Related Experiment Video
Updated: May 11, 2025
04:23
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
1.7K
Transformer-inspired training principles based breast cancer prediction: combining EfficientNetB0 and ResNet50
Tariq Shahzad1, Tehseen Mazhar2,3, Sheikh Muhammad Saqib4
1Department of Electrical and Electronic Engineering Science, University of Johannesburg, Johannesburg, 2006, South Africa. tariqshahzadd@gmail.com.
Scientific Reports
|April 18, 2025
Summary
A new hybrid deep learning model combining EfficientNetB0 and ResNet50 significantly improves breast cancer diagnosis from histopathology images. This accurate tool enhances early detection, addressing critical needs in cancer screening.
Area of Science:
- Medical Imaging
- Computational Pathology
- Artificial Intelligence in Oncology
Background:
- Breast cancer remains a leading cause of mortality, with diagnosis and treatment services impacted by global health crises like COVID-19.
- The need for rapid, efficient, and accurate diagnostic tools for breast cancer is critical, yet current machine learning approaches face challenges in diagnostic accuracy.
Purpose of the Study:
- To develop and evaluate a novel hybrid deep learning model for improved classification of breast histopathology images into invasive ductal carcinoma (IDC) and non-IDC categories.
- To enhance diagnostic accuracy and efficiency in breast cancer screening through advanced computational methods.
Main Methods:
- A hybrid model was developed by combining EfficientNetB0 and ResNet50 architectures.
- Histopathology images were preprocessed, including resizing to 128*128 pixels and normalization, to optimize model performance.
- The model leveraged EfficientNetB0's efficiency and ResNet50's deep residual connections to address vanishing gradients and improve classification.
Main Results:
- The proposed model achieved a high accuracy of 94% in classifying breast histopathology images.
- Performance metrics included a Mean Absolute Error (MAE) of 0.0628 and a Matthews Correlation Coefficient (MCC) of 0.8690.
- The model demonstrated superior performance compared to previous baselines, balancing precision and recall effectively.
Conclusions:
- The developed hybrid EfficientNetB0-ResNet50 model offers a resilient and accurate solution for breast cancer diagnosis.
- This ensemble approach shows superiority in accuracy and computational efficiency, making it suitable for practical breast cancer screening and diagnosis.
