Predictive classification of correlated targets with application to detection of metastatic cancer using functional CT imaging

Abstract

Summary Perfusion computed tomography (CTp) is an emerging functional imaging modality that uses physiological models to quantify characteristics pertaining to the passage of fluid through blood vessels. Perfusion characteristics provide physiological correlates for neovascularization induced by tumor angiogenesis. Thus CTp offers promise as a non-invasive quantitative functional imaging tool for cancer detection, prognostication, and treatment monitoring. In this paper, we develop a Bayesian probabilistic framework for simultaneous supervised classification of multivariate correlated objects using separable covariance. The classification approach is applied to discriminate between regions of liver that contain pathologically verified metastases from normal liver tissue using five perfusion characteristics. The hepatic regions tend to be highly correlated due to common vasculature. We demonstrate that simultaneous Bayesian classification yields dramatic improvements in performance in the presence of strong correlation among intra-subject units, yet remains competitive with classical methods in the presence of weak or no correlation.

Dose-volume correlates of mandibular osteoradionecrosis in Oropharynx cancer patients receiving intensity-modulated radiotherapy: Results from a case-matched comparison

“…To account for this limitation and utilize more of the acquired information, we implemented a multivariate approach based on Bayesian quadratic discriminant analysis [20, 21] (BQDA) to evaluate the extent to which DVH characteristics could be integrated via continuous multivariate analysis to discriminate ORN from non-ORN on the basis of characteristics of the mandibular dose-volume. The Bayesian multivariate approach estimates the mean and covariance for the set of predictors for each class (ORN versus non-ORN) and yields a posterior probability measuring the extent of evidence that an inter-feature distribution arises from each candidate class[20, 21].…”

“…The Bayesian multivariate approach estimates the mean and covariance for the set of predictors for each class (ORN versus non-ORN) and yields a posterior probability measuring the extent of evidence that an inter-feature distribution arises from each candidate class[20, 21]. Receiver-operating-characteristics (ROC) curves were evaluated with respect to the resultant posterior probabilities.…”

“…BQDA was carried out using statistical software R version 3.2.3 using the analysis scripts published in Wang et al[20]. All other statistical analysis was performed using commercial statistical analysis software (JMP v11Pro, SAS Institute, Cary, NC, USA).…”

CT Perfusion Characteristics Identify Metastatic Sites in Liver

“…Moreover, the extent of interdependence varies substantially in magnitude and direction between vasculatures surrounding malignant and healthy tissues, providing additional signal for detecting sites where angiogenesis is taking place within the tumor microenvironment. Wang et al [ 17 ] proposed a spatial multivariate Bayesian approach to quadratic discriminant analysis that can be used to predict the status of multiple ROIs simultaneously. The multivariate model was shown to dramatically improve performance for predicting the status of liver ROIs using the perfusion characteristics acquired in our study.…”

Detection of prostate cancer with multiparametric MRI utilizing the anatomic structure of the prostate

“…Here, we present our general framework for developing a voxel‐wise classifier that accounts for heterogeneity in the mpMRI parameters and cancer risk as a function of location. We developed our voxel‐wise classifiers by modifying the framework proposed by Wang et al (2015), which generated region‐wise classification for metastatic liver cancer . Our basic approach is to jointly model the mpMRI parameters Y and the corresponding cancer status C conditional on the location information and use Bayesian posterior predictive probabilities conditional on the observed mpMRI parameters to predict the unknown cancer status for voxels in a new prostate slice.…”