Click here to

Session: Outcome Modeling and Assessment [Return to Session]

From Person-Specific to Population-Based Colorectal Models: An Age-Scalable Computational Anatomical Colorectal Model for Radiation Dosimetry in Late Effects Studies of Childhood Cancer Survivors

C Owens1,2*, B Rigaud1, E Ludmir1, A Gupta1,2, S Shrestha1,2, A Paulino1, C Peterson1,2, S Kry1,2, S Smith1, K Brock1,2, R Howell1,2, (1) The University of Texas MD Anderson Cancer Center, Houston, TX, USA, (2) The University of Texas Graduate School of Biomedical Sciences at Houston, Houston, TX, USA


TU-A-TRACK 6-2 (Tuesday, 7/27/2021) 10:30 AM - 11:30 AM [Eastern Time (GMT-4)]

Purpose: To develop, validate and integrate into an age-scalable computational phantom an “average” colorectal model that incorporates anatomical variations of pediatric patients.

Methods: Whole-body non-contrast CT scans of fifty-five male pediatric patients (age range: 2.1-21.6years) were retrospectively selected. Manual colorectal contours were reviewed and approved by two radiation oncologists. Forty-five and ten patients were used for training and testing, respectively. All contours were spatially normalized using median colorectal length and registered to an anatomical template using the constrained symmetric thin-plate spline robust point matching (sTPS-RPM) method. Deformed contours were used to generate a principal component analysis (PCA)-based statistical shape model (SSM) of the colorectum to extract the dominant deformation modes of the population. The sTPS-RPM inter-patient registration was evaluated using Dice similarity coefficient (DSC), distance-to-agreement (DTA), and Hausdorff distance (HD). To validate the generalizability of the PCA model to represent unseen shapes, we computed the DSC between the original and PCA reconstructed contours for ten test patients. The population PCA-based colorectal model was integrated into our phantom, which can be scaled to any age (infant-to-adult). Additionally, the length and mass of the model was matched to ICRP reference data within 5% across the age range.

Results: The sTPS-RPM inter-patient registration mean(min-max) DSC, DTA and HD was 0.96(0.93-0.98), 1.5mm(1.2-2.0), and 6.0mm(3.2-19.7), respectively. On the training set, the PCA-based SSM was able to describe 95% of inter-patient colorectal variance with 11 of the 45 dominant principal components. Using the PCA-based SSM on the unseen 10 test patients, the mean(min-max) DSC between the validation and reconstructed colorectal contours was 0.80(0.76-0.84).

Conclusion: We developed, validated and integrated a population PCA-based SSM of the colorectum into our phantom. The model will be used to reconstruct colorectal doses for a study of subsequent colorectal cancers in childhood cancer survivors treated in the pre-CT era of RT.



    Phantoms, Radiation Dosimetry, Registration


    TH- Response Assessment: Radiation induced second cancers

    Contact Email