LNTP-MDBN: Big Data Integrated Learning Framework for Heterogeneous Image Set Classification.

Area of Science:

• Computer Science
• Medical Imaging
• Data Science

Background:

• Big Data analytics faces challenges in volume, variety, and velocity.
• Medical image processing requires effective image set classification.
• Multi-directional image views necessitate advanced analysis techniques.

Purpose of the Study:

• To propose Local N-ary Ternary Patterns (LNTP) and Modified Deep Belief Network (MDBN) for medical image analysis.
• To address dimensionality and robustness issues in big data analytics for images.
• To improve the accuracy and efficiency of medical image classification.

Main Methods:

• Utilizes filtering techniques to remove image noise.
• Applies smoothening and normalization to enhance image intensity.
• Employs LNTP for heterogeneous image categorization and feature extraction.
• Uses Modified Deep Belief Network (MDBN) for classifying normal and abnormal image categories.

Main Results:

• LNTP categorizes heterogeneous images and extracts relevant features.
• MDBN effectively classifies image sets into normal and abnormal categories.
• The proposed LNTP-MDBN method demonstrates improved classification accuracy.

Conclusions:

• The LNTP-MDBN approach effectively alleviates dimensionality and robustness issues in medical image analysis.
• Comparative analysis confirms the effectiveness of LNTP-MDBN over existing models.
• The study highlights the potential of big data analytics in medical image processing.