A systematic approach towards the objective evaluation of low-contrast performance in MDCT: Combination of a full-reference image fidelity metric and a software phantom

Structural similarity index family for image quality assessment in radiological images

The structural similarity index (SSIM) family is a set of metrics that has demonstrated good agreement with human observers in tasks using reference images. These metrics analyze the viewing distance, edge information between the reference and the test images, changed and preserved edges, textures, and structural similarity of the images. Eight metrics based on that family are proposed. This new set of metrics, together with another eight well-known SSIM family metrics, was tested to predict human performance in some specific tasks closely related to the evaluation of radiological medical images. We used a database of radiological images, comprising different acquisition techniques (MRI and plain films). This database was distorted with different types of distortions (Gaussian blur, noise, etc.) and different levels of degradation. These images were analyzed by a board of radiologists with a double-stimulus methodology, and their results were compared with those obtained from the 16 metrics analyzed and proposed in this research. Our experimental results showed that the readings of human observers were sensitive to the changes and preservation of the edge information between the reference and test images, changes and preservation in the texture, structural component of the images, and simulation of multiple viewing distances. These results showed that several metrics that apply this multifactorial approach (4-G-SSIM, 4-MS-G-SSIM, [Formula: see text], and [Formula: see text]) can be used as good surrogates of a radiologist to analyze the medical quality of an image in an environment with a reference image.

Virtual single-source computed tomography using dual-source acquisition: a new technique for the dose-neutral intraindividual comparison of different scan protocols

OBJECTIVES

The objective of this study was to compare the image quality of a standard single-source (SSS) computed tomography (CT) with that of a virtual single-source CT (VSS-CT) data set reconstructed from 2 raw data sets obtained by dual-source CT acquisition in abdominal CT to establish a radiation dose-neutral approach for the intraindividual comparison of 3 acquisition protocols at different radiation dose levels (RDLs).

MATERIALS AND METHODS

An abdominal phantom representing an 80-kg male was imaged using dual-source CT (SOMATOM Definition; Siemens Healthcare) at 3 RDLs with 120 kV(p) and different tube currents (low, standard, and high milliampere-second protocol). For each RDL, raw data were obtained once in single-source mode using x-ray tube A only and 5 times in dual-source mode using different ratios for tube current of x-ray tubes A and B (same total radiation dose; A/B: 90%/10%, 80%/20%, 70%/30%, 60%/40%, 50%/50%). For each RDL, SSS-CT and 5 virtual single-source image data sets (VSS-CT50 - 90) were reconstructed. To compare SSS-CT and VSS-CT data sets, image quality was assessed in terms of high- and low-contrast performance by calculating the modulation transfer function, image noise, noise power spectrum, and, for low contrast lesion detectability, the modified multiscale structural similarity index (MS-SSIM*). A maximum decrease of Δ = 5% of image quality compared with SSS-CT was defined as acceptable, and a noninferiority analysis with Δ was performed.

RESULTS

For modulation transfer function, noninferiority was observed for all VSS-CT data sets and RDL (P < 0.05). Image noise demonstrated an acceptable increase (<3.2%, P < 0.05) for each RDL and noise power spectrum showed only minor differences in the midfrequency range. The MS-SSIM* index demonstrated for the high RDL protocol a minor decrease for VSS-CT data sets (<2%, P < 0.05). For the standard and low RDL, the relative differences of the MS-SSIM* index increased and were only in 1 case above Δ (standard RDL, mean VSS-CT80 5.1%, P > 0.05).

CONCLUSIONS

The image quality obtained by virtual and SSS reconstruction using equivalent total radiation exposure to the patient showed only negligible differences in image quality. Therefore, this technique might allow an intraindividual comparison of full and reduced radiation dose protocols within 1 image acquisition step by simply splitting the radiation dose between the 2 x-ray tubes of a dual-source CT.

Effect of reduced radiation exposure and iterative reconstruction on detection of low-contrast low-attenuation lesions in an anthropomorphic liver phantom: an 18-reader study

PURPOSE

To measure the effect of reduced radiation exposure on low-contrast low-attenuation liver lesion detection in an anthropomorphic abdominal phantom by using filtered back projection (FBP) and sinogram-affirmed iterative reconstruction.

MATERIALS AND METHODS

Eighteen radiologists blinded to phantom and study design interpreted randomized image data sets that contained 36 spherical simulated liver lesions of three sizes and three attenuation differences (5-mm diameter: 12, 18, and 24 HU less than the 90-HU background attenuation of the simulated liver insert; 10- and 15-mm diameter: 6, 12, and 18 HU less than the 90-HU background attenuation) scanned with four discrete exposure settings and reconstructed by using FBP and sinogram-affirmed iterative reconstruction. Response assessment included region-level lesion presence or absence on a five-point diagnostic confidence scale. Statistical evaluation included multireader multicase receiver operating characteristic curve analysis, with nonparametric methods and noninferiority analysis at a margin of -0.10.

