Title :
Hierarchical and adaptive volume-growing methods for the PET based biologic target volume delineation for radiotherapy treatment planning
Author :
Liu, Guocai ; Zhu, Suyu ; Wang, Yaonan ; Zhang, Jiutang ; Hu, Bingqiang ; Yu, Zhihao ; Wu, Haiyan ; Yang, Weili
Author_Institution :
Coll. of Electr. & Inf. Eng, Hunan Univ., Changsha, China
Abstract :
To more accurately and precisely delineate a biologic target volume (BTV) for PET-guided radiotherapy treatment planning, we proposed a novel tumor segmentation method by hierarchically utilizing an adaptive volume-growing algorithm (AVGA). First, a rough volume of interest (VOI) is manually cut by a radiation oncologist that encloses a tumor volume. The voxel having the highest standard uptake values (SUV) in the manual VOI is chosen as a seed of the first AVGA. One of all 6-neighboring voxels (N6) of the current tumor volume (CTV) is successively appended to the CTV. The growing criterion of the AVAG is that the SUV of the appended voxel is higher than the mean SUV of the CTV multiplied by a threshold value, and its SUV subtracted from the mean SUV of the CTV is a minimum in the N6. We have always observed a sharp volume growth at a certain threshold value T1 between 0.3 and 0.7, in which case many voxels outside of the delineated tumor volume are appended to the CTV. So, the first AVGA is stopped just before the sharp volume growth and the second AVGA is launched to refine the BTV delineation. The resulting CTV by the first AVGA is chosen as a seed volume of the second AVGA, and the threshold varies from T1 to 0.30 with an adding growing criterion which the SUV of the current appended voxel is not higher than one of the last appended voxel. The idea behind is high-noise characteristics, partial volume effects due to the low spatial resolution, the tumor extension with lower SUV, and the spillover effects of some PET images. Patient studies were tested and evaluated by an experienced radiation oncologist. The results demonstrated that the proposed method can more accurately and precisely delineate a BTV due to the second AVGA. This method can automatically process PET images with different sizes of tumor volume and different signal to background ratios (SBR). The resulting BTV is reproducible and also independent of the- initial VOI.
Keywords :
image segmentation; medical image processing; planning; positron emission tomography; radiation therapy; tumours; PET based biologic target volume delineation; adaptive volume-growing methods; hierarchical volume-growing methods; high-noise characteristics; image processing; partial volume effects; radiation oncologist; radiotherapy treatment planning; rough volume-of-interest; sharp volume growth; spatial resolution; spillover effects; tumor extension; tumor segmentation; Image resolution; Image segmentation; Physiology; Planning; Robustness;
Conference_Titel :
Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC), 2011 IEEE
Conference_Location :
Valencia
Print_ISBN :
978-1-4673-0118-3
DOI :
10.1109/NSSMIC.2011.6153826