Modified tumor diagnosis by classification and use of canonical correlation and support vector machines methods

The main objective of this research is to investigate techniques for classifying tumor grades based on image processing. The algorithms to classify tumors are introduced, and their performance for the experimental results are investigated. In the proposed algorithm, first, the scan images of the lung are pre-processed, and then the histogram, texture, and geometric features are extracted. These features are then used in the Support Vector Machines (SVM) and Canonical Correlation Analysis (CCA) classifiers to diagnose tumors and classify benign and malignant types. These combined techniques in understanding medical images for researchers are an essential tool to increase the accuracy of diagnosis. In this paper, simulated and real medical images are used. The results obtained from the proposed methods in this paper were compared with the previous findings to approve the proposed approach's efficacy and reliability in diagnosing and classifying tumors. In addition to high accuracy in diagnosis, this method is also a low-cost and low-risk method. Due to its very high sensitivity and having the desired values of two criteria of precision and specificity, and the low number of features used for classification, the developed method was proposed as an efficient and appropriate method for tumor classification.