DCE-MRI model selection for investigating disruption of microvascular function in livers with metastatic disease

Authors

  • Anita Banerji MSc,

    1. Imaging Sciences, University of Manchester, Manchester Academic Health Science Centre, Manchester, UK
    2. Biomedical Imaging Institute, University of Manchester, Manchester, UK
    Search for more papers by this author
  • Josephine H. Naish PhD,

    1. Imaging Sciences, University of Manchester, Manchester Academic Health Science Centre, Manchester, UK
    2. Biomedical Imaging Institute, University of Manchester, Manchester, UK
    Search for more papers by this author
  • Yvonne Watson DCR(D),

    1. Imaging Sciences, University of Manchester, Manchester Academic Health Science Centre, Manchester, UK
    2. Biomedical Imaging Institute, University of Manchester, Manchester, UK
    Search for more papers by this author
  • Gordon C. Jayson FRCP, PhD,

    1. Biomedical Imaging Institute, University of Manchester, Manchester, UK
    2. School of Cancer and Enabling Sciences, University of Manchester, Manchester, UK
    Search for more papers by this author
  • Giovanni A. Buonaccorsi PhD,

    1. Imaging Sciences, University of Manchester, Manchester Academic Health Science Centre, Manchester, UK
    2. Biomedical Imaging Institute, University of Manchester, Manchester, UK
    Search for more papers by this author
  • Geoffrey J.M. Parker PhD

    Corresponding author
    1. Imaging Sciences, University of Manchester, Manchester Academic Health Science Centre, Manchester, UK
    2. Biomedical Imaging Institute, University of Manchester, Manchester, UK
    • Imaging Science and Biomedical Engineering, University of Manchester, Stopford Building, Oxford Road, Manchester, M13 9PT, UK
    Search for more papers by this author

Abstract

Purpose:

To evaluate the Akaike information criterion (AIC) model selection technique as a method for detecting differences in microvascular characteristics between tumorous and non-tumor liver tissue.

Materials and Methods:

The AIC was applied to six patient datasets with liver metastases to determine, on a per voxel basis, which of two physiologically plausible candidate models gave a more appropriate description of the data. The dual-input single-compartment Materne model, extended to incorporate a novel portal input function estimation method, was chosen to represent liver tissue and the single-input dual-compartment extended Kety model was used for tumor.

Results:

Median AIC probabilities when comparing tumor versus liver and tumor versus tumor-margins were significantly different (P ≤ 0.01) in five of the six patient datasets. Comparisons between tumor margins and liver regions were significantly different in four datasets. Median AIC probabilities selected for the extended Kety model in all tumor regions, with the Materne model being progressively more probable through tumor margins into liver.

Conclusion:

We present a viable method for assessing the spatially varying microvascular characteristics of tumor-bearing livers, with possible applications in lesion detection, assessment of tumor invasion, and measurement of drug efficacy. J. Magn. Reson. Imaging 2012;35:196-203. © 2011 Wiley Periodicals, Inc.

Ancillary