Lesion detection performance : comparative analysis of low-dose CT data of the chest on two hybrid imaging systems
Purpose: Incidental findings, discovered in low-dose CT images obtained during hybrid imaging, are an increasing phenomenon with advancing CT technology. Understanding their diagnostic value along with technical limitations is therefore important when reporting images and recommending follow-up, which may result in additional radiation dose from further diagnostic imaging and an increase in patient anxiety. This study assesses lesion detection in CT images obtained during attenuation correction (AC) acquisitions on two SPECT/CT systems. Methods: An anthropomorphic chest phantom, containing simulated lesions of varying size and density, was imaged on a GE Infinia Hawkeye 4 and a Siemens Symbia T6 with low-dose CT settings used during AC acquisitions in myocardial perfusion imaging. Twenty-two readers completed a lesion detection task, assessing 46 images (15 normal, 31 abnormal containing 41 lesions) from each SPECT/CT system. Data was evaluated using a jackknife alternative free-response receiver operating characteristic (JAFROC) analysis. Results: JAFROC analysis showed a significant difference (p<0.0001) in lesion detection with figures of merit 0.599 (95% CI 0.568, 0.631) and 0.810 (95% CI 0.781, 0.839) for GE Infinia Hawkeye 4 and Siemens Symbia T6 respectively. Lesion detection on the Infinia Hawkeye 4 was generally limited to larger, higher density lesions. The Siemens Symbia T6 images allowed improved detection rates with mid-sized lesions and some lower density lesions. However, readers struggled to detect small (5mm) lesions on both image sets, irrespective of density. Conclusions: Lesion detection is more reliable in low-dose CT images from the Symbia T6 than those from the Infinia Hawkeye4. This phantom based study gives an indication of potential lesion detection in the clinical context as shown by two commonly used SPECT/CT systems, which may assist the clinician in determining if further diagnostic imaging is justified
Thompson, John D.
De groot, Martin