The congruency of neuropsychological and F18-FDG brain PET/CT diagnostics of Alzheimer’s Disease (AD) in routine clinical practice: insights from a mixed neurological patient cohort

Abstract

Background Diagnostics of Alzheimer’s Disease (AD) require a multimodal approach. Neuropsychologists examine the degree and etiology of dementia syndromes and results are combined with those of cerebrospinal fluid markers and imaging data. In the diagnostic process, neuropsychologists often rely on anamnestic and clinical information, as well as cognitive tests, prior to the availability of exhaustive etiological information. The congruency of this phenomenological approach with results from FDG-PET/CT examinations remains to be explored. The latter yield highly accurate diagnostic information. Method A mixed sample of N = 127 hospitalized neurological patients suspected of displaying a dementia syndrome underwent extensive neuropsychological and FDG-PET/CT examinations. Neuropsychological examinations included an anamnestic and clinical interview, and the CERAD cognitive test battery. Two decisional approaches were considered: First, routine diagnostic results were obtained, i.e. the final clinical decision of the examining neuropsychologist (ADClinical vs. non-ADClinical). Secondly, a logistic regression model was implemented, relying on CERAD profiles alone. CERAD subscales that best predicted AD based on FDG-PET/CT were identified and a nominal categorization obtained (ADTest vs. non-ADTest). Congruency of results from both approaches with those of the FDG-PET/CT (ADPET vs. non-ADPET) were estimated with Cohen’s Kappa (κ) and Yule’s Y coefficient of colligation. Descriptive estimates of accuracy, sensitivity and specificity of CERAD relative to FDG-PET/CT diagnostics were derived. Results ADPET patients constituted N = 33/127 (26%) of the sample. The clinical decision approach (ADClinical vs. non-ADClinical) showed substantial agreement with the FDG-PET/CT classification (κ = .69, Y = .72) involving good accuracy (84.2%), moderate sensitivity (75.8%) and excellent specificity (92.6%). In contrast, the decisional approach that relied on CERAD data alone (ADTest vs. non-ADTest) involved only moderate agreement with the FDG-PET/CT (κ = .54, Y = .62) with lower accuracy (74.8%), attributable to decreased sensitivity (56.3%) and comparable specificity (93.3%). Conclusions It is feasible to identify AD through a comprehensive neuropsychological examination in a mixed sample of neurological patients. However, within the boundaries of methods applied here, decisions based on cognitive test results alone appear limited. One may conclude that the clinical impression based on anamnestic and clinical information obtained by the neuropsychological examiner plays a crucial role in the identification of AD patients in routine clinical practice.

Multimodal learning of clinically accessible tests to aid diagnosis of neurodegenerative disorders: a scoping review

“…When multimodal imaging data was used in the included studies, the workflow of the model usually consisted of feature extraction, feature selection, feature fusion and using multi-source discriminative features for classification [ 53 ]. Convolutional neural network (CNN) was the most widely used technique for feature extraction [ 17 , 42 , 43 , 53 , 57 , 68 , 71 ]. After extraction of biomedical image features, feature selection was used to explore deep common features among different image features and gain information sharing among multiple modal data [ 53 ].…”