MENet: A Mitscherlich function based ensemble of CNN models to classify lung cancer using CT scans

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Oncology
• Computational Pathology

Background:

• Lung cancer remains a leading cause of global cancer mortality.
• Early detection and accurate diagnosis are crucial for reducing lung cancer death rates.
• Computer-aided diagnosis (CADx) systems enhance diagnostic precision using medical image analysis.

Purpose of the Study:

• To develop an advanced ensemble model for improved lung cancer prediction accuracy.
• To integrate multiple deep learning models using a novel ensemble strategy.
• To enhance early lung cancer detection capabilities through AI.

Main Methods:

• Proposed an ensemble model named Mitscherlich function-based Ensemble Network (MENet).
• Combined prediction probabilities from Xception, InceptionResNetV2, and MobileNetV2 deep learning models.
• Utilized the Mitscherlich function for fuzzy rank-based combination of base classifier outputs.
• Trained and validated the model on two public CT scan datasets: IQ-OTH/NCCD and LIDC-IDRI.

Main Results:

• The proposed MENet model demonstrated superior performance compared to existing state-of-the-art methods.
• Achieved improved accuracy in lung cancer prediction based on standard evaluation metrics.
• The ensemble approach effectively leveraged the strengths of individual deep learning models.

Conclusions:

• MENet offers a promising approach for enhancing lung cancer diagnostic accuracy.
• The Mitscherlich function-based ensemble strategy effectively integrates deep learning models.
• This AI-driven method has the potential to improve early lung cancer detection and patient outcomes.