Fractal-Radiomics as Complexity Analysis of CT and MRI Cancer Images

Cancer is the second leading cause of death globally. Early diagnosis can allow intervention to reduce mortality but due to cancer complex structure and spatial heterogeneity among different tumors and within each lesion, it is difficult to differentiate it from healthy tissue using conventional imaging techniques. Quantification of its complexity can be a prognostic tool for fighting this disease. In recent years, clinical imaging allows this quantification thanks to Radiomics, which extracts features from images. In this study, Fractal Dimension (FD) and Lacunarity $(\pmb{L})$ in computed tomography (CT) and magnetic resonance (MR) images for different kinds of cancer were examined using box counting method. Our aim is to highlight the potentiality of features based on fractal analysis, in order to obtain new indicators able to detect tumor spatial complexity and heterogeneity. The results indicated that both FD and $\pmb{L}$ show problems linked to the lack of connection between complexity estimated with Radiomics and the underlying biological model.