Three-dimensional computed tomography dose optimisation and image quality improvement of abdomen-pelvis using adaptive-iterative dose reduction

Abstract

New development of computed tomography (CT) technology has made CT a versatile and efficient diagnostic modality. This has led to exponentially increased demand in clinical practice with increased risk of radiation exposure to the patients. Most research on CT optimisation concentrates on physical parameters, such as tube potential, tube current and pitch factor. Little research has been done on iterative reconstruction process in dose reduction without compromising image quality in clinical CT examination. Thus, this study investigates dose optimisation and image quality improvement using Adaptive-Iterative Dose Reduction Three-Dimensional (AIDR 3D) reconstruction, compared with conventional filtered back projection (FBP), in abdomen-pelvis CT. In a single-centre cohort study, 100 patients who underwent plain CT abdomen-pelvis using a 80-multidetector CT system were retrospectively analysed. Patients were divided into three groups according to the scanning protocol. Group 1 patients (n = 39) were scanned with 120 kVp standard dose FBP reconstruction. Iterative reconstruction was used for 120 kVp low dose group 2 (AIDR 3D standard, n = 28) and 100 kVp low dose group 3 (AIDR 3D strong, n = 33). Quantitative measures of radiation dose, objective image noise and signal to noise ratio (SNR) were obtained. The results were compared between all groups and correlated to body mass index (BMI). Subjective image quality evaluations were graded by two radiologists. The volume CT dose index (CTDIvol), dose length product (DLP), and effective dose (E) in low dose AIDR 3D studies (group 2 and group 3) were significantly lower than standard dose FBP CT (p < 0.05). Group 3 (100 kVp low dose AIDR 3D strong) obtained highest dose reduction with CTDIvol, DLP and E as low as 3.35 ± 1.04 mGy, 172.05 ± 63.32 mGy.cm and 2.58 ± 0.95 mSv respectively. In objective image quality analysis, group 2 and group 3 achieved significant image noise reduction (41.33% versus 52.62%) and SNR increment (62.25% versus 101.47%) compared to group 1. Subjective image noise, artifacts, sharpness and overall diagnostic confidence were greatly improved by AIDR 3D (group 2 and group 3). Moreover, AIDR 3D strong (group 3) was the most optimal iterative reconstruction to demonstrate fine anatomical structures. AIDR 3D could advance dose optimisation and improved image quality for wide range of BMI in the population. Thus, AIDR 3D is a useful algorithm to optimise scanning protocol and practicable in all routine CT examinations at the lowest radiation exposure.