Comparative assessment of three image reconstruction techniques for image quality and radiation dose in patients undergoing abdominopelvic multidetector CT examinations

Objective

To compare image quality and radiation dose of abdominal CT examinations reconstructed with three image reconstruction techniques.

Methods

In this Institutional Review Board-approved study, contrast-enhanced (CE) abdominopelvic CT scans from 23 patients were reconstructed using filtered back projection (FBP), adaptive statistical iterative reconstruction (ASiR) and iterative reconstruction in image space (IRIS) and were reviewed by two blinded readers. Subjective (acceptability, sharpness, noise and artefacts) and objective (noise) measures of image quality were recorded for each image data set. Radiation doses in CT dose index (CTDI) dose–length product were also calculated for each examination type and compared. Imaging parameters were compared using the Wilcoxon signed rank test and a paired t-test.

Results

All 69 CECT examinations were of diagnostic quality and similar for overall acceptability (mean grade for ASiR, 3.9±0.3; p=0.2 for Readers 1 and 2; IRIS, 3.9±0.4, p=0.2; FBP, 3.8±0.9). Objective noise was considerably lower with both iterative techniques (p<0.0001 and 0.0016 for ASiR and IRIS). Recorded mean radiation dose, i.e. CTDI_vol, was 24% and 10% less with ASiR (11.4±3.4 mGy; p<0.001) and IRIS (13.5±3.7 mGy; p=0.06), respectively, than with FBP: 15.0±3.5 mGy.

Conclusion

At the system parameters used in this study, abdominal CT scans reconstructed with ASiR and IRIS provide diagnostic images with reduced image noise and 10–24% lower radiation dose than FBP.

Advances in knowledge

CT images reconstructed with FBP are frequently noisy on lowering the radiation dose. Newer iterative reconstruction techniques have different approaches to produce images with less noise; ASiR and IRIS provide diagnostic abdominal CT images with reduced image noise and radiation dose compared with FBP. This has been documented in this study.CT continues to expand its role as an essential imaging modality [1]. However, with its increasing use, concerns of radiation overexposure have prompted efforts to reduce the cumulative dose to a patient [2]. Recent studies have highlighted increased utilisation of radiological examination and 10-fold increase in medical radiation exposure at the population level [3,4]. Therefore, lowering the CT radiation dose without compromising the image quality is desirable. Several technical approaches have been proposed to accomplish these goals including commonly used tube current modulation and adopting lower peak kilovoltage [5-8]. However, excessive dose reduction has remained difficult in the abdomen and pelvis CT due to increased levels of image noise and artefacts that lower the quality of the CT examination. Moreover, abdominopelvic CT demands higher image quality for confident detection of low-contrast lesions in various viscera [9]. The conventional technique of image reconstruction, filtered back projection (FBP), is an efficient method for image production, but makes several assumptions and therefore requires higher dose for delivering diagnostic quality images [10,11]. To overcome these limitations, iterative reconstruction (IR) techniques have been introduced, which have been shown to render optimal image quality at lower radiation dose [12-20]. Unlike advanced iterative techniques, partial IR approaches such as adaptive statistical iterative reconstruction (ASiR) and iterative reconstruction in image space (IRIS) are computationally less demanding and therefore faster to process images. In essence, both rely on mathematic modelling of the CT raw data to selectively identify image noise and reduce it. The ASiR technique models statistical variations in the distribution of noise from acquired image data and improves the signal-to-noise ratio while preserving image contrast [5,8,12]. Since its introduction, several investigators have confirmed its capabilities to deliver diagnostic quality images at 30–50% lower radiation dose. [8,12]. IRIS, on the other hand, reduces image noise by forming multiple iterations within the image space itself [10-14]. Phantom studies have demonstrated its ability in maintaining transverse and z-axis spatial resolution, as well as CT number accuracy and linearity while reducing image noise [18]. Its capability in preserving diagnostic accuracy and improving image quality at lower tube potential settings has been documented by Schindera et al [18] as has its ability to reduce noise and radiation dose in clinical studies by approximately 35–50% [11-16]. Owing to differences in image reconstruction approach by ASiR and IRIS, we investigated the performance of these two IR methods on image quality and radiation dose in patients undergoing contrast-enhanced (CE) abdominal CT examinations compared with the FBP technique.