(Saturday, 4/4/2020) [Mountain Time (GMT-6)]
Purpose: To develop an efficient image-driven method to automatically extract scan-related information and images from CT examination for size-specific dose estimation (SSDE).
Methods: The proposed automatic extraction approach for SSDE includes dicom information extraction, image pre-processing, dynamic thresholding, soft-tissue segmentation. Effective diameter and Water Equivalent Diameter (Dw) was determined based on AAPM reports 204 and 220. The dicom information (i.e.,Image Position Patient and Slice Location) were used to determine the scanned range and the mean SSDE over the entire scan range. Validation study was performed on ACR CT accreditation phantom. We retrospectively studied 37 Head CT with dual energy protocols (80kV+Sn150kV) and 41 body CT with single energy protocols (100kV) to test the performance. A variety of imaging parameters including detector configuration, section thickness, automatic exposure control, iterative reconstruction algorithm were used in clinical protocols. For each patient, AP, Lat, effective diameter and Dw at each slice location were measured and calculated from axial CT images automatically. The CTDIvol-to-SSDE conversion factor (f) and SSDE are calculated.
Results: There is good agreement between ACR CT phantom size and AP, Lat, effective diameter and Dw measured by the proposed method (Physical size: 20.0cm vs: AP 20.0±0.3cm, Lat 20.0±0.2cm, Eff.Dia 20.0±0.2cm, Dw 20.0±0.3cm). Significant correlations between patient effective diameter and Dw were found for both head and body CT group. CTDIvol shows an overestimation of the actual dose delivered to patient in head CT (CTDIvol: 36.8±8.4mGy vs SSDE: 34.7±6.7mGy) while CTDIvol shows a significant underestimation of the actual dose delivered to patient in body CT (CTDIvol: 8.0±5.4mGy vs SSDE: 12.8±6.6mGy). The f was 0.95±0.10 for head CT and 1.76±0.29 for body CT, respectively.
Conclusion: This work proposed an efficient image-driven method for automatic head and body SSDE. It provide a promising tool for patient dose calculation, dose optimization and image quality improvement.
ePosters
Contact Email
