Author: Beth Reed, Justin B. Solomon 👨🔬
Affiliation: Duke University, Clinical Imaging Physics Group, Department of Radiology, Duke University Health System 🌍
Purpose: The purpose of this study was to compare image quality between CT images of variable slice thickness reconstructed directly at the scanner verses reformatted at a PACS workstation.
Methods: The American College of Radiology (ACR) CT image quality phantom (Gammex 464) was imaged using the institution’s lung cancer screening protocol on two CT systems (GE Discovery CT750 HD, Siemens SOMATOM Definition Flash). Images were reconstructed at varying slice thicknesses (0.6-7mm). The thinnest axial series from each system were sent to a PACS workstation (Visage 7.0) and reformatted to matching slice thicknesses. Measurements - including noise power spectrum (NPS), contrast, and task transfer function (TTF) - were made using TG233 methodology with a standard publicly available software tool (IMQUEST) and compared between “native” (i.e. from scanner) vs PACS reconstructed images.
Results: Noise from native images was higher compared to PACS images by 21% on average for the GE scanner. Noise was comparable between native and PACS images for the Siemens scanner (< 6% difference on average). Native image noise texture was marginally higher than PACS images (< 11.5% difference on average). Contrast was comparable between native and PACS images (<8% difference on average). Spatial resolution was marginally improved on native images compared to PACS images (< 7% difference on average).
Conclusion: PACS reformatted images demonstrated only marginal difference from native reconstructed images in image quality properties of noise, noise texture, contrast, and spatial resolution.