(Monday, 7/15/2019) 9:30 AM - 10:00 AM
Room: Exhibit Hall | Forum 5
Purpose: In general, pancreatic gross tumor volume (GTV) is not adequately distinguishable from surrounding tissue from CT alone but may be identified using multimodality images, reliant on the accuracy of the registration which is often affected by anatomic variations between the images. This work aims to investigate the use of quantitative textures from dual-energy CT (DECT) to improve the differentiation between pancreatic adenocarcinoma (PDAC) and surrounding pancreatic tissues, thus improving GTV delineation directly on the planning CT used for radiation therapy planning and/or online adaptive radiation therapy.
Methods: DECT data for 7 PDAC patients acquired using a dual source DECT CT simulator (Drive, Siemens) were analyzed. For each case, monoenergetic decomposition CTs (MDCTs) with energies from 40 keV to 160 keV were reconstructed. The GTV and pancreas were delineated by experienced radiation oncologist and radiologist based on multimodality images and were populated on to these MDCTs. A series of CT texture features for the GTV and the pancreas were extracted from MDCTs. The differences in the texture features between PDAC and pancreas tissue were analyzed.
Results: The percent difference in the mean CT number increases with decreasing energy of MDCT from an average of -4.6% at 160 keV to -22.7% at 40 keV. The minimum difference in standard deviation over all patients and all energies is -64% and decreases with energy, leveling off after 120 keV for an average of -97%.
Conclusion: Differentiation of pancreas GTV from the surrounding pancreatic tissue can be substantially increased by using the texture features of MDCT from DECT and this increase is dependent on the energy of MDCT. With further studies of more patient data and texture analyses, DECT will be approved as an important image modality for radiation therapy planning and delivery guidance.
Keywords
CT, Quantitative Imaging, Dual-energy Imaging
Taxonomy
Contact Email
