Screening for Cognitive Impairment by Model-Assisted Cerebral Blood Flow Estimation

• Published in: IEEE transactions on biomedical engineering Vol. 65; no. 7; pp. 1654 - 1661
• Main Authors: Lassila, Toni, Marco, Luigi Yuri Di, Mitolo, Micaela, Iaia, Vincenzo, Levedianos, Giorgio, Venneri, Annalena, Frangi, Alejandro F.
• Format: Journal Article
• Language: English
• Published: United States IEEE 01-07-2018; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Aging; Alzheimer's disease; Biological system modeling; Biomedical monitoring; Blood flow; Blood pressure; Blood-brain barrier; Cerebral blood flow; Classifiers; Cognitive ability; Dementia disorders; Diagnosis; Impairment; Magnetic resonance imaging; Medical imaging; Model accuracy; Modelling; Neurodegenerative diseases; Neuroimaging; Neurological diseases; physiological modelling; Physiology; Predictive control; Predictive models; Regression analysis; Regression models; Screening; Spin labeling; Ultrasonic imaging; Ultrasonic variables measurement; Ultrasound
• ISSN: 0018-9294; 1558-2531
• DOI: 10.1109/TBME.2017.2759511

Objective: Alzheimer's disease (AD) is a progressive and debilitating neurodegenerative disease; a major health concern in the ageing population with an estimated prevalence of 46 million dementia cases worldwide. Early diagnosis is therefore crucial so mitigating treatments can be initiated at an early stage. Cerebral hypoperfusion has been linked with blood-brain barrier dysfunction in the early stages of AD, and screening for chronic cerebral hypoperfusion in individuals has been proposed for improving the early diagnosis of AD. However, ambulatory measurements of cerebral blood flow are not routinely carried out in the clinical setting. In this study, we combine physiological modeling with Holter blood pressure monitoring and carotid ultrasound imaging to predict 24-h cerebral blood flow (CBF) profiles in individuals. One hundred and three participants [53 with mild cognitive impairment (MCI) and 50 healthy controls] underwent model-assisted prediction of 24-h CBF. Model-predicted CBF and neuropsychological tests were features in lasso regression models for MCI diagnosis. Results: A CBF-enhanced classifier for diagnosing MCI performed better, area-under-the-curve (AUC) = 0.889 (95%-CI: 0.800 to 0.978), than a classifier based only on the neuropsychological test scores, AUC = 0.818 (95%-CI: 0.643 to 0.992). An additional cohort of 25 participants (11 MCI and 14 healthy) was recruited to perform model validation by arterial spin-labeling magnetic resonance imaging, and to establish a link between measured CBF that predicted by the model. Conclusion: Ultrasound imaging and ambulatory blood pressure measurements enhanced with physiological modeling can improve MCI diagnosis accuracy.