There is currently a search for an automated, objective and non-invasive system that would accurately diagnose pigmented skin tumours. Systems that measure either the spatial or the spectral characteristics of light reflected from the skin have shown promise for this purpose but only a few studies have combined spatial and spectral information. We plan to study this and consequently need to construct a cost-effective research spectral imaging system but the design will require a compromise between spatial and spectral information. Here, the effect, on diagnostic accuracy, of reducing the spectral resolution of spectrophotometry data was studied. Also studied was the effect of reducing the spectral range to that of the sensitivity range of low-cost detectors. There was no significant fall in the diagnostic power when the spectral resolution was reduced from 3.8nm to 50nm, and when the spectral range was reduced from 320-1100nm to 400-1000nm. Therefore, in the design of the spectral imaging system emphasis was placed on spatial resolution and a standard detector was used. The spectral imaging system contains a broadband light source, diffraction grating monochromator and CMOS camera and achieves 10nm spectral resolution over a spectral range of 400-1000nm, with a spatial resolution of 40 microns over a field of view of 2cm.