MRI Brain Classification Using the Quantum Entropy LBP and Deep-Learning-Based Features

Area of Science:

• Medical Imaging Analysis
• Quantum Calculus Applications
• Machine Learning in Healthcare

Background:

• Early brain tumor detection significantly improves patient prognosis.
• Medical imaging technologies are integral to modern diagnosis and treatment.
• Existing feature extraction methods can be enhanced for better classification.

Purpose of the Study:

• To generalize entropy difference using quantum calculus.
• To extend Local Binary Patterns (LBP) into Quantum Entropy LBP (QELBP).
• To improve MRI brain scan classification accuracy by combining QELBP and deep learning features.

Main Methods:

• Generalization of entropy difference using quantum calculus.
• Development of Quantum Entropy Local Binary Patterns (QELBP) feature extraction.
• Integration of QELBP and deep learning (DL) features for MRI brain scan classification.
• Utilizing a long short-term memory (LSTM) network for classification.

Main Results:

• A novel QELBP feature extraction method was proposed.
• Combining QELBP and DL features enhanced classification performance.
• The combined feature approach achieved a maximum accuracy of 98.80% on a dataset of 154 MRI brain scans.
• Experimental results validated the effectiveness of feature combination.

Conclusions:

• The proposed QELBP-DL approach significantly improves MRI brain tumor classification accuracy.
• Combining extracted features is a viable strategy for enhancing diagnostic performance.
• This method holds promise for earlier and more accurate brain tumor detection.