Initial performance evaluation of iterative model reconstruction in abdominal computed tomography

• Published in: Journal of computer assisted tomography Vol. 38; no. 3; pp. 408 - 414
• Main Authors: Suzuki, Shigeru, Haruyama, Takahiro, Morita, Hisashi, Takahashi, Yuzuru, Matsumoto, Reiko
• Format: Journal Article
• Language: English
• Published: United States 01-05-2014
• Subjects: Adult; Aged, 80 and over; Algorithms; Computer Simulation; Female; Hepatitis, Chronic - diagnostic imaging; Humans; Liver Cirrhosis - diagnostic imaging; Male; Middle Aged; Models, Biological; Phantoms, Imaging; Pilot Projects; Radiographic Image Enhancement - methods; Radiographic Image Interpretation, Computer-Assisted - methods; Radiography, Abdominal - instrumentation; Radiography, Abdominal - methods; Reproducibility of Results; Sensitivity and Specificity; Tomography, X-Ray Computed - instrumentation; Tomography, X-Ray Computed - methods
• ISSN: 0363-8715; 1532-3145
• DOI: 10.1097/rct.0000000000000062

The aim of this study was to present initial evaluation of the performance of the iterative model reconstruction algorithm (IMR) in abdominal computed tomography (CT).
Computed tomographic examinations were performed for clinical study of 36 patients and for phantom study. We reconstructed the raw data with 1.0- and 5.0-mm slice thicknesses using filtered back projection (FBP), iDose4, and IMR and evaluated image quality objectively and subjectively.
For almost all subjective characteristics, the image quality was better using IMR than iDose4. Objective image noise was significantly less using IMR than iDose4 (P < 0.0001). The contrast-noise ratio of both slice thicknesses increased in order from FBP to iDose4 to IMR. The spatial resolution of reconstructed images was almost identical using IMR, FBP, and iDose4.
The IMR can significantly improve image noise and low-contrast resolution and maintain edge sharpness in abdominal CT images compared with iDose4 or FBP.