A Modiﬁed LeNet CNN for Breast Cancer Diagnosis in Ultrasound Images - read paper summary

Aug 5, 2026

Convolutional neural networks support automated medical image feature extraction and classification for faster, more accurate diagnosis. This paper applies LeNet, with a corrected ReLU variant, to breast cancer analysis using ultrasound images, improving performance by addressing the “dying ReLU” effect and increasing feature discriminability. Batch normalization further enhances training stability and reduces internal covariate shift. The resulting classifier mitigates overfitting and lowers training time, achieving 89.91% recognition accuracy and improved detection capability.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages28

Reading Time1 h 11 min

AccessFree to download

Introduction

Breast cancer and imaging modalities

Computer-aided diagnosis pipeline

Abstract (model overview)

Method and improvements

Corrected ReLU and dying ReLU mitigation

Batch normalization for stability and efficiency

diagnostics Article A Modiﬁed LeNet CNN for Breast Cancer Diagnosis in Ultrasound Images Sathiyabhama Balasubramaniam 1,*, Yuvarajan Velmurugan 1, Dhayanithi Jaganathan 1 and Seshathiri Dhanasekaran 2, * Citation: Balasubramaniam, S.; Velmurugan, Y.; Jaganathan, D.; Dhanasekaran, S. A Modiﬁed LeNet CNN for Breast Cancer Diagnosis in Ultrasound Images. Diagnostics 2023, 13, 2746. [https://doi.org/10.3390/](https://doi.org/10.3390/)[ ](https://doi.org/10.3390/)diagnostics13172746 Academic Editor: Dechang Chen Received: 19 April 2023 Revised: 6 July 2023 Accepted: 11 July 2023 Published: 24 August 2023 Copyright: © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license ([https://](https://)[ ](https://)[creativecommons.org/licenses/by/](creativecommons.org/licenses/by/)[ ](creativecommons.org/licenses/by/)[4.0/](4.0/)) . 1 Computer Science and Engineering, Sona College of Technology, Salem 636005, India; [yuvi.cn@gmail.com](yuvi.cn@gmail.com) (Y.V.); [dhaya.j@sonatech.ac.in](dhaya.j@sonatech.ac.in) (D.J.) 2 Department of Computer Science, UiT The Arctic University of Norway, 9037 Tromso, Norway * Correspondence: [sathiyabhama@sonatech.ac.in](sathiyabhama@sonatech.ac.in) (S.B.); [seshathiri.dhanasekaran@uit.no](seshathiri.dhanasekaran@uit.no) (S.D.) Abstract: Convolutional neural networks (CNNs) have been extensively utilized in medical image processing to automatically extract meaningful features and classify various medical conditions, enabling faster and more accurate diagnoses. In this paper, LeNet, a classic CNN architecture, has been successfully applied to breast cancer data analysis. It demonstrates its ability to extract discriminative features and classify malignant and benign tumors with high accuracy, thereby supporting early detection and diagnosis of breast cancer. LeNet with corrected Rectiﬁed Linear Unit (ReLU), a modiﬁcation of the traditional ReLU activation function, has been found to improve the performance of LeNet in breast cancer data analysis tasks via addressing the “dying ReLU” problem and enhancing the discriminative power of the extracted features. This has led to more accurate, reliable breast cancer detection and diagnosis and improved patient outcomes. Batch normalization improves the performance and training stability of small and shallow CNN architecture like LeNet. It helps to mitigate the effects of internal covariate shift, which refers to the change in the distribution of network activations during training. This classiﬁer will lessen the overﬁtting problem and reduce the running time. The designed classiﬁer is evaluated against the benchmarking deep learning models, proving that this has produced a higher recognition rate. The accuracy of the breast image recognition rate is 89.91% . This model will achieve better performance in segmentation, feature extraction, classiﬁcation, and breast cancer tumor detection. Keywords: deep learning; breast cancer; convolutional neural networks; LeNet; medical image processing; batch normalization 1. Introduction Breast cancer is one of the most common causes of death in women worldwide. Early detection saves the life of many women. Healthcare practitioners extensively use mammograms for screening breast cancer. Breast ultrasound and diagnostic mammogram are the two imaging tests that are used to evaluate breast tissue for abnormalities or signs of breast cancer. While both are imaging studies of the breast, they differ in how they use sound and X-rays to obtain images and the information they provide [1] . A diagnostic mammogram is a low-dose X-ray that provides detailed images of the breast tissue. It can detect calciﬁcations, masses, and abnormalities indicating cancer or other conditions. Diagnostic mammograms are typically recommended for women with a breast lump, nipple discharge, a

FAQs

How does the modified LeNet improve breast cancer diagnosis from ultrasound images?

The model uses a corrected ReLU activation to address the “dying ReLU” problem, improving discriminative feature extraction for malignant vs. benign tumors.

What role does batch normalization play in the proposed classifier?

Batch normalization improves training stability and mitigates internal covariate shift, which helps reduce overfitting and training time in LeNet-like networks.

Which accuracy is reported for breast image recognition?

The paper reports an accuracy of 89.91% for breast image recognition using the designed classifier.

Aug 5, 2026

Upload by Olivia Brown