Research Journal of Medical Sciences
ISSN: Online 1993-6095ISSN: Print 1815-9346
Abstract
Breast cancer is a major global health concern, necessitating early detection and accurate diagnosis of breast lesions for optimal patient outcomes. In this project, we propose an automated breast lesion detection and classification system that uses ultrasound images and transfer learning with the Efficient Net B7 model. The model was trained on the Breast Ultrasound Images Dataset (BUSI), which consists of 780 ultrasound images of the breast. Preprocessing was carried out in the dataset by removing any mask images to ensure data quality. Our system achieved a remarkable classification accuracy of 98% for breast lesions, categorizing them as benign, malignant, or normal. A user‐friendly web application was developed that provides a quick and accurate classification of breast lesions based on ultrasound images. Medical professionals can easily access the web app, reducing their workload and improving patient care. Our proposed model showcases the potential of transfer learning and the Efficient Net B7 model for accurate and efficient breast lesion diagnosis using ultrasound images. The web app we created has the potential to revolutionize breast lesion diagnosis and improve clinical decision‐making.
How to cite this article:
T. Mangayarkarasi and S. Gowri Manohari. Enhancing Breast Lesion Assessment Using Efficient Net B7 Transfer Learning Models.
DOI: https://doi.org/10.36478/10.59218/makrjms.2023.411.416
URL: https://www.makhillpublications.co/view-article/1815-9346/10.59218/makrjms.2023.411.416