A new approach for restoration of NMR signals

NMR is a widely used technique for analysising the structure of matter, however the signals produced can be of very poor quality even with the averaging of hundreds or thousands of individual scans. It has therefore become common practice to use a restoration algorithm to improve the signal and remove degradation effects such as line broadening or blurring caused by the NMR instrument itself. The dominant restoration method currently used is maximum entropy. However this is very slow even with signals of 16000 data points and with newer NMR instruments the sampling rate is increasing dramatically enabling signals of over 100,000 data points or more to be collected. Clearly maximum entropy processing of these signals will take hours. Here we present an alternative nonlinear restoration method, projection onto convex sets (POCS), which is capable of producing restored signals of equal or better quality to maximum entropy and in much less time.