Ovarian cancer is one the common cancers that in women such pathological disease within an organ might lead to noticeable changes in the proteomic patterns in serum. Mass spectrometry is the most important tool to understand the proteomic profiles proteomic changes; mass spectrometry extracts complex and informative functional data; and the most significant features of it are the peaks. This article presents a comparison of 4 widely used machine learning (ML) algorithms and 2 feature selection algorithms. The ML algorithms were applied on low-resolution surface-enhanced laser desorption/ionization–time-of-flight data sets for ovarian cancer diagnosis, by extracting wavelet features from spectrometer data and feeding them to the classifiers. The comparison is done by fusion of both selected features using the different algorithms with the classifiers, and then they were compared by measuring their classification test accuracy, sensitivity, and specificity values. Results show that all the presented ML algorithms performed well, with different feature selection algorithms all exceeding 90% accuracy.
Corresponding author: Ali Mohammad Alqudah, MSc, is with the Department of Biomedical Systems and Informatics Engineering at Yarmouk University, P.O. Box 566, Irbid 21163, Jordan, and can be reached at email@example.com.
The author declares no conflicts of interest.