Comparative performance analysis for abdominal phantom ROI detectability according to CT reconstruction algorithm: ADMIRE

Purpose We compared and analyzed the detectability performance pertaining to an abdominal phantom including a region of interest (ROI) according to a computed tomography (CT) reconstruction algorithm. Methods Three types of reconstruction algorithms (FBP, SAFIRE, and ADMIRE) were used to evaluate the detectability performance using the abdominal phantom (phantom size: 25 × 18 × 28 cm3). The vendor default settings for routine multi‐detector computed tomography abdominal scans were used. As the quantitative evaluation method, the contrast‐to‐noise ratio (CNR), difference in coefficient of variation (COV) with the normalization based on the FBP data, and the noise power spectrum (NPS) were measured. Results The characteristic of the ADMIRE‐3 reconstructed image was higher than those of the FBP and SAFIRE‐3 reconstructed images. The CNR values of the SAFIRE and ADMIRE images were much higher than the corresponding values of the FBP images. The difference in COV values for the ADMIRE images was ~1.2 times lower than the corresponding values of the SAFIRE images. Conclusion The comparative analysis of the abdominal phantom low‐contrast resolution differences for each CT exposure parameters showed that ADMIRE demonstrated better results than SAFIRE and FBP in terms of contrast, CNR, COV difference, and 1D NPS. This indicates that ADMIRE can provide a clearer observation even with the same number of contrast objects as compared to SAFIRE and FBP owing to its better contrast resolution in the central part of the contrast hole at low kV.

The clinical field is actively performing CT examinations using the repetitive reconstruction algorithm. To this end, iterative reconstruction algorithms may allow a notable dose reduction as they facilitate a more precise modeling of the acquisition process. [2][3][4][5][6][7] In the advanced modeled iterative reconstruction algorithm (ADMIRE: Siemens Healthineers, Forchheim, Germany), not only are improvements in the statistical modeling applied to the raw projection data, a farther-reaching neighborhood analysis of voxel data in the image domain is performed to attain better preservation of the CT noise texture and artifact suppression. 5 Several recent reports incorporated some of the advantages of the advanced modeled iterative reconstruction algorithm [8][9][10] ; however, it may be necessary to compare the results of the image analysis at the reconstruction algorithm step.
In particular, past studies have shown that using SAFIRE [11][12][13] provides diagnostic quality images and reduced doses compared to FBP scans. Therefore, many clinical CT examinations try to use more advanced iterative reconstruction algorithm, such as ADMIRE.
We compared and analyzed the detectability performance corresponding to an abdominal phantom including a region of interest (ROI) according to the CT reconstruction algorithm. In this study, three types of reconstruction algorithms, namely, the FBP reconstruction method, SAFIRE, and ADMIRE, were used to evaluate the detectability performance.

2.A | Experimental conditions
Three types of reconstruction algorithms (FBP, SARFIRE, and ADMIRE) were used to evaluate the detectability performance using an abdominal phantom (phantom size: 25 × 18 × 28 cm 3 , see the (herein, we used the B40f medium kernel) was selected during the iterative reconstruction, 10 investigations were performed for each case considering 9 radiation exposure parameters (80 kV/51 mAs, 80 kV/153 mAs, 80 kV/511 mAs, 100 kV/24 mAs, 100 kV/72 mAs, 100 kV/242 mAs, 120 kV/17 mAs, 120 kV/44 mAs, and 120 kV/ 148 mAs). The pitch was 0.6 and gantry rotation time was 0.5 s. The volume CT dose index (CTDIvol) reported by the scanner console was recorded in a DICOM dose report file after each scan. The equipment used was the SOMATOM Definition Flash CT device (Siemens Healthineers, Forchheim, Germany), and the MDCT images were reconstructed using a matrix size of 512 × 512 mm and pixel spacing (size) as 0.586 mm, an active adaptive filter, small focus size, and reading per projection (RPP) 1 × 2 z-direction. A detailed description of the test conditions used in the reconstruction is presented in Table 1. We implemented the analysis by setting the region of interest (ROI) in MATLAB R2014a (2014a, the MathWorks Inc, USA.). The ROI size considered for the contrast-to-noise (CNR), coefficient of variation (COV) difference, and noise power spectrum (NPS) were 0.2 cm × 0.2 cm, and 1.5 cm × 1.5 cm, respectively.

2.B | The abdominal phantom
The abdominal phantom consisting of polyurethane, epoxy resin, and additional liver region was used to evaluate the image quality of MDCT. This unique anthropomorphic upper abdomen phantom allows obtaining CT images approximate to clinical data.
The elaborate anatomy of liver organs allows a multi-dimensional approach. Figure 2 Table 1.
To compare the results of abdominal phantom study, image quality was evaluated using Lungman phantom in Fig shows the location of the simulated tumor. More information on the Lungman phantom is found in Table 2.

2.C | Analysis methods
For the quantitative analysis of the reconstructed images in the MDCT, we measured the contrast-to-noise ratio (CNR), coefficient of variation (COV), and noise power spectrum (NPS). The CNR was obtained using Eq. (1), using the ROIs (ROI 1 and ROI 2 in Fig. 1) and the standard deviation from the mean values of the ROIs. The COV is defined as the ratio of the standard deviation, that is, the so-called coefficient of dispersion as follows (Eq. (2)) 14 .
where μ is the arithmetic mean (or its absolute value, x007C; μx007C;) and σ is the standard deviation in the ROI. A small COV indicates better image quality because the COV reflects the noise distribution in an X-ray image.
The NPS is expressed as the distribution of the noise frequency in the image and is defined as in Eq. (3) 15 : NPSðu; vÞ ¼ lim where X and Y indicate a distance in the x-and the y-directions, respectively, σ(x,y) is the difference between the average image sig-      Figure 10 shows the measured CNR difference and COV difference values from the z = 38th slice images indicated by boxes A and B in Fig. 9 for the FBP, SAFIRE, and ADMIRE cases.
For all conditions, the ADMIRE-based reconstructed images demonstrate much better quality than the SAFIRE-and FBP-based reconstructed images.
Difference values from the z = 38th slice images indicated by boxes A and B in Fig. 9 for the FBP, SAFIRE, and ADMIRE cases.
For all conditions, the ADMIRE-based reconstructed images demonstrate much better quality than the SAFIRE-and FBP-based reconstructed images. Figure 11 shows the resultant 1D NPS curves that gradually decrease as the spatial frequency increases from box C in

| CONCLUSION
In this work, we investigated the image performance of ADMIREbased reconstructed images using quantitative evaluation methods, compared to that of FBP-and SAFIRE-based reconstructed images.

CONF LICT OF I NTEREST
There is no conflict of interest for all contents in this paper.