Lung and colon cancer classification using medical imaging: a feature engineering approach

Area of Science:

• Oncology
• Medical Imaging
• Artificial Intelligence

Background:

• Lung and colon cancers are leading causes of death, with high metastasis risk when not diagnosed early.
• Accurate histopathological diagnosis is crucial for improving survival rates and reducing mortality.
• Artificial intelligence (AI) offers potential to assist specialists in cancer diagnosis, reducing effort, time, and cost.

Purpose of the Study:

• To develop a computer-aided diagnostic system for classifying five types of colon and lung cancer tissues.
• To analyze histopathological images for accurate cancer subtype identification.

Main Methods:

• Utilized machine learning, feature engineering, and image processing techniques.
• Evaluated six models: XGBoost, Support Vector Machine (SVM), Random Forest (RF), Linear Discriminant Analysis (LDA), Multilayer Perceptron (MLP), and LightGBM.
• Employed the LC25000 dataset for training and testing classification models.

Main Results:

• Machine learning models demonstrated satisfactory precision in identifying lung and colon cancer subtypes.
• The XGBoost model achieved the highest performance with 99% accuracy and a 98.8% F1-score.
• Feature-engineered machine learning models offered better interpretability compared to deep learning 'black box' networks.

Conclusions:

• The developed AI system, particularly the XGBoost model, accurately classifies lung and colon cancer subtypes from histopathological images.
• This system can significantly aid healthcare specialists in identifying cancer types, potentially improving patient outcomes.
• The study highlights the effectiveness of machine learning in computational pathology for cancer diagnosis.