IADR Abstract Archives
Image Quality Optimization of Mandibular Bone Assessment using Multislice CT Apparatus.
Objectives: Contemporary researchers have attempted to balance image quality optimisation in Computed Tomography (CT) and the minimum radiation dose necessary for diagnosis. The aim of this study was to investigate the optimisation of CT image quality in mandibular bone assessment using visual grading analysis (VGA) by applying ordinal regression analysis and comparing the results to default clinical exposures.
Methods: A 16-slice Toshiba Aquilion® CT was used to image a human skull phantom at various exposure settings (80,100, 120 kVp and 50-300 mA range). Dose calculations were based on dose length product. Blind review was undertaken by four qualified observers to grade the visibility of selected anatomical landmarks from a perspective of bone width, bone height, trabecular bone and cortical bone, using a five-point rating scale. An overall image quality score using a three-point rating scale of the image was also recorded. Acceptable image quality was defined by the grading values and ability to demonstrate clinical relevance. Inter-rater reliability was tested with Kappa statistics. An ordinal regression analysis method was applied to assess image quality versus dose.
Results: The minimum dose level required for an ‘acceptable’ image quality, when assessed by observers, was 315.77 mGy.cm (kVp=100/mA=200) with a 95% confidence interval of (178.44, 453.10). This dose was 50% less than the default exposure (717.5 mGy.cm (kVp=120/mA=300)) in the clinical setting for this CT unit. The difference was significant among all location and threshold coefficients (p<0.001).
Conclusions: Ordinal regression analysis could be used to numerically optimise CT image quality. This study has demonstrated that diagnostic relevance could be obtained with up to 50% of dose reduction. Further investigations are indicated.
Methods: A 16-slice Toshiba Aquilion® CT was used to image a human skull phantom at various exposure settings (80,100, 120 kVp and 50-300 mA range). Dose calculations were based on dose length product. Blind review was undertaken by four qualified observers to grade the visibility of selected anatomical landmarks from a perspective of bone width, bone height, trabecular bone and cortical bone, using a five-point rating scale. An overall image quality score using a three-point rating scale of the image was also recorded. Acceptable image quality was defined by the grading values and ability to demonstrate clinical relevance. Inter-rater reliability was tested with Kappa statistics. An ordinal regression analysis method was applied to assess image quality versus dose.
Results: The minimum dose level required for an ‘acceptable’ image quality, when assessed by observers, was 315.77 mGy.cm (kVp=100/mA=200) with a 95% confidence interval of (178.44, 453.10). This dose was 50% less than the default exposure (717.5 mGy.cm (kVp=120/mA=300)) in the clinical setting for this CT unit. The difference was significant among all location and threshold coefficients (p<0.001).
Conclusions: Ordinal regression analysis could be used to numerically optimise CT image quality. This study has demonstrated that diagnostic relevance could be obtained with up to 50% of dose reduction. Further investigations are indicated.