Deconvolution algorithm based on automatic cutoff frequency selection for EPR imaging

• Published in: Magnetic resonance in medicine Vol. 50; no. 2; pp. 444 - 448
• Main Authors: Deng, Yuanmu, He, Guanglong, Kuppusamy, Periannan, Zweier, Jay L.
• Format: Journal Article
• Language: English
• Published: Hoboken Wiley Subscription Services, Inc., A Wiley Company 01.08.2003; Williams & Wilkins
• Subjects: Algorithms; Animals; Biological and medical sciences; cutoff frequency; deconvolution; Digestive System - anatomy & histology; Electron Spin Resonance Spectroscopy - methods; EPRI; Fourier transform; image reconstruction; Imaging, Three-Dimensional; Medical sciences; Mice; Phantoms, Imaging; cutoff frequency; deconvolution; Fourier transform; EPRI; image reconstruction
• ISSN: 0740-3194; 1522-2594
• DOI: 10.1002/mrm.10533

The large line‐width associated with electron paramagnetic resonance imaging (EPRI) requires effective algorithms to deconvolve the true spatial profiles of spins from the measured projection data. The commonly used Fourier transform (FT) deconvolution algorithm is easy to implement but suffers from the division‐by‐zero problem. As a result, a couple of parameters are used to control the deconvolution performance. However, this is inconvenient and the deconvolution results are subject to the experience of the operators. In the present work we examined FT deconvolution for EPRI, and proposed an automatic algorithm to determine the cutoff frequency by calculating the piecewise variance of the division result of the Fourier amplitude spectra. The deconvolution algorithm and the filtered back‐projection image reconstruction algorithm were implemented and validated using 3D phantom and in vivo imaging data. It was clearly observed that the image resolution improved after deconvolution with the proposed algorithm. Magn Reson Med 50:444–448, 2003. © 2003 Wiley‐Liss, Inc.