IC‐P‐120: Association Between Brain MRI Diffusion Alterations and CSF Biomarkers in Amnestic MCI

Abstract

Background: The cause of cognitive impairment in acutely hospitalized geriatric patients is often unclear. The diagnostic process is challenging but important in order to effectively treat potentially life threatening etiologies or identify underlying neurodegenerative disease. We intended to evaluate the add-on diagnostic value of structural and metabolic neuroimaging in newly manifested cognitive impairment in elderly geriatric patients hospitalized due to acute or subacute admission indications (WHO Trials Registry DRKS00005041). Methods: Eighty-one inpatients (55 females, 86.1 6 5.5 years) without history of cognitive complaints prior to hospitalization were recruited in 10 different acute geriatric centers. Primary inclusion criterion was a clinical hypothesis of Alzheimer's disease (AD), cerebrovascular disease (CVD) or mixed AD+CVD etiology (MD) which remained uncertain after standard diagnostic workup. Additional diagnostic testing performed after enrollment into the study included detailed neuropsychological testing, structural magnetic resonance imaging (MRI) for characterization of cerebrovascular pathology and brain atrophy as well as positron emission tomography (PET) with the glucose analog 2-[F-18]-fluoro-2-deoxy-Dglucose (FDG) for the detection of a regional metabolic pattern of synaptic dysfunction indicative of AD. An interdisciplinary academic expert team established a probable etiological diagnosis (non-neurodegenerative, AD, CVD, MD) integrating all available data from diagnostic standard procedures and additional tests. Fully automatic multimodal classification based on a Random Undersampling Boosting (RUSBoost) approach was used for rater-independent assessment of the complementary contribution of the additional diagnostic procedures to the etiologic diagnosis. Results: Automatic classification based on all diagnostic standard procedures combined reproduced the etiologic expert diagnosis in 31.4 % of the patients (p ¼ 0.100, chance level is 25 %). Automatic classification based on MRI (cerebrovascular scores) or FDG PET alone provided an accuracy of about 45 % (both p ¼ 0.001). A statistically significant improvement to 76.4 % accuracy was achieved by the integration of all modalities (p ¼ 0.001). Conclusions: These findings indicate a substantial improvement of diagnostic accuracy in the assessment of uncertain de novo cognitive impairment in acutely hospitalized geriatric patients with the integration of structural MRI and brain FDG PET into the diagnostic process.Background: Subjects with mild cognitive impairment (MCI) and abnormal levels of CSF Ab42, Ab42/t-tau and/or Ab42/p-tau ratios are at higher risk of cognitive decline. Diffusion tensor imaging (DTI) is a quantitative MRI technique that enables the in vivo identification of white matter (WM) tissue alterations and can help in identifying early changes in MCI. The goal of this study was to characterize microstructural features of amnestic MCI patients with three different risk of cognitive decline, defined accordingly to their bas...

Identifying Early Mild Cognitive Impairment by Multi-Modality MRI-Based Deep Learning

“…In summary, multi-modality fusion diagnosis using the CNN is an effective medical image analysis method, and good classification performance can be achieved by selecting the most suitable neuroimaging modality according to the pathological characteristics of disease. Although it has been proven that DTI data is an effective imaging-biomarker for MCI diagnosis (Nowrangi et al, 2013;Marizzoni et al, 2016;Brueggen et al, 2017;Gyula, 2018), there are few studies to report. The high classification accuracy obtained in this paper again proves that DTI image can act as a remarkable biomarker for EMCI from the point of view of clinical diagnosis.…”

“…DTI data has been proven to be a useful diagnostic marker for distinguishing EMCI from NC, especially its measures, namely fractional anisotropy (FA) and mean diffusivity (MD). Moreover, DTI data can reflect brain microstructure changes by quantifying the integrity of white matter (Nowrangi et al, 2013;Marizzoni et al, 2016;Brueggen et al, 2017;Gyula, 2018), which makes it promising for identifying the subtle differences of EMCI compared with NC.…”