(Sunday, 7/14/2019) 4:00 PM - 5:00 PM
Room: 225BCD
Purpose: To evaluate the accuracy of tissue segmentation for generating synthetic CT from MRI for abdominal radiotherapy.
Methods: An IRB�approved volunteer study was performed on a 3T MRI scanner. In�phase, fat and water images were acquired for five volunteers with breath-hold using a Dixon pulse sequence. A method to classify different tissue types for synthetic CT generation in the abdomen was developed. Intensity non-uniformity corrections were applied to MR images. Three tissue clusters (fat, high-density tissue, and bone marrow/air/lungs) were generated using a fuzzy-c means clustering algorithm. The third cluster was further segmented into three sub-clusters that represented bone marrow, air and lungs using the features of intensity values, areas and spatial locations. Therefore, five segments were automatically generated. To evaluate the segmentation accuracy using the method, the five segments were manually contoured on MRI images as the ground truth, and the volume ratio, Dice coefficient and Hausdorff distance metric were calculated.
Results: The volume ratio of auto-segmentation to manual segmentation was 0.88-2.09 for the air segment while it was 0.71-1.01 for the remaining segments. The range of the Dice coefficient was 0.24-0.83, 0.84-0.93, 0.94-0.98, 0.9-0.95 and 0.74-0.79 for air, fat, lungs, high-density tissue and bone marrow, respectively. The range of the mean Hausdorff distance was 3-29.1, 0.5-1.4, 0.4-1, 0.9-1.8 and 1.1-1.4 mm for air, fat, lungs, high-density tissue and bone marrow.
Conclusion: The segmentation accuracy was better for fat, lungs and high-density tissue, while it was the worst for air. However, the volume of the air segment was smaller than other tissue segments and the location of the air segment may change in the treatment course so its dosimetric effect may not be as critical. Our future work will include the dosimetric effect of the segmentation accuracy when using MRI only in the treatment planning process.
Funding Support, Disclosures, and Conflict of Interest: This work was supported by a research grant from Varian Medical Systems, Inc.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
