Cross-Visit Tumor Sub-segmentation and Registration with Outlier Rejection for Dynamic Contrast-Enhanced MRI Time Series Data

Imaging Science and Biomedical Engineering, School of Cancer and Imaging Sciences, University of Manchester, Manchester, United Kingdom.
Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention 01/2010; 13(Pt 3):121-8. DOI: 10.1007/978-3-642-15711-0_16
Source: PubMed


Clinical trials of anti-angiogenic and vascular-disrupting agents often use biomarkers derived from DCE-MRI, typically reporting whole-tumor summary statistics and so overlooking spatial parameter variations caused by tissue heterogeneity. We present a data-driven segmentation method comprising tracer-kinetic model-driven registration for motion correction, conversion from MR signal intensity to contrast agent concentration for cross-visit normalization, iterative principal components analysis for imputation of missing data and dimensionality reduction, and statistical outlier detection using the minimum covariance determinant to obtain a robust Mahalanobis distance. After applying these techniques we cluster in the principal components space using k-means. We present results from a clinical trial of a VEGF inhibitor, using time-series data selected because of problems due to motion and outlier time series. We obtained spatially-contiguous clusters that map to regions with distinct microvascular characteristics. This methodology has the potential to uncover localized effects in trials using DCE-MRI-based biomarkers.

10 Reads
  • Source
    • "In general, the delivery and extravasation of CA is modeled with standard 2- or 3-compartment PK models (80) and, as mentioned above, these maps can be used to infer drug distribution in human tumors (77, 81). While many investigators use ROI (region of interest) analyses to derive a single perfusion value to describe a tumor, it is becoming increasingly appreciated that enhancement is heterogeneous and that quantitative descriptors of this heterogeneity improve the precision for diagnosis and monitoring of therapy response (82–86). We contend that perfusion heterogeneity is a key factor in the response of tumors to therapy, both in terms of drug delivery and in the establishment of specific habitats that select for cells with specific phenotypes and hence, therapy responses (87). "
    [Show abstract] [Hide abstract]
    ABSTRACT: Delivery of anti-cancer drugs to tumor tissues, including their interstitial transport and cellular uptake, is a complex process involving various biochemical, mechanical, and biophysical factors. Mathematical modeling provides a means through which to understand this complexity better, as well as to examine interactions between contributing components in a systematic way via computational simulations and quantitative analyses. In this review, we present the current state of mathematical modeling approaches that address phenomena related to drug delivery. We describe how various types of models were used to predict spatio-temporal distributions of drugs within the tumor tissue, to simulate different ways to overcome barriers to drug transport, or to optimize treatment schedules. Finally, we discuss how integration of mathematical modeling with experimental or clinical data can provide better tools to understand the drug delivery process, in particular to examine the specific tissue- or compound-related factors that limit drug penetration through tumors. Such tools will be important in designing new chemotherapy targets and optimal treatment strategies, as well as in developing non-invasive diagnosis to monitor treatment response and detect tumor recurrence.
    Frontiers in Oncology 11/2013; 3:278. DOI:10.3389/fonc.2013.00278
  • [Show abstract] [Hide abstract]
    ABSTRACT: PurposeTo compare fitting methods and sampling strategies, including the implementation of an optimized b-value selection for improved estimation of intravoxel incoherent motion (IVIM) parameters in breast cancer.Methods Fourteen patients (age, 48.4 ± 14.27 years) with cancerous lesions underwent 3 Tesla breast MRI examination for a HIPAA-compliant, institutional review board approved diffusion MR study. IVIM biomarkers were calculated using “free” versus “segmented” fitting for conventional or optimized (repetitions of key b-values) b-value selection. Monte Carlo simulations were performed over a range of IVIM parameters to evaluate methods of analysis. Relative bias values, relative error, and coefficients of variation (CV) were obtained for assessment of methods. Statistical paired t-tests were used for comparison of experimental mean values and errors from each fitting and sampling method.ResultsComparison of the different analysis/sampling methods in simulations and experiments showed that the “segmented” analysis and the optimized method have higher precision and accuracy, in general, compared with “free” fitting of conventional sampling when considering all parameters. Regarding relative bias, IVIM parameters fp and Dt differed significantly between “segmented” and “free” fitting methods.ConclusionIVIM analysis may improve using optimized selection and “segmented” analysis, potentially enabling better differentiation of breast cancer subtypes and monitoring of treatment. Magn Reson Med, 2014. © 2014 Wiley Periodicals, Inc.
    Magnetic Resonance in Medicine 10/2014; 74(4). DOI:10.1002/mrm.25484 · 3.57 Impact Factor