Spline-based deconvolution technique in electron paramagnetic resonance imaging

Abstract

The article presents a modification to the spline deconvolution technique. The method was developed within the framework of electron paramagnetic resonance imaging and is particularly useful in cases of low signal-to-noise ratio (S/N) and when the deconvolution kernel is a large fraction (about 50%) of the studied function. The modified spline deconvolution method works directly in the signal space and does not require a low-pass filter that causes loss of information and limits the resolution as is the case of Fourier deconvolution. The comparison was done with Fourier deconvolution and was based on three parameters that account for resolution, noise, and proportion stability. Experimental tests were done using two nitroxide-free radicals of different linewidths (0.05 and 0.16 mT, respectively) at different gradient values (between 0.05 and 0.2 mT/m). Modified spline deconvolution gave consistently better results in all the three parameters, particularly at the lower gradient values. For this reason the method is useful in in vivo experiments where the low S/N does not permit use of high gradient values.