Single and dual energy computed tomography in proton therapy

Purpose: The range uncertainties in proton therapy is a problem in current treatment planning of cancer, and is largely due to limitations in estimating stopping power (S) from computed tomography (CT) images. With a dual energy CT (DECT) scanning, monochromatic images can be provided that can potentially reduce noise and lead to better S estimates. In the current work, DECT is employed to investigate the impact of varying x-ray energy in conversion from Hounsfield Unit (HU) to S and proton treatment plan quality. Methods and materials: We reconstructed monochromatic images from 40 keV to 140 keV at 10 keV increments from data acquired by DECT and a 120 kVp CT image for two phantoms (Gammex and Quasar). We estimated Hounsfield Units (HU) values of different media (HE blood40, HE blood70, HE brain, CT HE solid water, HE blood100, HE adipose, lung, dense bone, water equivalent, inner bone, and polyethylene) and calculated the contrast to noise ratio (CNR) in each monochromatic image. From these calculations we found the monochromatic image with HU values and noise closest to the 120 kVp single energy CT (SECT) image. We then simulated a CT scanner in python and found the HU value for different media (Soft tissue, Blood, Brain, Inflated lung, Cortical bone, Adipose, and polyethylene) for both a monochromatic x-ray beam and 120 kVp x-ray beam. Stopping power ratio (SPR) was calculated for corresponding media, which was used to see investigate how conversion from HU to SPR changes with energies and noise in an image. DECT and CT was also done for two anthropomorphic head phantoms. The HU values for three media (soft tissue, bone, air) were collected in four different locations inside the head. These values were used to find a standard deviation between the locations. The 120 kVp image was then used to make a proton treatment plan for two different targets. The dose plans were recalculated in the monochromatic images to see how this would affect the dose calculation. A new HU-SPR CT calibration from the two first phantoms were used to see how the dose plan changes with CT calibration. Results: Measured and simulated Hus corresponded quite well across the different phantoms and material inserts. SPR was shown to depend roughly linearly with HU, but the relation changed with x-ray energy. In the DECT images a change in HU with energy and material compositions was seen a best fit to the 120 kVp SECT image. Comparing the HU-SPR relations for 40 keV, 80 keV, 140 keV, and 120 kVp showed a straighter line for the 80 keV DECT image. Proton planning with the SECT image series and different DECT images gave some variations in proton plan quality. Conclusion: Monochromatic images from DECT gave variations in Hus, leading to differences in HU-SPR conversion. If not accounted for, this may result in deterioration in proton plan quality. Still, DECT may improve proton plan quality if appropriate procedures are undertaken