Skip to Main Content
Complexity is considered with high interest to texture analysis, such as the application of lymphoma inspection. Previously, multiscale entropy was proposed to account for the multiple time scales inherent in complex biological signals, and complexity texture descriptors were considered for content-based medical image retrieval system. However, in the past, these complexity texture descriptors were derived by linear algorithms of Fourier and wavelet transforms. In this study, we introduce a new approach of multiscale complexity (MSC) for texture analysis of lymphomas based on a nonlinear parameter of entropy. An extended algorithm of 2-D entropy was used in the calculations of complexities for the coarse-grained images of different scale factors. Furthermore, five parameters are defined to characterize 58 images of reactive lymphadenopahty and five categories of lymphomas using complexity on three scales and the changes of complexity between two successive scales. MSC analysis performed well to figure out the statistical differences of paired comparisons picked from reactive lymphadenopahty and five categories of lymphomas.