-
Notifications
You must be signed in to change notification settings - Fork 203
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge pull request #1108 from inkerton/main
COVID Detection from CXR Using Explainable CNN
- Loading branch information
Showing
3 changed files
with
102 additions
and
0 deletions.
There are no files selected for viewing
99 changes: 99 additions & 0 deletions
99
Explainable-AI/Covid_Detection_from_Explainable_cnn/Readme.md
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,99 @@ | ||
# COVID-19 Chest X-Ray Classification using Depthwise Separable Convolutional Neural Networks (CNN) | ||
|
||
This project focuses on building a convolutional neural network (CNN) model to classify Chest X-Ray (CXR) images into three categories: **COVID-19**, **Viral Infection**, and **Normal**. The model uses image enhancement techniques like **White Balance** and **CLAHE** (Contrast Limited Adaptive Histogram Equalization) for better image processing before classification. | ||
|
||
## Dataset | ||
|
||
The dataset used in this project consists of Chest X-Ray (CXR) images classified into three categories: | ||
- **COVID-19**: Images of chest X-rays from COVID-19 patients. | ||
- **Viral Infection**: Images of chest X-rays from patients with other viral infections. | ||
- **Normal**: Images of chest X-rays from healthy individuals. | ||
|
||
The dataset is stored in the following directories: | ||
- `/kaggle/input/covid-cxr-image-dataset-research/COVID_IEEE/covid/` for COVID-19 images. | ||
- `/kaggle/input/covid-cxr-image-dataset-research/COVID_IEEE/normal/` for normal images. | ||
- `/kaggle/input/covid-cxr-image-dataset-research/COVID_IEEE/virus/` for viral infection images. | ||
|
||
## Project Setup | ||
|
||
1. **Dependencies**: The project requires several Python libraries such as `numpy`, `cv2`, `matplotlib`, `PIL`, `keras`, and `sklearn`. These can be installed via pip: | ||
```bash | ||
pip install numpy opencv-python matplotlib Pillow scikit-learn tensorflow | ||
``` | ||
|
||
2. **Image Enhancement**: The dataset images are enhanced using **White Balance** and **CLAHE** to improve the clarity of the Chest X-rays for better feature extraction. | ||
|
||
3. **Model Architecture**: The model uses **Depthwise Separable Convolutional Neural Networks (CNN)**, which reduces the number of parameters while maintaining high performance for image classification tasks. | ||
|
||
## Data Preprocessing | ||
|
||
1. **White Balance**: Each image channel is processed by adjusting the intensity to standardize the color across images. | ||
```python | ||
def wb(channel, perc=0.05): | ||
mi, ma = (np.percentile(channel, perc), np.percentile(channel, 100.0-perc)) | ||
channel = np.uint8(np.clip((channel - mi) * 255.0 / (ma - mi), 0, 255)) | ||
return channel | ||
``` | ||
2. **CLAHE (Contrast Limited Adaptive Histogram Equalization)**: This technique is applied to enhance the contrast in images, especially useful in medical imaging. | ||
```python | ||
clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(16, 16)) | ||
img_clahe1 = clahe.apply(gray_image) | ||
``` | ||
3. **Resizing**: All images are resized to a fixed size of `224x224` pixels to ensure uniformity when feeding them into the model. | ||
4. **Normalization**: The pixel values of images are normalized by dividing by 255.0 to scale the values between 0 and 1. | ||
## Model Architecture | ||
The model architecture includes several convolutional layers, both traditional and depthwise separable, followed by dense layers for final classification: | ||
- **Conv2D Layers**: Initial convolution layers to extract features. | ||
- **MaxPooling Layers**: To reduce the spatial dimensions. | ||
- **SeparableConv2D Layers**: Depthwise separable convolutions to reduce parameters. | ||
- **BatchNormalization**: To normalize activations and improve training stability. | ||
- **Dropout**: To prevent overfitting. | ||
- **Dense Layers**: For classification of the images into three categories. | ||
```python | ||
inputs = Input(shape=(224, 224, 3)) | ||
x = Conv2D(filters=4, kernel_size=(3, 3), activation='relu', padding='same')(inputs) | ||
x = MaxPool2D(pool_size=(2, 2))(x) | ||
... | ||
output = Dense(units=3, activation='softmax')(x) | ||
model = Model(inputs=inputs, outputs=output) | ||
``` | ||
## Training and Evaluation | ||
1. **Model Compilation**: The model is compiled using the Adam optimizer and sparse categorical cross-entropy loss function. | ||
```python | ||
model.compile(loss="sparse_categorical_crossentropy", optimizer="adam", metrics=["accuracy"]) | ||
``` | ||
2. **Callbacks**: A `ModelCheckpoint` callback is used to save the best model based on validation loss during training. | ||
3. **Train-Test Split**: The data is split into training and testing sets using `train_test_split` from `sklearn`. | ||
4. **Data Augmentation**: An augmentation generator is used to artificially increase the size of the training data by applying random transformations. | ||
```python | ||
trainAug = ImageDataGenerator(rotation_range=15, fill_mode="nearest") | ||
``` | ||
## Results and Observations | ||
- **Normal CXR Images**: Clear lung patterns. | ||
- **Viral Infection CXR Images**: Slight congestion in the lungs. | ||
- **COVID-19 CXR Images**: Serious lung congestion. | ||
## Conclusion | ||
This project demonstrates how CNNs, particularly Depthwise Separable CNNs, can be used effectively to classify Chest X-ray images into categories such as COVID-19, Viral Infection, and Normal. The image enhancement techniques of White Balance and CLAHE significantly improve the quality of the input images, contributing to better model performance. | ||
## Future Work | ||
- Fine-tuning hyperparameters. | ||
- Exploring more advanced CNN architectures. | ||
- Integrating the model into a real-world application for automated CXR analysis. | ||
1 change: 1 addition & 0 deletions
1
...Covid_Detection_from_Explainable_cnn/covid-detection-from-cxr-using-explainable-cnn.ipynb
Large diffs are not rendered by default.
Oops, something went wrong.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters