Limits...
Correlation of clinical and physical-technical image quality in chest CT: a human cadaver study applied on iterative reconstruction.

De Crop A, Smeets P, Van Hoof T, Vergauwen M, Dewaele T, Van Borsel M, Achten E, Verstraete K, D'Herde K, Thierens H, Bacher K - BMC Med Imaging (2015)

Bottom Line: Potential dose reduction based on clinical image quality varied from 27 to 37.4%, depending on the strength of SAFIRE.Our results demonstrate that noise assessments in a uniform phantom overestimate the potential dose reduction for the SAFIRE IR algorithm.In conclusion, one should be cautious to evaluate the performance of CT equipment taking into account only physical-technical parameters as noise and CNR, as this might give an incomplete representation of the actual clinical image quality performance.

View Article: PubMed Central - PubMed

Affiliation: Department of Basic Medical Sciences, Ghent University, Proeftuinstraat 86, B-9000, Ghent, Belgium. An.decrop@ugent.be.

ABSTRACT

Background: The first aim of this study was to evaluate the correlation between clinical and physical-technical image quality applied to different strengths of iterative reconstruction in chest CT images using Thiel cadaver acquisitions and Catphan images. The second aim was to determine the potential dose reduction of iterative reconstruction compared to conventional filtered back projection based on different clinical and physical-technical image quality parameters.

Methods: Clinical image quality was assessed using three Thiel embalmed human cadavers. A Catphan phantom was used to assess physical-technical image quality parameters such as noise, contrast-detail and contrast-to-noise ratio (CNR). Both Catphan and chest Thiel CT images were acquired on a multislice CT scanner at 120 kVp and 0.9 pitch. Six different refmAs settings were applied (12, 30, 60, 90, 120 and 150refmAs) and each scan was reconstructed using filtered back projection (FBP) and iterative reconstruction (SAFIRE) algorithms (1,3 and 5 strengths) using a sharp kernel, resulting in 24 image series. Four radiologists assessed the clinical image quality, using a visual grading analysis (VGA) technique based on the European Quality Criteria for Chest CT.

Results: Correlation coefficients between clinical and physical-technical image quality varied from 0.88 to 0.92, depending on the selected physical-technical parameter. Depending on the strength of SAFIRE, the potential dose reduction based on noise, CNR and the inverse image quality figure (IQF(inv)) varied from 14.0 to 67.8%, 16.0 to 71.5% and 22.7 to 50.6% respectively. Potential dose reduction based on clinical image quality varied from 27 to 37.4%, depending on the strength of SAFIRE.

Conclusion: Our results demonstrate that noise assessments in a uniform phantom overestimate the potential dose reduction for the SAFIRE IR algorithm. Since the IQF(inv) based dose reduction is quite consistent with the clinical based dose reduction, an optimised contrast-detail phantom could improve the use of contrast-detail analysis for image quality assessment in chest CT imaging. In conclusion, one should be cautious to evaluate the performance of CT equipment taking into account only physical-technical parameters as noise and CNR, as this might give an incomplete representation of the actual clinical image quality performance.

No MeSH data available.


Related in: MedlinePlus

Mean noise, CNR and IQFinv versus mean VGAS. The error bars in the x direction represent the standard deviation between the scores of the different radiologists. For noise, the error bars in the y direction represent the standard deviation between noise measurements in 11 following slices in the Catphan uniformity module. For CNR, the error bars in the y direction represent the standard deviation between CNR measurements in four consecutive slices of the Catphan CT number linearity and CT number accuracy module. For IQFinv, the error bars in the y direction represent the standard deviation between the six readers of the contrast-detail module in the Catphan phantom. Regression lines were plotted resulting in an r2 of 0.90, 0.88 and 0.92, p < 0.001, for noise, CNR and IQFinv respectively
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC4541737&req=5

Fig3: Mean noise, CNR and IQFinv versus mean VGAS. The error bars in the x direction represent the standard deviation between the scores of the different radiologists. For noise, the error bars in the y direction represent the standard deviation between noise measurements in 11 following slices in the Catphan uniformity module. For CNR, the error bars in the y direction represent the standard deviation between CNR measurements in four consecutive slices of the Catphan CT number linearity and CT number accuracy module. For IQFinv, the error bars in the y direction represent the standard deviation between the six readers of the contrast-detail module in the Catphan phantom. Regression lines were plotted resulting in an r2 of 0.90, 0.88 and 0.92, p < 0.001, for noise, CNR and IQFinv respectively

Mentions: To evaluate the correlation between clinical and physical-technical image quality, regression curves were plotted for noise, CNR and IQFinv as a function of VGA scores for the different refmAs settings (Fig. 3). Good correlation was found between noise and VGAS, 0.90, p < 0.001. A correlation coefficient of 0.88, p < 0.001 was obtained for CNR and VGAS. Contrast-detail (IQFinv) and VGAS resulted in a correlation coefficient of 0.92, p < 0.001.Fig. 3


Correlation of clinical and physical-technical image quality in chest CT: a human cadaver study applied on iterative reconstruction.

De Crop A, Smeets P, Van Hoof T, Vergauwen M, Dewaele T, Van Borsel M, Achten E, Verstraete K, D'Herde K, Thierens H, Bacher K - BMC Med Imaging (2015)

Mean noise, CNR and IQFinv versus mean VGAS. The error bars in the x direction represent the standard deviation between the scores of the different radiologists. For noise, the error bars in the y direction represent the standard deviation between noise measurements in 11 following slices in the Catphan uniformity module. For CNR, the error bars in the y direction represent the standard deviation between CNR measurements in four consecutive slices of the Catphan CT number linearity and CT number accuracy module. For IQFinv, the error bars in the y direction represent the standard deviation between the six readers of the contrast-detail module in the Catphan phantom. Regression lines were plotted resulting in an r2 of 0.90, 0.88 and 0.92, p < 0.001, for noise, CNR and IQFinv respectively
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC4541737&req=5

Fig3: Mean noise, CNR and IQFinv versus mean VGAS. The error bars in the x direction represent the standard deviation between the scores of the different radiologists. For noise, the error bars in the y direction represent the standard deviation between noise measurements in 11 following slices in the Catphan uniformity module. For CNR, the error bars in the y direction represent the standard deviation between CNR measurements in four consecutive slices of the Catphan CT number linearity and CT number accuracy module. For IQFinv, the error bars in the y direction represent the standard deviation between the six readers of the contrast-detail module in the Catphan phantom. Regression lines were plotted resulting in an r2 of 0.90, 0.88 and 0.92, p < 0.001, for noise, CNR and IQFinv respectively
Mentions: To evaluate the correlation between clinical and physical-technical image quality, regression curves were plotted for noise, CNR and IQFinv as a function of VGA scores for the different refmAs settings (Fig. 3). Good correlation was found between noise and VGAS, 0.90, p < 0.001. A correlation coefficient of 0.88, p < 0.001 was obtained for CNR and VGAS. Contrast-detail (IQFinv) and VGAS resulted in a correlation coefficient of 0.92, p < 0.001.Fig. 3

Bottom Line: Potential dose reduction based on clinical image quality varied from 27 to 37.4%, depending on the strength of SAFIRE.Our results demonstrate that noise assessments in a uniform phantom overestimate the potential dose reduction for the SAFIRE IR algorithm.In conclusion, one should be cautious to evaluate the performance of CT equipment taking into account only physical-technical parameters as noise and CNR, as this might give an incomplete representation of the actual clinical image quality performance.

View Article: PubMed Central - PubMed

Affiliation: Department of Basic Medical Sciences, Ghent University, Proeftuinstraat 86, B-9000, Ghent, Belgium. An.decrop@ugent.be.

ABSTRACT

Background: The first aim of this study was to evaluate the correlation between clinical and physical-technical image quality applied to different strengths of iterative reconstruction in chest CT images using Thiel cadaver acquisitions and Catphan images. The second aim was to determine the potential dose reduction of iterative reconstruction compared to conventional filtered back projection based on different clinical and physical-technical image quality parameters.

Methods: Clinical image quality was assessed using three Thiel embalmed human cadavers. A Catphan phantom was used to assess physical-technical image quality parameters such as noise, contrast-detail and contrast-to-noise ratio (CNR). Both Catphan and chest Thiel CT images were acquired on a multislice CT scanner at 120 kVp and 0.9 pitch. Six different refmAs settings were applied (12, 30, 60, 90, 120 and 150refmAs) and each scan was reconstructed using filtered back projection (FBP) and iterative reconstruction (SAFIRE) algorithms (1,3 and 5 strengths) using a sharp kernel, resulting in 24 image series. Four radiologists assessed the clinical image quality, using a visual grading analysis (VGA) technique based on the European Quality Criteria for Chest CT.

Results: Correlation coefficients between clinical and physical-technical image quality varied from 0.88 to 0.92, depending on the selected physical-technical parameter. Depending on the strength of SAFIRE, the potential dose reduction based on noise, CNR and the inverse image quality figure (IQF(inv)) varied from 14.0 to 67.8%, 16.0 to 71.5% and 22.7 to 50.6% respectively. Potential dose reduction based on clinical image quality varied from 27 to 37.4%, depending on the strength of SAFIRE.

Conclusion: Our results demonstrate that noise assessments in a uniform phantom overestimate the potential dose reduction for the SAFIRE IR algorithm. Since the IQF(inv) based dose reduction is quite consistent with the clinical based dose reduction, an optimised contrast-detail phantom could improve the use of contrast-detail analysis for image quality assessment in chest CT imaging. In conclusion, one should be cautious to evaluate the performance of CT equipment taking into account only physical-technical parameters as noise and CNR, as this might give an incomplete representation of the actual clinical image quality performance.

No MeSH data available.


Related in: MedlinePlus