Multi-model Ensemble Learning Architecture Based on 3D CNN for Lung Nodule Malignancy Suspiciousness Classification.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Accurate classification of lung nodules in CT images is crucial for early lung cancer diagnosis and improving patient survival rates.
• Conventional deep learning methods struggle with nodule diversity and visual similarity to surrounding tissues, hindering robust classification.
• Developing advanced diagnostic models is essential to overcome these limitations in lung cancer detection.

Purpose of the Study:

• To propose a robust multi-model ensemble learning architecture based on 3D convolutional neural networks (MMEL-3DCNN) for classifying benign and malignant lung nodules.
• To address the challenges posed by nodule heterogeneity and improve the discriminative capacity of feature extraction.
• To enhance the generalization ability and robustness of lung nodule classification models.

Main Methods:

• Developed a multi-model ensemble learning architecture (MMEL-3DCNN) utilizing 3D convolutional neural networks.
• Incorporated concatenated input images (intensity, original, enhanced) for advanced feature extraction.
• Implemented dynamic model selection based on nodule size for improved prediction and generalization.

Main Results:

• The MMEL-3DCNN architecture demonstrated effectiveness in classifying lung nodules.
• Experimental validation on the LIDC-IDRI dataset yielded satisfactory classification results.
• The ensemble learning approach significantly improved the robustness of the classification model.

Conclusions:

• The proposed MMEL-3DCNN architecture offers a promising solution for accurate lung nodule classification.
• This approach effectively handles nodule heterogeneity and enhances feature extraction capabilities.
• The study highlights the potential of ensemble learning in improving the reliability of AI-driven lung cancer diagnosis.