Artificial intelligence breakthroughs in pioneering early diagnosis and precision treatment of breast cancer: A multimethod study

Mohammad Reza Darbandi,Mahsa Darbandi,Sara Darbandi,Igor Bado,Mohammad Hadizadeh,Hamid Reza Khorram Khorshid

DOI: https://doi.org/10.1016/j.ejca.2024.114227

Abstract:This article delves into the potential of artificial intelligence (AI) to enhance early breast cancer (BC) detection for improved treatment outcomes and patient care. Utilizing a multimethod approach comprising literature review and experiments, the study systematically reviewed 310 articles utilizing 30 diverse datasets. Among the techniques assessed, recurrent neural network (RNN) emerged as the most accurate, achieving 98.58 % accuracy, followed by genetic principles (GP), transfer learning (TL), and artificial neural networks (ANNs), with accuracies exceeding 96 %. While conventional machine learning (ML) methods demonstrated accuracies above 90 %, DL techniques outperformed them. Evaluation of BC diagnostic models using the Wisconsin breast cancer dataset (WBCD) highlighted logistic regression (LR) and support vector machine (SVM) as the most accurate predictors, with minimal errors for clinical data. Conversely, decision trees (DT) exhibited higher error rates due to overfitting, emphasizing the importance of algorithm selection for complex datasets. Analysis of ultrasound images underscored the significance of preprocessing, while histopathological image analysis using convolutional neural networks (CNNs) demonstrated robust classification capabilities. These findings underscore the transformative potential of ML and DL in BC diagnosis, offering automated, accurate, and accessible diagnostic tools. Collaboration among stakeholders is crucial for further advancements in BC detection methods.