Classification of Chest Lesions with Using Fuzzy C-Means Algorithm and Support Vector Machines

Abstract

Abstract. The specification of the nature of the lesion detected is a hard task for chest radiologists. While there are several studies reported in developing a Computer Aided Diagnostic system (CAD), they are limited to the distinction between the cancerous lesions from the non-cancerous. However, physicians need a system which is significantly analogous to a human judgment in the process of analysis and decision making. They need a classifier which can give an idea about the nature of the lesion. This paper presents a comparative analysis between the classification results of the Fuzzy C Means (FCM) and the Support Vector Machines (SVM) algorithms. It discusses also the possibility to increase the interpretability of SVM classifier by its hybridization with the Fuzzy C method.

A Hybrid Fuzzy-SVM classifier for automated lung diseases diagnosis

“…Here, we only describe the final stage of our CAD system which is a continuation of our already published work in [11]: The segmentation and the detection are discussed in our previous works [12] and [13] that are briefly presented in the next sections.…”

“…To this aim, we computed several features based on characteristics given by service of medical imaging of CHU Charles Nicolle (described in our published work [11]). …”

“…The commonly used fuzzy clustering methods are: FCM Fuzzy C-Means and its variants. The detail of this part is already published in our previous work in [11]. …”

An Improved Hybrid Canopy-Fuzzy C-Means Clustering Algorithm Based on MapReduce Model

“…The classification of application areas in this survey has been made in the following manner: neural network, clustering and classification, image analysis, structural analysis of algorithms together in various application domains. A comparative study between FCM and SVM has been performed by Hassen et al [2] for classification of chest lesions and elaborates the possibility of increasing the interpretability of SVM classifier by hybridizing with FCM. Bharill and Tiwari [3] introduced a random sampling iterative optimization fuzzy c-means (RSIO-FCM) clustering algorithm which partitions large datasets into various subsets and results in formation of effective clusters for elimination of the problem of overlapping cluster centers.…”