Classification of Medical Image Modeling Methods: A Review

Abstract

Image modeling can be concerned as a basic core of many medical image analysis/processing systems. Indeed, proposed model for a medical image defines the required process such as coding, compression, contrast enhancement, denoising, feature extraction, classification, etc. In this paper, we present a comprehensive classification of the models used in medical image processing. Assortment of various models can be done in different manners. In a wide categorization, each model can be applied in spatial or transform domain. In transform domain, we divide models into two subgroups subject to the choice of the basis function as a data adaptive and non-data adaptive transform models. Beside this classification, we categorize all the models in both spatial or transform domain, as a deterministic, stochastic, geometric, or partial differential equation (PDE) based models. After describing each of these models, we provide a tree structure figure to display the classification of these models and the relations between them. Although we attempt to present an all-around classification of different models used in medical images, it is necessary to note that these models are not entirely distinct from each other and in some cases they may overlap with each other. In addition, we try to illustrate by two examples on retinal Optical Coherence Tomography (OCT) and color fundus images that how this categorization of models can be used in different image processing applications and conclude that considering this classification can help researchers to have a cognitive selection of models for their own specified goals.