CT Simulation Optimization for Prostate Cancer Patients in Radiation Therapy: A General Strategy
H Li1*, L Yu2, D Low3, H Gay1, J Michalski1, S Mutic1, (1) Washington University School of Medicine, Saint Louis, MO, (2) Mayo Clinic, Rochester, MN, (3) UCLA, Los Angeles, CASU-E-J-169 Sunday 3:00PM - 6:00PM Room: Exhibit Hall
To develop a general strategy that allows for automatically, prospectively, and objectively determine the optimal CT simulation technique for individual prostate cancer patients based on radiation-therapy goals: maintenance of contouring quality and integrity while minimizing patient CT simulation dose.
An image quality index (IQI) was defined to characterize the simulation performance on target delineation required in radiation therapy. The optimal IQI is the one requiring the minimum dose to provide an image dataset that yields target contours that are equivalent to those segmented from greater-dose-technique scans. An anthropomorphic pelvis phantom with added-bolus layers were used to mimic prostate cancer patients with 38-53 cm lateral diameters for the experiments and the prediction model development. The manual prostate contour delineated from the highest dose CT scan was used as the ground truth to determine the optimal IQI for each patient size. A clinical workflow was recommended for selecting optimal CT simulation techniques, incorporating patient size, target delineation, and dose efficiency.
The median optimal IQI for accurate manual segmentation (compared to the ground truth) was 5.27 (range 4.8 to 6.32) for patient sizes of 38-53 cm. 120 kVp scan protocols could not reach the optimal IQI due to the tube power limit for patient sizes above 53 cm. The optimal tube potentials (yielding the lowest radiation dose) for patient sizes of 38-53 cm were 120, 140, 140, and 140, respectively. The corresponding minimum CTDIvol for achieving the optimal image quality were 16.7, 24.7, 48.7, and 96.9 mGy without tube current modulation enabled.
The proposed strategy can predict the optimal CT simulation scan protocols with minimum radiation dose while taking into account patient sizes and treatment planning tasks. We expect that the optimal image quality index will vary for automatic segmentation algorithms compare to the manual method.