Efficient 3D AlexNet Architecture for Object Recognition Using Syntactic Patterns from Medical Images

Area of Science:

• Computer Vision
• Medical Image Processing
• Artificial Intelligence

Background:

• Object recognition in medical images, particularly brain tumors, is crucial for diagnosis.
• Human visual recognition is rapid, motivating the development of efficient computer-based methods.
• Current methods require enhancement for accurate pattern recognition in noisy medical images.

Purpose of the Study:

• To develop a computer-specific pattern recognition method for identifying objects in medical images, specifically brain tumors.
• To enhance the accuracy and efficiency of brain tumor classification using advanced algorithms.
• To improve upon existing models for brain tumor detection and classification.

Main Methods:

• Utilized an adaptive median filter for noise reduction in MRI images.
• Applied contrast image enhancement techniques to improve image quality.
• Employed a cellular logic array processing (CLAP)-based algorithm for wireframe model evaluation and 3D pattern identification.
• Integrated syntactic pattern recognition for feature vector extraction and 3D AlexNet for brain tumor classification.

Main Results:

• Successfully identified basic patterns in 3D medical images.
• Achieved object classification based on pattern frequency.
• Demonstrated superior performance of the proposed 3D AlexNet model for brain tumor classification.
• Validated the model using benchmark datasets: Figshare, Brain MRI Kaggle, Medical MRI, and BraTS 2019.

Conclusions:

• The proposed syntactic pattern recognition and 3D AlexNet model offers a highly effective approach for brain tumor classification.
• The method significantly enhances accuracy compared to existing models.
• This work advances computer vision applications in medical image analysis for improved diagnostic tools.