Skin Cancer Classification using HAM10000 Dataset

Overview

This project is focused on classifying different types of skin cancer using the HAM10000 dataset from Kaggle. The dataset consists of 10,015 dermatoscopic images belonging to 7 different classes. The goal of this project is to develop a machine learning model that can accurately classify skin cancer types using convolutional neural networks (CNNs).

Dataset

The HAM10000 dataset contains images of skin lesions categorized into the following 7 classes:

1. Actinic keratoses (akiec)
2. Basal cell carcinoma (bcc)
3. Benign keratosis-like lesions (bkl)
4. Dermatofibroma (df)
5. Melanoma (mel)
6. Melanocytic nevi (nv)
7. Vascular lesions (vasc)

I utilized the hmnist_28_28_RGB.csv file, which contains 28x28 pixel representations of the images, to preprocess and train our model.

Model Architecture

The model consists of a fully connected neural network with dropout regularization. The architecture is as follows:

• Input Layer (28x28 images)
• Flatten Layer (Converts 2D images into a 1D vector)
• Dense Layer (128 units, ReLU activation)
• Dropout Layer (0.5 dropout rate to prevent overfitting)
• Output Layer (7 units, softmax activation)

Model Compilation

model = tf.keras.Sequential([
    tf.keras.layers.InputLayer(shape=(28, 28)), 
    tf.keras.layers.Flatten(), 
    tf.keras.layers.Dense(128, activation='relu'),
    tf.keras.layers.Dropout(0.5), 
    tf.keras.layers.Dense(7, activation='softmax')
])

model.compile(loss='sparse_categorical_crossentropy',
              optimizer=Adam(learning_rate=0.0001),
              metrics=['accuracy'])

Training Strategy

To improve model performance, early stopping was implemented to halt training if validation loss does not improve for three consecutive epochs.

early_stopping = EarlyStopping(
    monitor='val_loss',  
    patience=3,         
    restore_best_weights=True,  
    verbose=1            
)

history = model.fit(X_train_resampled, Y_train_resampled, 
                    validation_split=0.2, 
                    batch_size=128, 
                    epochs=20, 
                    callbacks=[early_stopping])

Training Details

• Learning Rate: 0.0001
• Loss Function: Sparse Categorical Crossentropy
• Optimizer: Adam
• Batch Size: 128
• Epochs: 20 (early stopping enabled)

Evaluation Metrics

To assess the model's performance, we used:

• Accuracy: Measures overall correctness
• Confusion Matrix: Visualizes misclassification patterns

Results & Improvements

The model showed promising results, but there is still room for improvement. Some of the key areas for future work include:

• Data Augmentation: Enhancing the dataset with transformations to improve generalization.
• Hyperparameter Tuning: Experimenting with different learning rates, batch sizes, and optimizers.
• Transfer Learning: Utilizing pre-trained models like ResNet or VGG for better feature extraction.

References

• HAM10000 Dataset

License

This project is open-source under the MIT License.