REGULAR CONTENT
Final ID
452
Type
Original Scientific Research-Oral or Pos
Authors
M Forsberg1, D Eschelman1, J Dave1
Institutions
1Thomas Jefferson University, Philadelphia, PA
Purpose
The purpose of this study was to compare vendor specific fluoroscopic image processing using quantitative metrics derived from clinical images.
Materials & Methods
Institutional Review Board approved this retrospective study. Images from 30 patients undergoing repeat hepatic embolization procedures (with identical contrast injections rates and procedural set ups) in different interventional fluoroscopic suites (vendors: Philips AlluraClarity and Siemens ArtisQ; total 60 cases) were used. Circular ROIs (2 cm diameter) were used to measure signal (mean) and noise (standard deviation) in bone and liver parenchyma in fluoroscopic images. Signal and noise values were also obtained using a circular ROI within the hepatic vessels (till 4 bifurcations) in both fluoroscopic images and DSA acquisitions. Using these measurements contrast to noise ratio (CNR) for low-contrast detectability and contrast to tissue ratio (CTR) for contrast visibility were calculated. Full width at half maximum (FWHM) was computed from a line profile across the vessel wall as a measure of vessel-edge sharpness. Tissue landmarks in the images were used to ensure repeatable measurements. Paired t-tests and ANOVA were used to compare CNR, CTR and FWHM for the images from both vendors and CTR after vessel bifurcation.
Results
The CNR, CTR and FWHM values from the images obtained with different vendors showed a statistically significant difference (p<0.05). The average rib CNR was 6.4±3.0 and 14.2±11.2, and the average spine CNR was 6.9±2.4 and 20.6±12.8 for images from Philips and Siemens suites, respectively. On average, CTR values decreased from the main vessel to the fourth bifurcation by 30% and 27% for images from the Philips and Siemens suites, respectively. The ratio of CTR measurements between bifurcations in both fluoroscopic images and DSA acquisitions were not significantly different (p>0.05). The FWHM was 0.7±0.2 and 1.4±0.4 for images from the Philips and Siemens suites, respectively.
Conclusions
In conclusion, the use of quantitative metrics derived from clinical images for image quality comparisons has been shown. Such measurements allow comparison of vendor-specific fluoroscopic image quality enhancement algorithms.
Final ID
452
Type
Original Scientific Research-Oral or Pos
Authors
M Forsberg1, D Eschelman1, J Dave1
Institutions
1Thomas Jefferson University, Philadelphia, PA
Purpose
The purpose of this study was to compare vendor specific fluoroscopic image processing using quantitative metrics derived from clinical images.
Materials & Methods
Institutional Review Board approved this retrospective study. Images from 30 patients undergoing repeat hepatic embolization procedures (with identical contrast injections rates and procedural set ups) in different interventional fluoroscopic suites (vendors: Philips AlluraClarity and Siemens ArtisQ; total 60 cases) were used. Circular ROIs (2 cm diameter) were used to measure signal (mean) and noise (standard deviation) in bone and liver parenchyma in fluoroscopic images. Signal and noise values were also obtained using a circular ROI within the hepatic vessels (till 4 bifurcations) in both fluoroscopic images and DSA acquisitions. Using these measurements contrast to noise ratio (CNR) for low-contrast detectability and contrast to tissue ratio (CTR) for contrast visibility were calculated. Full width at half maximum (FWHM) was computed from a line profile across the vessel wall as a measure of vessel-edge sharpness. Tissue landmarks in the images were used to ensure repeatable measurements. Paired t-tests and ANOVA were used to compare CNR, CTR and FWHM for the images from both vendors and CTR after vessel bifurcation.
Results
The CNR, CTR and FWHM values from the images obtained with different vendors showed a statistically significant difference (p<0.05). The average rib CNR was 6.4±3.0 and 14.2±11.2, and the average spine CNR was 6.9±2.4 and 20.6±12.8 for images from Philips and Siemens suites, respectively. On average, CTR values decreased from the main vessel to the fourth bifurcation by 30% and 27% for images from the Philips and Siemens suites, respectively. The ratio of CTR measurements between bifurcations in both fluoroscopic images and DSA acquisitions were not significantly different (p>0.05). The FWHM was 0.7±0.2 and 1.4±0.4 for images from the Philips and Siemens suites, respectively.
Conclusions
In conclusion, the use of quantitative metrics derived from clinical images for image quality comparisons has been shown. Such measurements allow comparison of vendor-specific fluoroscopic image quality enhancement algorithms.