A Comparative Exploration of ML Classifiers in Recognition of Ovarian Cancer

Ovaries are an essential aspect of a woman's reproductive system. The synthesis of female sex hormones and female gametes gives these little glands great significance. These small glandular structures, shaped like almonds and devoid of ducts, exist on either side of the uterus, connected by the ovarian ligament. Ovarian cancer can develop for several causes, but experts have identified a variety of distinct ways to categorize the disease. Predicting the onset of ovarian cancer at an early stage has the potential to save many lives by slowing the disease's progression. The use of CAD (Computer-aided diagnosis) systems is a noninvasive procedure for early detection of ovarian cancer that can spare patients the stress and inconvenience of a biopsy. Different methods are discussed in this review study for categorizing ovarian cancer tumors. We conducted an analysis of field data from Mass Spectrometry (MS) in order to develop a computer-assisted method. By using machine learning approaches, we were able to distinguish between malignant and benign cancerous cells by classifying the data into different groups. We conducted a comparison analysis to examine several approaches on different datasets and operational settings in order to determine which worked best. Our comparison analyses clearly show that the Multilayer Perceptron (MLP) is the best option for these kinds of detection tasks, outperforming other models in terms of accuracy, sensitivity, specificity, and errors.