The Application Of Neural Imaging Tools Of Deep Learning Algorithm In Brain Volume Measurement Of AD-MCI

Objective:By using a deep learning-based quantitative calculation tool for whole brain in neural imaging,to analyze the changes of brain structure in patients with Mild Cognitive Impairment（MCI）caused by Alzheimer’s disease,and to explore the correlation between the changes and neurocognition.Methods:Collected 90 cases of memory impairment patients admitted to the neurology department of Zhongshan Hospital Affiliated to Dalian University from September 2017 to March 2019,and 65 cases were enrolled after exclusion criteria,including 29 males（44.6%）and 36 females（55.4%）.The patients ranged in age from45 to 80 years old,with an average age of 69.20±10.25 years old.There were 60normal controls（NC group）,including 24 males（40.0%）and 36 females（60.0%）,aged from 53 to 80 years old,with an average age of 64.90±6.67 years old.First,the two groups of subjects were evaluated for their clinical cognitive function in a quiet state,using the Montreal cognitive assessment scale（MoCA）.3.0T MR scanning was performed on the subjects within 24 hours after the test,and sagittal thin layer images of T1WI were collected.The image processing used the quantitative calculation tool of neural image to segment the whole brain tissue rapidly and accurately,and obtained the absolute and relative volumes of 64 brain regions.Spss20.0 statistical software was used for statistical analysis and processing.First,clinical data and imaging data of the two groups of subjects were analyzed,and T test of two independent samples was used to obtain the differences in age,MoCA score and absolute volume and relative volume of each brain region,andc^（17）test was used for gender.Pearson correlation analysis was performed between different brain regions and MoCA scores.P<0.05 was considered statistically significant.Results:1.There were statistically significant differences in age and MoCA scores between MCI group and NC group（P<0.05）,MoCA score of MCI group was lower than NC group.Gender was not statistically significant（P>0.05）.2.The absolute volume difference between MCI group and NC group showed that:the total hippocampus,the left hippocampus,the right hippocampus,the left temporal lobe,the right temporal lobe,the left cingulate gyrus,the right cingulate gyrus,the left insula,the right insula and the total volume of gray matter decreased.The Ventricular system,the left ventricle,the right ventricle and the third ventricle were enlarged.The difference was statistically significant（P<0.05）.3.The relative volume difference between MCI group and NC group showed that:the total volume of hippocampus,the left hippocampus,the right hippocampus,the left amygdala,the left temporal lobe,the right temporal lobe,the left cingulate gyrus,the left insula,the right insula and the total volume of gray matter decreased.The Ventricular system,the left ventricle,the right ventricle and the third ventricle were enlarged.The difference was statistically significant（P<0.05）.4.Correlation analysis between absolute volume changes in MCI patients and MoCA scores showed that:Absolute volume of left hippocampus（r=0.258,P=0.038）,right hippocampus（r=0.248,P=0.047）,left temporal lobe（r=0.258,P=0.038）and left cingulate gyrus（r=0.262,P=0.035）were positively correlated with MoCA score.Absolute volume of ventricular system（r=-0.417,P=0.001）,lateral ventricle（r=-0.445,P<0.001）and third ventricle（r=-0.348,P=0.004）was negatively correlated with MoCA score.5.Correlation analysis between relative volume changes in MCI patients and MoCA scores showed that:Relative volume of left hippocampus（r=0.400,P=0.001）,right hippocampus（r=0.353,P=0.004）,left temporal lobe（r=0.311,P=0.012）,left amygdala（r=0.324,P=0.008）,left cingulate gyrus（r=0.275,P=0.026）and total gray matter（r=0.461,P<0.001）were positively correlated with MoCA score.The relative volume of ventricular system（r=-0.479,P<0.001）,lateral ventricle（r=-0.467,P<0.001）and third ventricle（r=-0.369,P=0.002）was negatively correlated with MoCA score.Conclusion:1.The whole brain quantitative calculation tool for neural imaging based on deep learning algorithm can effectively detect the volume difference of brain structure in patients with MCI.2.MCI patients at the early stage of AD have experienced changes in the volume of multiple brain regions,which are associated with clinical manifestations of cognitive dysfunction.