Constructing multi-modality and multi-classifier radiomics predictive models through reliable classifier fusion

Radiomics aims to extract and analyze large numbers of quantitative features from medical images and is highly promising in staging, diagnosing, and predicting outcomes of cancer treatments. Nevertheless, several challenges need to be addressed to construct an optimal radiomics predictive model. First, the predictive performance of the model may be reduced w hen features extracted from an individual imaging modality (e.g. PET, CT, MRI) are blindly combined into a single predictive model. Second, because many different types of classif iers are available to construct a predictive model, selecting an “optimal” classif ier for a particular application is still challenging. In this w ork, w e developed multi-modality and multi-classif ier radiomics predictive models that address the aforementioned issues in currently available models. Specif ically, a new reliable classif ier fusion strategy w as proposed to optimally combine output from different modalities and classif iers. In this strategy, modalityspecif ic classifiers were f irst trained, and an analytic evidential reasoning (ER) rule w as developed to fuse the output scor e from each modality to construct an optimal predictive model. One public datasets and tw o clinical case studies w ere performed to validate model performance. The experimental results indicated that the proposed ER rule based radiomics models outperformed the traditional models that rely on a single classif ier or simply use combined features from different modalities.