Farooq Kamal, Cassandra Morrison, Michael D. Oliver, Mahsa Dadar
{"title":"探索大脑结构磁共振成像和临床测量在预测老年痴呆症神经病理学方面的作用:一种机器学习方法","authors":"Farooq Kamal, Cassandra Morrison, Michael D. Oliver, Mahsa Dadar","doi":"10.1101/2024.02.28.24303519","DOIUrl":null,"url":null,"abstract":"Importance: Vascular and structural brain changes are increasingly recognized for their role in cognitive decline and progression of neurodegenerative conditions including Alzheimer's disease (AD). Despite advances in imaging technologies, the exact contribution of these brain changes to disease processes remains a subject of ongoing research. Objective: To apply machine learning techniques to determine critical features of AD-related neuropathologies in vivo. Main Outcomes and Measures: A total of 127 participants (95 females, mean age=87.3) from the RUSH dataset and 65 participants (17 females, mean age=79.0) from the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset were included. In the RUSH dataset, machine learning models were applied towards feature selection of MRI, clinical, and demographic data to identify the best performing set of variables that could predict neuropathology outcomes (e.g., Braak neurofibrillary tangle stage, neurofibrillary tangle burden; NFT). The best-performing neuropathology predictors using the top seven MRI, clinical, and demographic features were then validated in ADNI to compare results and ensure that the feature selection process did not lead to overfitting. For continuous measures, gradient boosting, bagging, support vector regression, and linear regression were implemented. For binary outcomes, logistic regression, gradient boosting, support vector machine, and bagging classifiers were utilized. Results: Applying feature ranking methods using similar information criteria, four machine learning models consistently ranked white matter hyperintensity (WMHs), gray matter (GM), and white matter (WM) volumes as important features in predicting all neuropathology measures. In the RUSH dataset, prediction accuracy was highest for Braak stage, NFT, and tangles (i.e., cross-validated correlation between actual measures and predictions was above 0.8). The best-performing model achieved r=0.83 (RMSE=0.50) in predicting tangles. The best-performing binary classifier achieved 82% accuracy, 86% sensitivity, and 78% specificity in predicting NIA-Reagan (measure of neurofibrillary tangles and neuritic plaques). Similar results were observed in the ADNI dataset. Conclusion and Relevance: These results highlight the efficacy of machine learning models, particularly when incorporating structural MRI features (e.g., GM, WM) alongside WMHs, in accurately predicting AD neuropathology. The use of machine learning may prove beneficial in early detection of AD pathology.","PeriodicalId":501025,"journal":{"name":"medRxiv - Geriatric Medicine","volume":"12 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-02-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Exploring the power of structural brain MRI and clinical measures in predicting AD neuropathology: a machine learning approach\",\"authors\":\"Farooq Kamal, Cassandra Morrison, Michael D. Oliver, Mahsa Dadar\",\"doi\":\"10.1101/2024.02.28.24303519\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Importance: Vascular and structural brain changes are increasingly recognized for their role in cognitive decline and progression of neurodegenerative conditions including Alzheimer's disease (AD). Despite advances in imaging technologies, the exact contribution of these brain changes to disease processes remains a subject of ongoing research. Objective: To apply machine learning techniques to determine critical features of AD-related neuropathologies in vivo. Main Outcomes and Measures: A total of 127 participants (95 females, mean age=87.3) from the RUSH dataset and 65 participants (17 females, mean age=79.0) from the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset were included. In the RUSH dataset, machine learning models were applied towards feature selection of MRI, clinical, and demographic data to identify the best performing set of variables that could predict neuropathology outcomes (e.g., Braak neurofibrillary tangle stage, neurofibrillary tangle burden; NFT). The best-performing neuropathology predictors using the top seven MRI, clinical, and demographic features were then validated in ADNI to compare results and ensure that the feature selection process did not lead to overfitting. For continuous measures, gradient boosting, bagging, support vector regression, and linear regression were implemented. For binary outcomes, logistic regression, gradient boosting, support vector machine, and bagging classifiers were utilized. Results: Applying feature ranking methods using similar information criteria, four machine learning models consistently ranked white matter hyperintensity (WMHs), gray matter (GM), and white matter (WM) volumes as important features in predicting all neuropathology measures. In the RUSH dataset, prediction accuracy was highest for Braak stage, NFT, and tangles (i.e., cross-validated correlation between actual measures and predictions was above 0.8). The best-performing model achieved r=0.83 (RMSE=0.50) in predicting tangles. The best-performing binary classifier achieved 82% accuracy, 86% sensitivity, and 78% specificity in predicting NIA-Reagan (measure of neurofibrillary tangles and neuritic plaques). Similar results were observed in the ADNI dataset. Conclusion and Relevance: These results highlight the efficacy of machine learning models, particularly when incorporating structural MRI features (e.g., GM, WM) alongside WMHs, in accurately predicting AD neuropathology. The use of machine learning may prove beneficial in early detection of AD pathology.\",\"PeriodicalId\":501025,\"journal\":{\"name\":\"medRxiv - Geriatric Medicine\",\"volume\":\"12 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-02-29\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"medRxiv - Geriatric Medicine\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1101/2024.02.28.24303519\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"medRxiv - Geriatric Medicine","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1101/2024.02.28.24303519","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Exploring the power of structural brain MRI and clinical measures in predicting AD neuropathology: a machine learning approach
Importance: Vascular and structural brain changes are increasingly recognized for their role in cognitive decline and progression of neurodegenerative conditions including Alzheimer's disease (AD). Despite advances in imaging technologies, the exact contribution of these brain changes to disease processes remains a subject of ongoing research. Objective: To apply machine learning techniques to determine critical features of AD-related neuropathologies in vivo. Main Outcomes and Measures: A total of 127 participants (95 females, mean age=87.3) from the RUSH dataset and 65 participants (17 females, mean age=79.0) from the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset were included. In the RUSH dataset, machine learning models were applied towards feature selection of MRI, clinical, and demographic data to identify the best performing set of variables that could predict neuropathology outcomes (e.g., Braak neurofibrillary tangle stage, neurofibrillary tangle burden; NFT). The best-performing neuropathology predictors using the top seven MRI, clinical, and demographic features were then validated in ADNI to compare results and ensure that the feature selection process did not lead to overfitting. For continuous measures, gradient boosting, bagging, support vector regression, and linear regression were implemented. For binary outcomes, logistic regression, gradient boosting, support vector machine, and bagging classifiers were utilized. Results: Applying feature ranking methods using similar information criteria, four machine learning models consistently ranked white matter hyperintensity (WMHs), gray matter (GM), and white matter (WM) volumes as important features in predicting all neuropathology measures. In the RUSH dataset, prediction accuracy was highest for Braak stage, NFT, and tangles (i.e., cross-validated correlation between actual measures and predictions was above 0.8). The best-performing model achieved r=0.83 (RMSE=0.50) in predicting tangles. The best-performing binary classifier achieved 82% accuracy, 86% sensitivity, and 78% specificity in predicting NIA-Reagan (measure of neurofibrillary tangles and neuritic plaques). Similar results were observed in the ADNI dataset. Conclusion and Relevance: These results highlight the efficacy of machine learning models, particularly when incorporating structural MRI features (e.g., GM, WM) alongside WMHs, in accurately predicting AD neuropathology. The use of machine learning may prove beneficial in early detection of AD pathology.