Open Collections

Prostate cancer detection from magnetic resonance images : a data-driven approach

University of British Columbia

The combination of Dynamic Contrast Enhanced (DCE) images with Diffusion Tensor Images (DTI) has shown great potential in prostate cancer detection. The parametrization of DCE images to generate cancer markers is traditionally performed based on pharmacokinetic modeling. However, pharmacokinetic models make simplistic assumptions about the tissue perfusion process, require the knowledge of contrast agent concentration in a major artery, and the modeling process is sensitive to noise and fitting instabilities. In this work, this issue is addressed by extracting features directly from the DCE T1-weighted signal intensities without modeling the physical perfusion phenomenon. In this work, a novel set of data-driven features are proposed which are generated by mapping the DCE T1-weighted signal intensity to its principal component space. The optimal set of components is extracted with sparse regularized regression through a Least Absolute Shrinkage and Selection Operator (LASSO) model. It is shown that when the proposed features are used to replace pharmacokinetic parameters, the Area under receiver operating characteristics Curve (AUC) is 0.86, with a support vector machine classifier trained on the peripheral zone of prostate. When the proposed features are used within the multiparametric MRI (mpMRI) protocol with the DTI feature, the area under ROC was 0.91 for the peripheral zone classifier, and 0.87 for the whole gland classifier. We showed that in 85.0% cases, the mpMRI whole gland classifier detected more than 50% area of the tumor. The proposed features were used to generate cancer likelihood maps for the prostate gland. These likelihood maps show the likelihood of cancer for each pixel and hence highlight the regions from where the biopsy samples should be taken. The generated cancer likelihood maps have the potential to be used as a reference in MRI-targeted biopsies to decrease the possibility of missing clinically significant and potentially aggressive tumors.

Text

2014-08-19

Attribution-NonCommercial-NoDerivs 2.5 Canada

http://hdl.handle.net/2429/50058

Biomedical Engineering

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc-nd/2.5/ca/