RETRACTED: A novel spectral transformation technique based on special functions for improved chest X-ray image...

Area of Science:

• Medical Imaging Analysis
• Machine Learning in Healthcare
• Signal Processing

Background:

• Chest X-ray classification is vital for medical diagnostics.
• Machine learning has advanced diagnostic capabilities.
• Data transformation into reduced feature spaces is common in classification.

Purpose of the Study:

• To propose a new chest X-ray image classification method using spectral coefficients.
• To develop a mechanism for converting medical images into a spectral space.
• To evaluate the effectiveness of spectral moments in classifying X-ray images for different patient conditions.

Main Methods:

• Developed a method to convert medical images into a spectral space using Legendre type smooth polynomials.
• Calculated spectral moments in the Legendre polynomial space.
• Trained Support Vector Machine (SVM) and Random Forest classifiers using these spectral moments on a dataset of normal, COVID-19, and pneumonia X-ray images.
• Performed simulations using Matlab for image preprocessing and spectral moment generation, and Python for classifier implementation.

Main Results:

• The proposed spectral moments, when used with SVM and Random Forest, significantly improved classification accuracy.
• Achieved a high accuracy of 0.975 in classifying chest X-ray images.
• Demonstrated the efficiency and accuracy of the spectral moment-based approach through a parametric study.

Conclusions:

• The developed spectral coefficient-based method offers a promising approach for accurate chest X-ray image classification.
• The method shows satisfactory results, particularly in distinguishing between normal, COVID-19, and pneumonia cases.
• Further research is needed to develop a more accurate and faster version of this approach.