametrics for differentiating NBE from control newborns were obtained with Receiver Operating Characteristic(ROC)curve analysis.Results:The MP2RAGE T1-weighted contrast could display the structure of basal ganglia clearly,and the resolution is higher than that of the conventional T1W images(P<0.001).Relative hyper-intensity signals were found in the bilateral GP on T1-weighted images in all patients and sixteen(16/21)control neonates.No abnormal T2WI signals were observed in all NBE and control neonates.Compared with the control group,the T1 relaxation time in GP was decreased in NBE group,but there was no statistical significance difference(P>0.05).The T1-relaxation time ratio of GP and putamen in NBE were found to be consistently shorter compared with those in control neonates(t=2.38,P<0.05).There was a significant correlation between the T1-relaxation time ratio of GP and putamen and the peak level of total serum bilirubin(TSB)in the NBE group(r=-0.571).ROC curve analysis showed that the cut-off point of the ratio for diagnosis of NBE was 0.7748,and the AUC,Yuden index J,sensitivity and specificity were 0.690,0.444,68.18%,76.19%,respectively.None of the patients with NBE and control neonates demonstrated abnormal signals in diffusion weighted images.In addition,there was no significant difference of the ADC value between the two groups(P>0.05).FA,MK,AK,and RK of all VOIs were significantly decreased in the NBE group as compared with the control group,whereas MD increased in NBE.Compared with MK and AK,the change of RK was largest between NBE and normal newborns.Between NBE and control newborns,the MK,AK,RK,and MD more markedly varied in the globus pallidus(9.2%,10.5%,15.2%,3.3%,respectively)than in the putamen(6.9%,7.1%,8.9%,2.8%,respectively)and thalamus(5.5%,6.9%,8.1%,2.7%,respectively).However,FA more obviously changed in the thalamus(reduced 9.6%)than in the globus pallidus(reduced 7.5%)and putamen(reduced 6.9%).ROC curve analysis showed that MK in globus pallidus has the largest AUC and Yuden index J,and the cut-off point of MK in globus pallidus for diagnosis of NBE was 0.4238.The AUC,Yuden index J,sensitivity and specificity were 0.840,0.671,90.91%,76.19%,respectively.Conclusions:The MP2RAGE and DKI sequences can evaluate the microstructural changes in globus pallidus,putamen,and thalamus in NBE,which are consistent with previous theories.The changes of T1-relaxation time ratio of GP and putamen is associated with serum bilirubin level.The quantitative T1 relaxation time maps obtained with MP2RAGE and DKI parameter map can provide quantitative information,thus providing a theoretical basis for diagnosis of bilirubin encephalopathy.