Tensor Radiomics: Paradigm for Systematic Incorporation of Multi-Flavoured Radiomics Features

Radiomics features extract quantitative information from medical images, towards the derivation of biomarkers for clinical tasks such as diagnosis, prognosis, or treatment response assessment. Different image discretization parameters (e.g. bin number or size), convolutional filters, segmentation perturbation, or multi-modality fusion levels can be used to generate radiomics features and ultimately signatures. Commonly, only one set of parameters is used, resulting in only one value or ‘flavour’ for a given feature. We propose ‘tensor radiomics’ (TR) where tensors of features calculated with multiple combinations of parameters (i.e. flavours) are utilized to optimize the construction of radiomics signatures. We have applied TR to PET/CT, MRI, and CT imaging, invoking machine learning or deep learning solutions, and reproducibility analyses. Given space limitations, here we present example results on PET/CT imaging: (1) TR via varying bin sizes on PET-CT images of head & neck cancer (HNC) for overall survival prediction. A hybrid deep neural network, referred to as ‘TR-Net’, along with two ML-based flavour fusion methods showed improved accuracy compared to regular radiomics features; and (2) TR via multiple PET/CT fusions in HNC: flavours were built from different fusions using methods such as Laplacian pyramids and wavelet transforms. We also tried TR based on segmentation perturbation flavours and preprocessing-filter flavours in CT and MR images. TR showed improved reproducibility as well as overall survival prediction compared to single-flavoured radiomics. Our results suggest that the proposed TR paradigm has the potential to improve performance capabilities in different medical imaging tasks.