The present study investigated (a) the agreement between computerized tomography (CT) and 3 T magnetic resonance imaging (MRI) in the visual grading of medial temporal atrophy (MTA); and (b) whether MTA on CT differentiated patients with dementia from no dementia in memory clinics.Materials and Methods:
Data were obtained from patients who underwent CT scans at the National University Hospital of Singapore, and from a subsample who subsequently underwent 3 T MRI scans in a research study. Agreements and disagreements between CT and MRI were determined. Area under the curve (AUC) analyses determined if CT-graded MTA distinguished patients with dementia from no dementia.Results:
Of the 107 patients in the subsample, MTA scores of 71 agreed on both CT and MRI. The true positive rate between CT and MRI for MTA scores ≥2 was 79.7%. The true negative rate for MTA scores between 0 and 1 was 96.4%. CT underestimated MTA severity in 33 of 36 disagreements with the MRI. MTA scores ≥2 on CT distinguished dementia from no dementia in both discovery [n=263; AUC (95% confidence interval)=0.77 (0.72-0.83); sensitivity=0.69; specificity=0.74] and validation [n=264; AUC (95% confidence interval)=0.77 (0.71-0.82); sensitivity=0.72; specificity=0.72] groups.Conclusions:
MTA graded on CT is a viable alternative to MRI to aid in the diagnosis of dementia in memory clinics.