Boosting in Nonlinear Regression Models with an Application to DCE-MRI Data^[*]

Summary

Background: For the statistical analysis of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) data, compartment models are a commonly used tool. By these models, the observed uptake of contrast agent in some tissue over time is linked to physiologic properties like capillary permeability and blood flow. Up to now, models of different complexity have been used, and it is still unclear which model should be used in which situation. In previous studies, it has been found that for DCE-MRI data, the number of compartments differs for different types of tissue, and that in cancerous tissue, it might actually differ over a region of voxels of one DCE-MR image. Objectives: To find the appropriate number of compartments and estimate the parameters of a regression model for each voxel in an DCE-MR image. With that, tumors in an DCE-MR image can be located, and for example therapy success can be assessed. Methods: The observed uptake of contrast agent in a voxel of an image of some tissue is described by a concentration time curve. This curve can be modeled using a nonlinear regression model. We present a boosting approach with nonlinear regression as base procedure, which allows us to estimate the number of compartments and the related parameters for each voxel of an DCE-MR image. In addition, a spatially regularized version of this approach is proposed. Results: With the proposed approach, the number of compartments – and with that the complexity of the model – per voxel is not fixed but data-driven, which allows us to fit models of adequate complexity to the concentration time curves of all voxels. The parameters of the model remain nevertheless interpretable because of the underlying compartment model. Conclusions: The proposed boosting approaches outperform all competing methods considered in this paper regarding the correct localization of tumors in DCE-MR images as well as the spatial homogeneity of the estimated number of compartments across the image, and the definition of the tumor edge.

Keywords

Statistical computing - computer-assisted image processing - statistical models - algorithms - regression analysis - spatial regularization - biological models - magnetic resonance imaging

