Обробка даних для магнітно-резонансної томографії методом періодограм

Автор(и)

DOI:

https://doi.org/10.20535/S002134702412001X

Анотація

Magnetic resonance (MR) images are often influenced by undesirable noise that is combined with the relevant signal. Therefore, the resulting images may not provide the quality needed for its interpretation. This study proposes the periodogram as the nonparametric method for processing and reconstruction of noise-impacted MR images without prior knowledge of their spectral characteristics. Here we simulate the real-life noise contamination of the images by artificially adding the Gaussian noise. By then transforming the corrupted MR image into the frequency domain, applying the Daniell periodogram and performing an inverse transformation, the noise impact is reduced. Authors aim to evaluate method’s possible applications as well as its limitations regarding its future use in noise elimination in MRI scans utilizing Structural Similarity Index Measure (SSIM), Peak Signal-to-Noise Ratio (PSNR) as well as Fractal Dimension (FD) metrics. The values of the said metrics of the denoised images largely correspond to authors’ visual evaluation of the images.

Опубліковано

2025-07-15

Як цитувати

Мамотенко, С. П., & Нетреба, А. В. (2025). Обробка даних для магнітно-резонансної томографії методом періодограм. Вісті вищих учбових закладів. Радіоелектроніка. https://doi.org/10.20535/S002134702412001X

Номер

Розділ

Оригінальні статті