My introduction to NMR began as a student in the laboratory of Sir Peter Mansfield in 1973. At that time a few initial reports had started to appear that were to be significant for the early development of MR imaging. Dr. Paul Lauterbur had demonstrated that NMR could be used for mapping spin density distributions in large objects [1], NMR had already become established as a valuable tool for the study of biochemical problems including the use of excised tissues [2], and the concept that measurement of spin relaxation times could provide a valuable diagnostic indicator for cancer detection was widely known [3,4]. At the same time, X-ray computerized tomography had shown that tomographic imaging methods were a valuable tool in the hospital environment, and there was an expectation that NMR could be used in a similar manner. What was clearly lacking before NMR could be used as a diagnostic imaging technique was the availability of good imaging methods and suitable instrumentation, though in these areas several important concepts had already been demonstrated. This included the use of selective RF excitation [5], the availability of rapidly switched field gradient coils [6,7], and the increasing availability of minicomputers that were capable of performing automated data acquisition and data processing.
My initial training with Peter Grannell included an investigation of a projection-reconstruction technique [8] for image formation. In Fig. 1a and b are shown two 1D projections and a reconstructed image through a finger (of P.K.G.) which was obtained in 1974 [9,10]. Being limited by the available computer capabilities, this image was obtained using only five projections and reconstructed to 10 grey levels. Since suitable image display devices were not available the image was then formed by pasting up appropriately sized pieces of black paper according to the reconstructed image data values.