A hybrid supervised learning nasal tumor discrimination system for DMRI

Abstract

Dynamic magnetic resonance imaging (DMRI) has become one of the major tools for diagnosing nasal tumors in recent years. The purpose of this research is to develop a system that can discriminate between and enhance the differences between the tumor region and healthy tissue in DMRI automatically during the testing phase. Three supervised learning methods, the Adaboost, support vector machines (SVM), and Bayesian classifiers, are used for discriminating the tumor tissue from the normal tissue. A hybrid method, called ASB, is used to detect whether a given section of tissue is normal or tumor tissue based on the results obtained from the above three classifiers. The first stage of the method is a training process in which a tumor region is manually defined as the ground truth. Subsequently, a testing process is implemented with three typical classifiers, where the results of two out of the three classifiers are applied as a criterion for the discrimination. The experimental results show that ASB has a better differentiating rate than the Adaboost, SVM, and Bayesian classifiers. The ASB system shows an average sensitivity of 98.7%, an average specificity of 93.47%, and an average accuracy of 95.89%.

Tumor Segmentation in Contrast-Enhanced Magnetic Resonance Imaging for Nasopharyngeal Carcinoma: Deep Learning with Convolutional Neural Network

“…Huang et al [ 8 ] performed semisupervised NPC lesion extraction in MR images by using spectral clustering-based method with the positive predictive value up to 0.71. Huang et al [ 9 ] performed NPC tumor segmentation by using Bayesian classifiers and SVM method with average specificity of 0.93.…”