Home // AIHealth 2025, The Second International Conference on AI-Health // View article
Authors:
Caitlin Dosch
Shilpi Shaw
Keywords: breast lesions; deep learning; image classification
Abstract:
Breast cancer remains a global health concern, with a 13.1% lifetime diagnosis rate among women. Early and accurate diagnosis plays a critical role in improving patient outcomes. Traditional diagnostic methods, such as MRIs, ultrasounds, CT scans, and mammograms, are widely used for detecting and characterizing breast lesions. In recent years, Artificial Intelligence has shown great promise in enhancing diagnostic accuracy, with models such as K-Nearest Neighbors (KNN), Random Forest Classifier (RFC), and Convolutional Neural Networks (CNN) being applied to breast cancer diagnosis. In this study, we explore the application of deep learning models, specifically MobileNetV2 and ResNet50, for breast cancer detection using ultrasound images from The Cancer Image Archive. A dataset comprising 522 breast lesion images was used, split into training, validation, and test sets. We implemented both image classification and segmentation tasks, optimizing hyperparameters such as learning rate and number of epochs. Our comparative analysis aims to evaluate the efficiency and diagnostic performance of the two models. We highlight key insights into their effectiveness in breast cancer detection and provide recommendations based on their application to ultrasound imaging. The findings of this study contribute to the ongoing efforts to improve AI-based diagnostic tools for breast cancer.
Pages: 61 to 67
Copyright: Copyright (c) IARIA, 2025
Publication date: March 9, 2025
Published in: conference
ISBN: 978-1-68558-247-0
Location: Lisbon, Portugal
Dates: from March 9, 2025 to March 13, 2025