Premium
P3‐210: T2 AND PROTON DENSITY SIGNAL‐ AND TEXTURE‐RELATED FEATURES FOR THE PREDICTION OF MCI TO ALZHEIMER'S DISEASE PROGRESSION
Author(s) -
MartínezTorteya Antonio,
Treviño Víctor,
TamezPeña José
Publication year - 2014
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2014.05.1301
Subject(s) - feature selection , biomarker , multivariate statistics , artificial intelligence , segmentation , feature (linguistics) , population , pattern recognition (psychology) , disease , medicine , computer science , machine learning , biology , biochemistry , linguistics , philosophy , environmental health
cognitive complaints who had received both HMPAO-SPECT and DaTscan imaging in a dementia diagnostic centre (26 DaTscan positive, 40 DaTscan negative) and 31 age-matched controls were analysed. Regional differences in brain perfusion in DaTscan positive compared with DaTscan negative patients were identified. Measures of perfusion were calculated from HMPAO-SPECT images for frontal, temporal, parietal and occipital lobe regions of interest using SPM8. An experienced rater identified whether DaTscan images were normal or abnormal, independent of HMPAO-SPECT results. Discriminant analysis was used to classify DaTscan status based on regional perfusion values. The presence or absence of occipital hypoperfusion (regional value <-2 S.D. from 31 controls) was cross-tabulated with the DaTscan result. Results: An SPM analysis of HMPAO-SPECT images identified a local minimum for reduced perfusion in the left superior occipital lobe in DaTscan negative patients (p<0.01). There was significantly reduced perfusion in occipital areas in individuals with abnormal DaTscan compared to other cortical regions (Occipital-L: p<0.01, -R: p<0.05; Parietal-R: p<0.05; Temporal: not sig; Frontal: not sig). Discriminant analysis correctly classified 68.2% of DaTscan images based on the degree of left occipital lobe hypoperfusion (Wilks l1⁄40.881, P<0.01). The specificity for presumed LBD or Parkinsonian disorder with occipital hypoperfusion was 73%, but the sensitivity was low (50%). The positive predictive value was relatively high at 69%. Conclusions: Occipital hypoperfusion is prominent in individuals with abnormal DaTscans who have cognitive complaints, but although it is a relatively specific finding with reasonable positive predictive value, the sensitivity is low. A positive finding should prompt clinical review of features typical for LBD.