Enhancing Acurate Diagnosis of Inflammatory Breast Cancer Based on Polar Fox Optimization

Abstract

Inflammatory breast cancer (IBC) presents a significant challenge due to its aggressive nature, necessitating a more precise definition, which can hinder clinical care, trials, and the development of IBC-specific biomarkers and treatments. Our goal is to create a deep learning-based computer-aided diagnosis (CAD) system that can detect and classify LABC and IBC based on histopathological features of H&E whole-slide images (WSIs) of breast cancer patients. In our study, a total of 100 histopathologic H&E WSIs were collected and classified into two categories: LABC and IBC. The proposed model starts by dividing the data into training and testing sets, which represent 70% and 30% of the total samples of collected data. Then we used the feature extraction technique to extract the features from the two datasets. In the next process, we developed a Polar Fox Optimization (PFO) feature selection model for LABC and IBC prediction. The proposed PFO algorithm has high average accuracy, followed by GWCA and AROA. It is more stable than other algorithms and has the smallest fitness value in terms of mean and best value. The AROA model has the smallest fitness value, while GWCA and PFO are more stable. The model's ability to select the smallest number of relevant features is observed. This study focused on improving the accuracy of machine learning classification methods by applying a Polar Fox Optimization (PFO) feature selection algorithm that selects the most effective features for LABC and IBC diagnosis among all their histopathological features. The proposed PFO has a high ability to detect LABC and IBC cases based on the accuracy and fitness value measures.