An Enhanced LightGBM-Based Breast Cancer Detection Technique Using Mammography Images

Abdul Rahaman Wahab Sait,Ramprasad Nagaraj

DOI: https://doi.org/10.3390/diagnostics14020227

IF: 3.6

2024-01-23

Diagnostics

Abstract:Breast cancer (BC) is the leading cause of mortality among women across the world. Earlier screening of BC can significantly reduce the mortality rate and assist the diagnostic process to increase the survival rate. Researchers employ deep learning (DL) techniques to detect BC using mammogram images. However, these techniques are resource-intensive, leading to implementation complexities in real-life environments. The performance of convolutional neural network (CNN) models depends on the quality of mammogram images. Thus, this study aimed to build a model to detect BC using a DL technique. Image preprocessing techniques were used to enhance image quality. The authors developed a CNN model using the EfficientNet B7 model's weights to extract the image features. Multi-class classification of BC images was performed using the LightGBM model. The Optuna algorithm was used to fine-tune LightGBM for image classification. In addition, a quantization-aware training (QAT) strategy was followed to implement the proposed model in a resource-constrained environment. The authors generalized the proposed model using the CBIS-DDSM and CMMD datasets. Additionally, they combined these two datasets to ensure the model's generalizability to diverse images. The experimental findings revealed that the suggested BC detection model produced a promising result. The proposed BC detection model obtained an accuracy of 99.4%, 99.9%, and 97.0%, and Kappa (K) values of 96.9%, 96.9%, and 94.1% in the CBIS-DDSM, CMMD, and combined datasets. The recommended model streamlined the BC detection process in order to achieve an exceptional outcome. It can be deployed in a real-life environment to support physicians in making effective decisions. Graph convolutional networks can be used to improve the performance of the proposed model.

medicine, general & internal

What problem does this paper attempt to address?

The paper aims to address several key issues in breast cancer detection:

1. **Improving detection efficiency and accuracy**: Existing breast cancer detection technologies, especially those based on deep learning methods, face resource-intensive issues in practical applications, leading to increased implementation complexity. Additionally, the effectiveness of these methods often depends on the quality of the images.

2. **Applicability in resource-limited environments**: In resource-limited environments, complex deep learning models may be difficult to deploy and run. Therefore, researchers are committed to developing a lightweight yet efficient breast cancer detection model.

3. **Model generalization ability**: To improve the model's generalization ability, researchers combined two different datasets (CBIS-DDSM and CMMD) to ensure that the model can adapt to diverse image data.

In summary, the main goal of this study is to construct a deep learning-based breast cancer detection model that can operate efficiently in resource-limited environments while maintaining high detection accuracy. By employing techniques such as image preprocessing, enhancement, feature extraction, and optimized classification algorithms, the research team aims to overcome the challenges present in traditional breast cancer detection methods and propose a more reliable and efficient solution.