Influence of spectral detector CT based monoenergetic images on the computer-aided detection of pulmonary artery embolism

Abstract

Objective: To evaluate the influence of monoenergetic reconstructions using a spectral detector CT (SDCT) on the computer aided detection (CAD) of pulmonary artery embolism (PAE) on CT pulmonary angiography (CTPA) and CT in venous contrast phase (CTV).

Material and methods: A retrospective data base search identified 15 patients with CTPA and 18 patients with CTV and diagnosis of PAE. For these patients, monoenergetic (monoE) images at different energy levels or with a fixed attenuation in the pulmonary artery were generated and independently analyzed using a commercially available computer aided detection (CAD) tool. Attenuation in the pulmonary artery and in the embolus was measured.

Results: For CTPA and CTV, in monoenergetic images the difference in attenuation between vessel and embolus was significantly higher than in conventional images. In CTPA the detection rate was highest in the >500 HU monoE images with 67,9% detected emboli and 93,3% of patients correctly identified as having PAE. At the same time the false positive rate could be significantly reduced by using monoE images compared to conventional images. Detection rates for CTV were lower than in CTPA but were raised significantly by monoE reconstructions.

Conclusions: The combination of SDCT and CAD improves the diagnostic accuracy of CAD and enables CAD interpretation of CTV.

Keywords: Computed tomography; Computer-assisted; Dual energy CT; Image analysis; Pulmonary embolism; Spectral detector CT; Spiral.