Automated nerve fiber counting using an array processor in a multi-minicomputer system

It has been suggested that recovery of motor and sensory function in the site distal to a peripheral nerve lesion should be improved if the nerve bundles (fasciculi) are matched and individually sutured. Three parameters are proposed to provide quantitative data: the count of the nerve fibers that regenerate, the number of functional regenerated nerve fibers, and a measurement of end organ reinnervation. A thin cross section of a transected and repaired sciatic nerve of a mongrel cat is fixed, stained, photographed, and digitized through a microscope 6 months following nerve repair. The data arrays are then subjected to four basic processing routines: edge enhancing, thresholding, template matching, and peak detection. Finally, the peaks are counted and provide an estimate of the number of nerve fibers in the nerve under study. Comparing counts of nerve fibers proximal and distal to the transection site of the nerve provide data on the proportion of regeneration present at various times. The content of this paper is, to a large extent, describing the implementation of the needed image-processing algorithms for automated counting on the Multi-MiniComputer System (MMCS). Optimal use of the AP-120B array processor and the pipeline processing provided by using the Eclipse 200s and the Nova 3 make a marked improvement in overall throughput.