Mamotenko S. Nonparametric Analysis of Response Signal Spectra for the Recovery of Magnetic Resonance Imaging Data.

Українська версія

Thesis for the degree of Doctor of Philosophy (PhD)

State registration number

0826U005112

Applicant for

Specialization

  • 105 - Прикладна фізика та наноматеріали

11-09-2026

Specialized Academic Board

PhD 16586

Taras Shevchenko National University of Kyiv

Essay

Magnetic resonance imaging (MRI) is a leading method of non-invasive medical imaging due to its ability to provide high soft-tissue contrast. However, the presence of, in particular, thermal noise generated by electronic components affects the quality of the obtained diagnostic data. In k-space, where the primary registration of the MRI signal is performed, this noise has an additive character, a white spectrum, and follows a normal distribution (Gaussian distribution). Since the energy of the useful signal decreases at high spatial frequencies, which correspond to contours and fine anatomical details, the relative contribution of noise in these regions increases significantly. The application of traditional linear and nonlinear spatial filtering methods to final magnitude images has several limitations. The nonlinear operation of calculating the modulus of a complex signal changes the noise statistics from Gaussian to Rician, which may lead to a positive intensity bias in regions with a low signal-to-noise ratio. In addition, although spatial averaging algorithms are capable of reducing the noise level, they often demonstrate excessive smoothing of high-frequency components, which complicates the identification of, in particular, pathological brain lesions.

Research papers

Mamotenko, S.P. and Netreba, A.V., Magnetic resonance image data processing by the periodogram method, Radioelectron. Commun. Syst. (2025).

Mamotenko, S.P. and Netreba, A.V., Reconstruction for MRI by autoencoder and periodogram method, Molecular Crystals and Liquid Crystals, 769(15-18), 1547– 1557 (2025)

S. P. Mamotenko and A. V. Netreba, Multitaper Spatial Frequency Domain Signal Noise Reduction in Magnetic Resonance Imaging, J. Nano-Electron. Phys. 18(1), 01008 (2026)

Reconstruction of magnetic resonance images using multiple periodogram methods / Mamotenko S. P. and Netreba A. V. // Conference Proceedings Book of XX International Scientific Conference Electronics and Applied Physics, 2024, Kyiv, Ukraine

Reconstruction for MRI by Autoencoder and Periodogram Method / Mamotenko S. P. and Netreba A. V. // International research and practice conference: Nanotechnology and Nanomaterials 21-24 of August 2024 Uzhhorod, Ukraine

Multitaper k-Space Denoising of Magnetic Resonance Images / Mamotenko S. P. and Netreba A. V. // Conference Proceedings Book of XXI International Scientific Conference Electronics and Applied Physics, 2025, Kyiv, Ukraine

Magnetic Resonance Image Denosing Using Frequency-Domain Windowing / Mamotenko S. P. and Netreba A. V. // Conference Proceedings Book of XXV International Young Scientists Conference on Applied Physics, 2025, Kyiv, Ukraine

Wiener-Based Spectral Shrinkage for K-Space Denoising of Magnetic Resonance Images / Mamotenko S. P. and Netreba A. V. // Conference Proceedings Book of XXVI International Young Scientists Conference on Applied Physics, 2026, Kyiv, Ukraine

Files

Similar theses