AUTOMATING THE QUANTITATIVE-ANALYSIS OF 2-D NEURAL DENDRITIC TREES

Neurons in the central and peripheral nervous system vary widely in th eir dendritic branching patterns. Quantification of the morphological characteristics used to identify different classes of neurons and rela te neural structure to function requires that accurate metric and non- metric data be obtained from neural images obtained by camera-lucida d rawing or from digitized video images made with transmitted, fluoresce nce or confocal microscopy. This paper describes a largely automated p rocedure for determining the dendritic tree structure of largely plana r cells (such as retinal ganglion cells or cells in tissue culture mon olayers) from an initial pictorial representation or digitized image. From this structure, non-metric data (such as the ordered 'tree' of br anches) and metric information (such as total dendritic length and den dritic:field area) can be automatically computed. The use of this meth od is specifically illustrated in the capture of the dendritic tree st ructure of retinal ganglion cells from the rabbit retina.