Classification of Depression, Cognitive Disorders, and Co-Morbid Depression and Cognitive Disorders with Perfusion SPECT Neuroimaging.

BACKGROUND

Depression and cognitive disorders (CDs) are two common co-morbid afflictions that commonly present with overlapping symptoms.

OBJECTIVE

To evaluate if perfusion neuroimaging with brain SPECT can distinguish persons with depression from those with CDs or both conditions.

METHODS

Inclusion criteria were DSM-IV defined depression or CDs (Alzheimer's disease, vascular dementia, dementia not otherwise specified, and amnestic disorders not otherwise specified) including persons with both (total n = 4,541; 847 CDs, 3,269 depression, 425 with both). Perfusion differences between the groups were calculated using two-sampled t-tests corrected for multiple comparisons. Diagnostic separation was determined with discriminant analysis. Feature selection revealed predictive regions in delineating depression from CDs and comorbid cases.

RESULTS

Persons with CDs had lower cerebral perfusion compared to depression. In co-morbid persons, cerebral hypoperfusion was additive, with regions showing lower regional cerebral blood flow compared to either diagnosis alone. Both baseline and concentration SPECT regions yielded correct classification of 86% and leave one out cross-validation of 83%. AUC analysis for SPECT regions showed 86% accuracy, 80% sensitivity and 75% specificity. Discriminant analysis separated depression and CDs from comorbid cases with correct classification of 90.8% and cross validated accuracy of 88.6%. Area under the curve was 83% with sensitivity of 80% and specificity of 70%. Feature selection identified the most predictive regions in left hippocampus, right insula, cerebellar, and frontal lobe regions.

CONCLUSION

Quantitative perfusion SPECT neuroimaging distinguishes depression from dementia and those with both co-morbidities. Perfusion brain SPECT can be utilized clinically to delineate between these two disorders.