Analysis of Machine Learning Algorithms for Classification of Osteosarcoma

The exploration of histopathological osteosarcoma data is a profitable technique to observe the pathological textures of osteosarcoma. Osteosarcoma classification is complex and inefficient due to inter-class similarities and variations. In this research, we employ two ML methods to classify OST: RF alias Random Forest and SVM alias Support Vector Machine. Sensitivity, Accuracy, F1-score, AUC, Kappa, and Specificity are the performance metrics used in the evaluation of ML models. Based on these classifiers, osteosarcoma can be classified as three categories which are explained in this work. The overall accuracy of the RF and SVM are 94.70% and 93.2% respectively. RF algorithm outperforms SVM algorithm with specificity 98.5%, F1 score 93.60%, sensitivity 92.30%, overall accuracy 94.70%, AUC 0.99 and Kappa 0.92.