PubMed1975
It is the quantitative study of the central nervous system to enumerate neurone populations. Since neurones are generally arranged in non-random patterns the conventional method of counting with a square lattice graticule in the ocular of the microscope is severely restricted. Appropriately tailored methods of (a) counting, or (b) estimating the density of a neurone population must therefore be used. Four types of neuronal distribution pattern are discussed and morphometric methods adapted to them are presented, as follows: (1) Where the whole population is circumscribed, as in the anterior horn motor neurones of the cord, direct counting is feasible. Only those cells whose nucleoli are visible should be included; this diminishes the bias in favour of large cells, and if one corrects for the fact that larger neurones have larger nucleoli the bias can be eliminated completely. (2) Where neurones are situated on an "interface", as Purkinje cells are, the length of its profile (in this instance the boundary between the molecular and granular zones) is estimated by superimposing parallel lines on the microscopical image and counting intersections as these lines cross the interface. The number of Purkinje cells in each particular field is counted at the same time. A "linear dinsity" (PL) is then calculated from the formula pl = 2n/pid where n is the quotient of (a) the number of cells counted along (b) an estimated length of profile, and d is the distance between the parallel lines of the graticule. For comparative studies between cerebella of different sizes a correction factor is easily introduced. Those who prefer to work on photomicrographs can use a mapping wheel to measure a length of profile and then proceed as before. (3) Where neurones are organised in a curving band, as in the dentate nucleus, another form of cell density can be established. The microscope is focused on a neurone at random, and the distance from this cell to its nearest neighbour is measured, preferably by means of a screw-micrometer eyepiece. The mean value r of a number of such measurements is substituted in a formula which gives the "areal density" of the neurone population: pa = 1/4r2. This method can be applied to photomicrographs by using a pair of dividers and/or an accurate rule to measure the set of r values. (4) In large collections of neurones, e.g., in the thalamus, three methods are available: (a) the nearest neighbour method; (b) a conventional squared graticule count, and (c) a count of cells intersected by a line probe as in Haug's (1972) technique (fig. 5), or a modified form of Strong's (1966) transect method.