Tumor growth prediction with hyperelastic biomechanical model, physiological data fusion, and nonlinear optimization

Monday, September 22, 2014 — Poster Session I
12:00 p.m. – 2:00 p.m.	FAES Academic Center	CC	BIOENG-11

* FARE Award Winner

Authors

• K. Wong
• R. Summers
• E. Kebebew
• J. Yao

Abstract

Tumor growth prediction is to accurately model the tumor growth process, which can be achieved by combining physiological modeling with medical images. Several issues limit the accuracy of existing frameworks, including the unrealistic infinitesimal strain assumption of mass effect, complicated model personalization, and the lack of functional information. Therefore, we propose a framework which comprises a hyperelastic biomechanical model for physiological plausibility, gradient-free nonlinear optimization for flexibility, and physiological data fusion of structural and functional images for subject-specificity. Our model is driven by longitudinal dual-phase CT and FDG-PET images. Experiments were performed on data sets from eight patients with pancreatic neuroendocrine tumors. Each data set contains three time points, with the first two for model personalization, and the last one for prediction performance evaluation. Each time point contains a tumor volume segmented from post-contrast CT, intracellular volume fractions computed from dual-phase CT, and standardized uptake values computed from FDG-PET. The results show that through the nonlinear optimization, the complementary structural and functional information can be combined to improve the patient-specificity of the tumor growth model, and thus the prediction accuracy. The recall, precision, and relative volume difference between predicted and measured tumor volumes are 84.85+/-6.15%, 87.08+/-7.83%, and 13.81+/-6.64%, respectively.

back to top