Statistical analysis of corticopontine neuron distribution in visual areas 17, 18, and 19 of the cat.

The spatial organization of visual corticopontine neurons was studied both at a "large scale" (in relation to cortical visual field maps) and at a "small scale" (in relation to cortical modular organization). Large injections of horse-radish peroxidase-wheat germ agglutinin were made in the pontine nuclei. In complete series of sections from parts of areas 17, 18, and 19, the position of each retrogradely labeled neuron was recorded with an x-y plotter connected to the microscope stage. Each cell was thus given a set of x, y, and z coordinates. After alignment of the sections, three-dimensional computer reconstructions of the distribution of the labeled cells were made. With program RPOP (developed by Blackstad and Bjaalie, '88), the reconstructions were studied with different rotations, scaling, etc. In addition, section-independent parts of reconstructions were isolated ("windows") and further analyzed. Curved parts were automatically unfolded for inspection of distribution patterns and determination of cell densities. The spatial distribution of the labeled cells was analyzed within small windows, where density gradients are negligible. We confirm and extend previous demonstrations of a large-scale aggregation of visual corticopontine cells due to density gradients by showing that densities of corticopontine neurons increase linearly as a function of distance from paracentral to lower visual field representations in area 17 (and partly in areas 18 and 19). We demonstrate that density gradients are steeper in area 17 than in area 18. For example, clear-cut differences between the areas in mediolateral density gradients are found. These findings are discussed in relation to the different visual field maps of the areas and the existence of a similar visual field representation in corticopontine projections from different visual areas. The type of small-scale distribution (randomness or non-randomness, aggregation into clusters, bands, etc.) was studied with statistical methods. Such analysis shows that the labeled cells within small zones are non-randomly distributed in all three areas. In most cases, the analysis indicates an aggregated spatial distribution. A possible relationship to the cortical map of direction selectivity is discussed. To our knowledge, this study is the first to combine the use of three-dimensional computer reconstructions of a population of labeled neurons, with subsequent statistical analysis of spatial point (cell distribution) patterns.