Previous research has established that breast cancer is a prevalent cancer type, with Invasive Ductal Carcinoma (IDC) being the most common subtype. The incidence of this dangerous cancer continues to rise, making accurate and rapid diagnosis, particularly in the early stages, critically important. While modern Computer-Aided Diagnosis (CAD) systems can address most cases, medical professionals still face challenges in quickly adapting CAD systems or using them in the field without powerful computing resources. In this paper, we enhance the traditional Convolutional Neural Network (CNN) architecture by integrating Batch Normalization and Dropout layers, tailoring the model to meet the specific demands of IDC detection. Furthermore, we introduce a novel CNN called BCDNet, which effectively detects IDC in histopathological images with an accuracy of up to 89.5% and reduces training time by up to 82.1%.
Download the code and install the dependencies, for which conda environment is recommended.
git clone https://github.com/404-UnknownUsername/BCDNet
conda create -n bcdnet python==3.8.19
conda install pytorch torchvision pytorch-cuda=12.1 -c pytorch -c nvidia
pip install -r requirements.txt
Attention: If your CUDA version is earlier than 12.1, please visit PyTorch to find the corresponding installation command.
The Breast Histopathology Images Dataset can be downloaded from kaggle. Then, you should store it in the data folder under the BCDNet folder.
To train BCDNet on your devices, you can use
python train.py
Remember to change the configuration in train.py based on your requirements and devices.
To test the model, you can use
python test.py
Our test results are as follows:
BCDNet:
ResNet 50:
ViT-B-16:
