🧬 Lung Cancer Prediction using CNN and Transfer Learning

Deep Learning • Medical Image Classification • Transfer Learning • Computer Vision

Author: K. Siddhartha

This project aims to build a Lung Cancer Prediction System using Convolutional Neural Networks (CNN) and transfer learning. The model classifies lung cancer images into four categories:

• Normal
• Adenocarcinoma
• Large Cell Carcinoma
• Squamous Cell Carcinoma

---

📑 Table of Contents

• Introduction
• Dataset
• Dependencies
• Project Structure
• Training the Model
• Using the Model
• Results
• Acknowledgements
• License

---

🧠 Introduction

Lung cancer is one of the leading causes of cancer-related deaths worldwide. Early detection and accurate classification are crucial for effective treatment and patient survival. This project leverages deep learning techniques to develop a robust lung cancer classification model using chest X-ray images.

---

📊 Dataset

The dataset used in this project consists of lung cancer images categorized into four classes:

• Normal
• Adenocarcinoma
• Large Cell Carcinoma
• Squamous Cell Carcinoma

The dataset should be organized into training (train), validation (valid), and testing (test) folders with the following subfolders for each class:

train/
 ├── normal/
 ├── adenocarcinoma/
 ├── large_cell_carcinoma/
 └── squamous_cell_carcinoma/

valid/
 ├── normal/
 ├── adenocarcinoma/
 ├── large_cell_carcinoma/
 └── squamous_cell_carcinoma/

test/
 ├── normal/
 ├── adenocarcinoma/
 ├── large_cell_carcinoma/
 └── squamous_cell_carcinoma/

Alternatively, you can download a similar dataset from Kaggle:

https://www.kaggle.com/datasets/mohamedhanyyy/chest-ctscan-images

---

▶ Google Colab Link

To replicate and run the project in Google Colab:

https://colab.research.google.com/drive/1kMTghEwVoJaFmlKydxuhhoyzHluIUjoV?usp=sharing

---

⚙ Dependencies

Required libraries:

• Python 3.x
• pandas
• numpy
• seaborn
• matplotlib
• scikit-learn
• tensorflow
• keras

Install dependencies:

pip install pandas numpy seaborn matplotlib scikit-learn tensorflow keras

---

📁 Project Structure

.
├── Lung_Cancer_Prediction.ipynb
├── README.md
├── dataset/
│ ├── train/
│ ├── test/
│ └── valid/
└── best_model.hdf5

File Description

• Lung_Cancer_Prediction.ipynb — Notebook for training and evaluation
• dataset/ — Image dataset grouped by cancer type
• best_model.hdf5 — Saved trained model weights

---

🏗 Training the Model

The Jupyter Notebook contains the full training pipeline:

• Mount Google Drive
• Load and preprocess data using ImageDataGenerator
• Define the model using Xception transfer learning
• Compile the model
• Train with callbacks (EarlyStopping, LR scheduler, checkpoints)
• Save the trained model

Example Code

pretrained_model = tf.keras.applications.Xception(
    weights='imagenet',
    include_top=False,
    input_shape=[*IMAGE_SIZE, 3]
)
pretrained_model.trainable = False

model = Sequential([
    pretrained_model,
    GlobalAveragePooling2D(),
    Dense(4, activation='softmax')
])

model.compile(
    optimizer='adam',
    loss='categorical_crossentropy',
    metrics=['accuracy']
)

---

🔬 Using the Model

Steps:

• Load trained .h5 model
• Preprocess input image
• Predict class probabilities
• Display prediction

Example

model = load_model('/content/drive/MyDrive/dataset/trained_lung_cancer_model.h5')
predictions = model.predict(img)
predicted_class = np.argmax(predictions[0])

---

📈 Results

After training and evaluation:

• Final model accuracy: 93%

The model demonstrates strong performance in classifying lung cancer categories using transfer learning.

---

🙏 Acknowledgements

Chest CT Scan Images Dataset:

https://www.kaggle.com/datasets/mohamedhanyyy/chest-ctscan-images

---

📜 License

This project is licensed under the MIT License.

Feel free to use or modify this code for educational and non-commercial purposes.