Fast thoracic MRI as an alternative to chest x-ray: A retrospective evaluation of 287 patients

Abstract

PurposeTo compare half-Fourier acquisition single-shot turbo spin-echo thoracic MRI (HASTE-MRI) with chest x-ray (CXR). MethodsA total of 287 patients (median age 53 years) undergoing both axial HASTE-MRI and CXR (<7 days apart) were retrospectively analyzed. Pulmonary nodules and consolidations, pleural effusion, and cardiac or aortic enlargement were considered. Detection rate (DR) and intra- and inter-technique agreement for both HASTE-MRI and CXR, were assessed. ResultsPulmonary nodule DR was 1.7% (5/287) at both techniques (perfect inter-technique agreement); inter-reader agreement was perfect for HASTE-MRI (κ = 1.000) and modest for CXR (κ = 0.393). Pleural effusion DR at HASTE-MRI (19.1%, 55/287) was higher (p < 0.001) than at CXR (13.9%, 40/287), but their agreement was almost perfect (κ = 0.812); inter-reader agreement was almost perfect for HASTE-MRI (κ = 0.902) and substantial for CXR (κ = 0.773). Pulmonary consolidation DR at HASTE-MRI (8.4%, 24/287) was similar (p = 0.625) to that at CXR (7.7%, 22/287), with an almost perfect inter-technique agreement (κ = 0.905); inter-reader agreement was moderate for both HASTE-MRI (κ = 0.639) and CXR (κ = 0.554). Cardiac enlargement DR at HASTE-MRI (45.3%, 130/287) was borderline significantly lower (p = 0.057) than that at CXR (48.1%, 138/287), with an almost perfect inter-technique agreement (κ = 0.902); inter-reader agreement was substantial for both HASTE-MRI (κ = 0.798) and CXR (κ = 0.744). Aortic dilation DR at HASTE-MRI (17.7%, 51/287) was significantly higher (p < 0.001) than that at CXR (9.7%, 28/287), with a substantial inter-technique agreement (κ = 0.648); inter-reader agreement was moderate for both HASTE-MRI (κ = 0.724) and CXR (κ = 0.346). ConclusionHASTE-MRI and CXR showed at least a comparable DR for thoracic cardiac and non-cardiac findings.