RESULTS

Pooled accuracy at 75% exposure for both FBP and sinogram-affirmed iterative reconstruction was noninferior to 100% exposure (P = .002 and P < .001, respectively). Subsequent exposure reductions resulted in a significant decrease in accuracy. When the smallest (5-mm-diameter) lesions were excluded from analysis, sinogram-affirmed iterative reconstruction was superior to FBP at 100% exposure (P = .011), and sinogram-affirmed iterative reconstruction at 25% and 50% exposure reduction was noninferior to FBP at 100% exposure (P ≤ .013). Reader confidence was greater with sinogram-affirmed iterative reconstruction than with FBP for 10- and 15-mm lesions (2.94 vs 2.76 and 3.62 vs 3.52, respectively).

CONCLUSION

In this low-contrast low-attenuation liver lesion model, a 25% exposure reduction maintained noninferior diagnostic accuracy. However, detection was inferior with each subsequent exposure reduction, regardless of reconstruction method. Sinogram-affirmed iterative reconstruction and FBP performed equally well at modest exposure reduction (25%-50%). Readers had higher confidence levels with sinogram-affirmed iterative reconstruction for the 10- and 15-mm lesions.

Development and evaluation of a software tool for the generation of virtual liver lesions in multidetector-row CT datasets

RATIONALE AND OBJECTIVES

Development and evaluation of a software tool for the insertion of simulated hypodense liver lesions in multidetector-row computed tomography (CT) datasets.

MATERIALS AND METHODS

Forty software-generated hypodense liver lesions were inserted at random locations in 20 CT datasets by using the "alpha blending" technique and compared with 40 real metastatic lesions. The location, diameter (5-20 mm) and density of the simulated lesions were individually adjusted to closely resemble real lesions in each patient. Three blinded readers evaluated all 80 lesions twice in a 2-week interval using a five-point Likert confidence scale under standardized conditions. Nonparametric tests were used to statistically evaluate possible differences in scoring between real and simulated lesions. The correctness of the observer rating for real and simulated lesions was compared to chance distribution using the chi-squared statistics. The inter- and intraobserver variability was determined using Kendall's coefficient of concordance.

RESULTS

The observer study did not reveal significant differences between the scoring for real versus simulated lesions for any of the readers (P > .05). The distribution of correct and false scoring of the lesions was not significantly different from chance distribution (P > .05). Inter- and intraobserver agreement was poor (Kendall W coefficient = 0.12/0.13).

CONCLUSION

The proposed algorithm is suitable for creating realistic virtual liver lesions in CT datasets.

Influence of sinogram affirmed iterative reconstruction of CT data on image noise characteristics and low-contrast detectability: an objective approach

Objectives

To utilize a novel objective approach combining a software phantom and an image quality metric to systematically evaluate the influence of sinogram affirmed iterative reconstruction (SAFIRE) of multidetector computed tomography (MDCT) data on image noise characteristics and low-contrast detectability (LCD).

Materials and Methods

A low-contrast and a high-contrast phantom were examined on a 128-slice scanner at different dose levels. The datasets were reconstructed using filtered back projection (FBP) and SAFIRE and virtual low-contrast lesions (-20HU) were inserted. LCD was evaluated using the multiscale structural similarity index (MS-SIM*). Image noise texture and spatial resolution were objectively evaluated.

Results

The use of SAFIRE led to an improvement of LCD for all dose levels and lesions sizes. The relative improvement of LCD was inversely related to the dose level, declining from 208%(±37%), 259%(±30%) and 309%(±35%) at 25mAs to 106%(±6%), 119%(±9%) and 123%(±8%) at 200mAs for SAFIRE filter strengths of 1, 3 and 5 (p<0.05). SAFIRE reached at least the LCD of FBP at a relative dose of 50%. There was no statistically significant difference in spatial resolution. The use of SAFIRE led to coarser image noise granularity.

Conclusion

A novel objective approach combining a software phantom and the MS-SSIM* image quality metric was used to analyze the detectability of virtual low-contrast lesions against the background of image noise as created using SAFIRE in comparison to filtered back-projection. We found, that image noise characteristics using SAFIRE at 50% dose were comparable to the use of FBP at 100% dose with respect to lesion detectability. The unfamiliar imaging appearance of iteratively reconstructed datasets may in part be explained by a different, coarser noise characteristic as demonstrated by a granulometric analysis.