A Novel Hybrid Feature Extraction Model for Classification on Pulmonary Nodules

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Early diagnosis of pulmonary nodules is crucial for effective lung cancer treatment.
• Computer Tomography (CT) imaging is a primary tool for detecting lung abnormalities.
• Existing Computer Aided Design (CAD) systems require improvement for enhanced diagnostic accuracy.

Purpose of the Study:

• To develop an improved CAD system for assisting radiologists in the early diagnosis of pulmonary nodules.
• To enhance the accuracy of pulmonary nodule detection and classification using advanced AI techniques.

Main Methods:

• A Deep Convolutional Neural Network (DCNN) was employed for feature extraction.
• Hybrid feature extraction combined Convolutional Neural Network (CNN), Histogram of Oriented Gradients (HOG), Extended Histogram of Oriented Gradients (ExHOG), and Local Binary Pattern (LBP).
• Features including shape, texture, scaling, rotation, and translation were extracted and fed into classifiers like Support Vector Machine (SVM), K-Nearest Neighbour (KNN), Decision Tree, and Random Forest.

Main Results:

• The proposed DCNN-based CAD system demonstrated superior performance in classifying pulmonary nodules and non-nodules.
• Experimental results indicated that the developed method achieved higher accuracy compared to competing approaches.
• The combination of diverse features (shape, texture, etc.) contributed to improved classification efficacy.

Conclusions:

• The improved CAD system offers a valuable second opinion for radiologists in early pulmonary nodule diagnosis.
• The hybrid DCNN approach effectively extracts and utilizes relevant features for accurate nodule classification.
• This AI-driven system shows significant potential for improving diagnostic outcomes in lung imaging.