• References

• 1 Buckley D, Parker G. Measuring contrast agent concentration in Ti-weighted dynamic contrast-enhanced MRI. In: Jackson A, Buckley D, Parker G. editors. Dynamic Contrast-Enhanced Magnetic Resonance Imaging in Oncology. New York: Springer; 2005. pp 69-79.
• 2 Padhani AR, Ah-See ML, Makris A. MRI in the detection and management of breast cancer. Expert Rev Anticancer Ther 2005; 5 (Suppl. 02) 239-252.
• 3 Liu G, Rugo HS, Wilding G, Mc-Shane TM, Evelhoch JL, Ng C. et al. Dynamic contrast-enhanced magnetic resonance imaging as a pharmacodynamic measure of response after acute dosing of AG-013736, an oral angiogenesis inhibitor, in patients with advanced solid tumors: results from a phase I study. J Clin Oncol 2005; 23 (Suppl. 24) 5464-5473.
• 4 Tofts P, Kermode A. Measurement of the blood-brain barrier permeability and leakage space using dynamic MR imaging – 1. Fundamental concepts. Magn Reson Med 1991; 17: 357-367.
• 5 Schmid V, Whitcher B, Padhani A, Yang G. Quantitative analysis of dynamic contrast-enhanced MR images based on Bayesian P-Splines. IEEE Trans Med Imaging 2009; 28 (Suppl. 06) 789-798.
• 6 Brix G, Zwick S, Kiessling F, Griebel J. Pharmacokinetic analysis of tissue microcirculation using nested models: multimodel inference and parameter identifiability. Med Phys 2009; 36 (Suppl. 07) 2923-2933.
• 7 Sourbron S, Buckley D. On the scope and interpretation of the Tofts models for DCE-MRI. Magn Reson Med 2011; 66 (Suppl. 03) 735-745.
• 8 Sommer J, Schmid V. Spatial two-tissue compartment model for dynamic contrast-enhanced magnetic resonance imaging. J R Stat Soc C-App 2014; 63: 695-713.
• 9 Port R, Knopp M, Hoffmann U, Milker-Zabel S, Brix G. Multicompartment analysis of gadolinium chelate kinetics: blood-tissue exchange in mam-mary tumors as monitored by dynamic MR imaging. J Magn Reson Imaging 1999; 10: 233-241.
• 10 Sommer J, Gertheiss J, Schmid V. Spatially regularized estimation for the analysis of dynamic contrast-enhanced magnetic resonance imaging data. Stat Med 2014; 33: 1029-1041.
• 11 Freund Y, Shapire R. Experiments with a new boosting algorithm. In: Proceedings of the Thirteenth International Conference on Machine Learning Theory. 3. San Francisco, CA: Morgan Kaufmann Publishers Inc; 1996. pp 148-156.
• 12 Mayr A, Binder H, Gefeller O, Schmid M. Extending statistical boosting. An overview of recent methodological developments. Methods Inf Med 2014; 53: 428-435.
• 13 Friedman J, Hastie T, Tibshi-rani R. Additive logistic regression: a statistical view of boosting (with Discussion). Ann Stat 2000; 28: 337-407.
• 14 Friedman J. Greedy function approximation: A gradient boosting machine. Ann Stat 2001; 29: 1189-1232.
• 15 Biihlmann P, Yu B. Boosting with the L2-L0SS: Regression and classification. J Am Stat Assoc 2003; 98: 324-338.
• 16 Biihlmann P. Boosting for high-dimensional linear models. Anna Stat 2006; 34: 559-583.
• 17 Biihlmann P, Hothorn T. Boosting algorithms: Regularization, prediction and model fitting. Stat Sci 2007; 22: 477-522.
• 18 Mayr A, Binder H, Gefeller O, Schmid M. The evolution of boosting algorithms. From machine learning to statistical modelling. Methods Inf Med 2014; 53: 419-427.
• 19 Feilke M. dcemriboost: Boosting for DCE-MRI data. 2015 R package version 0.1. Available from: https://github.com/feilke/dcemriboost.
• 20 Tofts P. Modeling tracer kinetics in dynamic GdDTPA MR imaging. J Magne Reson Imaging 1997; 7 (Suppl. 01) 91-101.
• 21 Tofts P, Brix G, Buckley D, Evelhoch J, Henderson E, Knopp M. et al. Estimating kinetic parameters from dynamic contrast-enhanced T1-weighted MRI of a diffusable tracer: Standardized quantities and symbols. J Magn Reson Imaging 1999; 10: 223-232.
• 22 Sommer J. Regularized estimation and model selection in compartment models. Dissertation at the Faculty of Mathematics, Computer Science and Statistics of the LMU Munich. Munich, Germany: Verlag Dr. Hut; 2013
• 23 Yang X, Kopp MV. Quantifying tumor vascular heterogeneity with dynamic contrast-enhanced magnetic resonance imaging: a review. J Biomed Biotech 2011; 2011: 1-12.
• 24 Schmid V, Whitcher B, Padhani A, Taylor N, Yang GZ. A Bayesian hierarchical model for the analysis of a longitudinal dynamic contrast-enhanced MRI oncology study. Magn Reson Med 2009; 61 (Suppl. 01) 163-174.
• 25 Seber GAF, Wild CJ. Nonlinear Regression. New York: Wiley; 1989
• 26 Schmid VJ, Whitcher B, Padhani AR, Taylor NJ, Yang GZ. Bayesian methods for pharmacokinetic models in dynamic contrast-enhanced magnetic resonance imaging. IEEE Trans Med Imaging 2006; 25 (Suppl. 12) 1627-1636.
• 27 Parker G, Suckling J, Tanner S, Padhani A, Husband J, Leach M. MRIW: Parametric analysis software for contrast-enhanced dynamic MR imaging in cancer. Radiographics 1998; 18 (Suppl. 02) 497-506.
• 28 Weinmann H, Laniado M, Miitzel W. Pharmokinetics of Gd-DTPA/dimeglumine after intravenous injection into healthy volunteers. Physiol Chem Phys Med NMR 1984; 16: 167-172.
• 29 Hofner B, Mayr A, Robinzonov N, Schmid M. Model-based boosting in R: a hands-on tutorial using the R package mboost. Comput Stat 2014; 29: 3-35.
• 30 R Core Team.. R: A language and environment for statistical computing. Vienna, Austria: 2013 ISBN 3-900051-07-0. Available from: http://www.Rproject.org.
• 31 Elzhov T, Mullen K, Spiess AN, Bolker B. minpack.lm: R interface to the Levenberg-Marquardt nonlinear least-squares algorithm found in MIN-PACK, plus support for bounds. 2013 R package version 1.1-8. Available from: http://CRAN.Rproject.org/package=minpack.lm.
• 32 Ahearn T, Staff R, Redpath T, Semple S. The use of the Levenberg-Marquardt curve-fitting algorithm in pharmacokinetic modelling of DCE-MRI data. Phys Med Biol 2005; 50 (Suppl. 09) N85-N92.
• 33 Reich J. On parameter redundancy in curve fitting of kinetic data. Kinetic data analysis: Design and analysis of enzyme and pharmacokinetic experiments. New York: Plenum Press; 1981
• 34 Schmid M, Hothorn T. Boosting additive models using componentwise P-splines. Comput Stat Data An 2008; 53: 298-311.
• 35 Mayr A, Hofner B, Schmid M. The importance of knowing when to stop. A sequential stopping rule for component-wise gradient boosting. Methods Inf Med 2012; 51 (Suppl. 02) 178-186.
• 36 Hastie T, Tibshirani R, Friedman J. The elements of statistical learning. 2nd ed. New York: Springer; 2009
• 37 Hansen M, Yu B. Model selection and the principle of minimum description length. J Am Stat Assoc 2001; 96 (Suppl. 454) 746-774.
• 38 Candes E, Tao T. The Dantzig selector: Statistical estimation when p is much larger than n. Ann Stat 2007; 35 (Suppl. 06) 2313-2351.
• 39 Gertheiss J, Tutz G. Sparse modeling of categorial explanatory variables. Ann Appl Stat 2010; 4 (Suppl. 04) 2150-2180.
• 40 Tutz G, Groll A. Likelihood-based boosting in binary and ordinal random effects models. J Comput Graph Stat 2013; 22 (Suppl. 02) 356-378.
• 41 Groll A. GMMBoost: Likelihood-based Boosting for Generalized mixed models. 2013 R package version 1.1.2. Available from: http://CRAN.Rproject.org/package=GMMBoost.
• 42 Turlach B, Weingessel A. quad-prog: Functions to solve Quadratic Programming Problems. 2013 R package version 1.5–5. Available from: http://CRAN.R-project.org/package=quadprog.
• 43 Zou H, Hastie T. Regularization and variable selection via the elastic net. J Roy Stat Soc B 2005; 67: 301-320.
• 44 Ah-See ML, Makris A, Taylor N, Burcombe R, Harrison M, Stirling J. et al. Does vascular imaging with MRI predict response to neoadjuvant chemo-therapy in primary breast cancer?. J Clin Oncol 2004; 22 14S 582.
• 45 Schmid V, Whitcher B, Yang G, Taylor N, Padhani A. Statistical analysis of pharmacokinetical models in dynamic contrast-enhanced magnetic resonance imaging. In: Duncan J, Gerig S. editors. Medical Imaging Computing and Computer-Assisted Intervention – MICCAI 2005. New York: Springer; 2005. pp 886-893.
• 46 Bischl B, Lang M, Mersmann O, Rahnenfuehrer J, Weihs C. BatchJobs and BatchExperiments: Abstraction mechanisms for using R in batch environments. J Stat Softw 2015; 64: 1-25.