An integrated convolutional neural network for classifying small pulmonary solid nodules.

Area of Science:

• Medical imaging analysis
• Artificial intelligence in healthcare
• Pulmonary nodule classification

Background:

• Accurate classification of benign and malignant pulmonary nodules is crucial for disease treatment.
• Traditional methods struggle with small pulmonary solid nodules due to noise and feature loss from downsampling.

Purpose of the Study:

• To develop an improved diagnostic method for small pulmonary solid nodules in CT images.
• To enhance the accuracy of pulmonary nodule classification using advanced computational techniques.

Main Methods:

• Implemented Otsu thresholding algorithm for data preprocessing and noise reduction.
• Integrated parallel radiomics with a 3D convolutional neural network to capture detailed nodule features.
• Utilized a classifier combining visual and radiomic features for improved diagnostic accuracy.

Main Results:

• The proposed method demonstrated superior performance in classifying small pulmonary solid nodules compared to existing methods across multiple datasets.
• Ablation experiments confirmed the significant contribution of Otsu thresholding and radiomics to the improved classification accuracy.
• Otsu thresholding algorithm showed greater flexibility than manual thresholding in this application.

Conclusions:

• The novel approach effectively addresses challenges in small pulmonary nodule classification, offering a more accurate diagnostic tool.
• The combination of image preprocessing, radiomics, and deep learning provides a robust framework for medical image analysis.
• This method holds promise for improving early diagnosis and treatment planning for pulmonary nodules.