z-logo
Premium
SU‐C‐WAB‐01: Computerized Production of Statistical Clinical Target Volume Models in Prostate Cancer Radiation Treatment Planning
Author(s) -
matsushita N,
Arimura H,
Nakamura K,
Kagami Y,
Shioyama Y,
Nakamura Y,
Honda H,
Hirata H
Publication year - 2013
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4813953
Subject(s) - contouring , prostate cancer , principal component analysis , nuclear medicine , medicine , prostate , mathematics , computer science , statistics , cancer , computer graphics (images)
Purpose: Inter‐and intra‐observer variability of clinical target volume (CTV) contours for prostate cancer is a critical issue in radiation treatment planning. Automated contouring based on statistical CTV models could be one of possible approaches for reducing the variability. Therefore, the purpose of our study is to develop a computerized method for producing the statistical CTV models to be used as CTV templates for automated contouring in prostate cancer radiation treatment planning. Methods: Eighteen, seventeen and sixteen radiation oncologists belonging to different facilities delineated CTV contours for low‐, intermediate‐and high‐risk prostate cancers, respectively. The low‐risk, intermediate‐risk, and high‐risk CTVs include only a prostate, a prostate plus a 1‐cm seminal vesicle base, and a prostate plus 2‐cm seminal vesicle, respectively. The statistical CTV models for the three risk types were derived based on a principal component analysis (PCA), which statistically took into account the inter‐observer variability. CTV regions were triangulated by using a marching cubes method. For matching the number of points on the surfaces of all CTV regions, the number of vertices on each CTV region was reduced to 1,000 with a quadric error metrics. All CTV regions were registered with a reference CTV by using an iterative closest point algorithm for calculation of a covariance matrix to be employed for the PCA‐based CTV modeling. Results: CTV models of three risk types were produced, which consisted of a mean CTV and PCA coefficients multiplied by eigenvectors, i.e., eigen‐CTV image. Average Euclidean distances between corresponding points of a mean CTV and the other CTVs for three risk types were 2.05 mm, 3.64 mm, 2.55 mm, respectively. Conclusion: The proposed method could provide probabilistic CTV models to be used as CTV templates in delineating CTV contours.

This content is not available in your region!

Continue researching here.

Having issues? You can contact us here