Alterations Of Cerebral Morphology In Patients With Early Parkinson’s Disease And The Value Of Machine Learning In Predicting The Freezing Of Gait

Part Ⅰ: Study on cerebral morphology in patients with early Parkinson’s disease based on MRIObjective: The diagnosis of Parkinson’s disease(PD)mainly depends on clinical symptoms.Structural magnetic resonance imaging(MRI)can identify the morphological changes of brain structure and find the damaged brain structure related to PD,which provides objective evidence for the early diagnosis of PD.The purpose of this study was to investigate the alterations of cerebral morphology in patients with early drug na?ve PD and its relationship with clinical and laboratory examination and to explain its neural mechanism.Methods: The cortical thickness,cortical surface area and cortical curvature were measured by CIVET-pipeline,and gray matter volume and white matter volume were calculated by VBM method in 158 patients with early drug na?ve PD and 73 healthy controls(HC)from the open database of Parkinson’s disease progression markers program(PPMI).Sex,age and center were taken as covariables,and the differences of cerebral morphological indexes between the two groups were analyzed by double-sample t-test.The correlation between the above-mentioned indexes of brain regions with statistical differences and clinical and laboratory indexes were analyzed.Results: In patients with early drug na?ve PD,the cortical thickness and cortical surface were increased,while the volume of gray matter decreased,which mainly distributed in the frontaltemporal-parietal lobe and limbic system.Specifically,the thickness of right dorsolateral superior frontal gyrus,right middle frontal gyrus,right inferior frontal gyrus,orbital part,left insula and left posterior cingulate gyrus cortex increased in PD patients(P < 0.01,uncorrected),the cortical surface of right inferior frontal gyrus,orbital part,right medial superior frontal gyrus,left posterior cingulate gyrus,right precuneus,right temporal pole:superior temporal gyrus,right middle temporal gyrus(P < 0.01,uncorrected)and bilateral olfactory cortex(P < 0.05,FDR correction)increased in PD patients,while the gray matter volume of the left inferior frontal gyrus,orbital part and left temporal pole: superior temporal gyrus decreased(P < 0.01,uncorrected).The damaged brain regions mentioned above were related to cognitive impairment,olfactory disorders,anxiety,depression,motor disorders,cerebrospinal fluid protein and serum urate abnormalities.Conclusion: We found that the cortical morphological alterations of frontal lobe and limbic system were the most significant in early drug na?ve PD patients,and were related to cognitive impairment,olfactory disorders,anxiety,depression and motor disorders,cerebrospinal fluid protein and serum urate abnormalities.The results of this study were expected to provide imaging markers for early diagnosis of PD patients.Part Ⅱ: The value of machine learning in predicting the freezing of gait in Parkinson’s disease with baseline structural morphologyObjective: Freezing of gait(FOG)is a common disabled gait disorder in patients with advanced PD.It is of great significance to effectively predict the occurrence of FOG in the early stage of PD.We aim to construct a FOG prediction model based on baseline cerebral structural MRI morphology,clinical and laboratory information of patients with PD.Methods: Ninety-two patients without FOG and 66 patients with FOG with PD were included during the follow-up.The cerebral morphological features calculated in the first part were incorporated into model 1,while adding clinical,cerebrospinal fluid and blood examination information to model 2.Feature selection was carried out through elastic network,and 10-folder cross-validation were used to predict FOG.Results: Thirty-three features were contributed to model one with 72.63% of prediction accuracy,and 44 features were contributed to model two with 77.47% of prediction accuracy.The first 10 contribution features to the two models were the structural imaging features,including gray and white matter volume and cortical curvature,where the increase volume of gray matter in the left lingual gyrus,the left anterior cingulate and paracingulate gyri,and the left angular gyrus played a key role in predicting the FOG.Conclusion: We believed that cerebral morphological imaging information could effectively predict the occurrence of FOG in early drug na?ve PD patients,which provides biomarkers and a new disease diagnosis and treatment model for clinical management and early nursing care of FOG in patients with PD.