نهمین کنگره نقشه برداری مغز

نمایش مقاله

Title:

Brain MRI Texture Analysis for Diagnosing Alzheimer's Disease Using Fractional Fourier Transform

Subject:

Advanced Techniques in Signal Acquisition and Image Processing

Presenter Author:
Name:	Shahyar Saramad

Author:
Name 1:	Shahyar Saramad - Corresponding - Presenter
Affiliation 1:	Department of Physics and Energy Engineering, Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran.

Name 2:	Nader Razmara
Affiliation 2:	Department of Physics and Energy Engineering, Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran.

Name 3:	Mojtaba Shamsaei
Affiliation 3:	Department of Physics and Energy Engineering, Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran.

Name 4:	Maryam Noroozian
Affiliation 4:	Memory & Behavioral Neurology Department, Roozbeh Hospital, Tehran University of Medical Sciences

Name 5:
Affiliation 5:

Name 6:
Affiliation 6:

Name 7:
Affiliation 7:

Name 8:
Affiliation 8:

Name 9:
Affiliation 9:

Name 10:
Affiliation 10:

Abstract

Original Research

introduction :

Early diagnosis of Alzheimer’s Disease (AD) requires sensitive biomarkers capable of detecting microstructural changes before observable atrophy. Conventional MRI analyses and Fourier Transform (FT)-based methods often have limited sensitivity in capturing subtle structural and textural alterations. In this study, a novel application of the Fractional Fourier Transform (FrFT) is introduced for brain MRI texture analysis. The FrFT allows for the extraction of both phase and magnitude information at fractional domains between the spatial and frequency representations, providing enhanced sensitivity to early microstructural changes. This approach aims to improve the detection of AD before macroscopic atrophy becomes evident.

Methods:

A total of 98 MRI scans from patients with AD and 121 scans from cognitively normal (CN) individuals were obtained from the MNI152 atlas. At the group level, Pearson correlation across slices with the MNI152 atlas was calculated, and the optimal fractional order (α) for the FrFT was determined to maximize the differences between AD and CN groups. For subject-level analysis, a dual-correlation framework was implemented, calculating both the mean and standard deviation of slice correlations for each individual. Performance metrics included Recall, F1-score, Accuracy, Precision, and Specificity. Comparisons were made between the FrFT components (Real, Imaginary, and Phase) and conventional methods, including the original image and the standard Fourier Transform (FT), to evaluate sensitivity to microstructural alterations.

Results:

At the group level, the FrFT Phase component with an optimal fractional order of α = 0.1π/2 exhibited the largest relative difference between AD and CN groups at 18.64%, whereas the main image (α = 0) and the Fourier Transform (α = π/2) showed 3.6% and 1% differences, respectively. Overall, the optimal FrFT achieved a relative difference of 14.56%, demonstrating superior sensitivity to subtle changes. At the subject level, using mean correlation, the Phase component achieved a Recall of 83.33%, F1-score of 90.91%, and Accuracy of 92.86%, while the Real and Imaginary components also performed well, with Accuracy above 88%. Precision and Specificity were 100% for all components. Using the standard deviation of slice correlations, the Real component achieved perfect performance with Recall, F1-score, and Accuracy of 100%, while the Phase component showed high performance (Recall 88.9%, F1 94.1%, Accuracy 95.2%). In contrast, the Image and Imaginary components demonstrated reduced sensitivity, highlighting the robustness of the FrFT Real and Phase components.

Conclusion:

The Real and Phase components of the Fractional Fourier Transform reliably capture subtle structural and textural alterations in brain MRI while maintaining stability across slices. The novel dual-correlation framework, combining mean and standard deviation of slice correlations at the subject level, provides a robust approach for early detection of Alzheimer’s Disease. These findings indicate that FrFT-based analysis outperforms conventional image and Fourier Transform methods and represents a promising tool for identifying microstructural changes prior to visible atrophy, potentially enabling earlier clinical intervention.

Keywords:

Alzheimer’s Disease, Early Diagnosis, Fractional Fourier Transform, MRI Texture Analysis, Dual-Correlation, Biomarkers.