Growth of the rat somatic sensory cortex and its constituent parts during postnatal development.

We have compared the size and arrangement of the primary somatic sensory cortex (SI) and its constituent parts in juvenile (1 week old) and mature (10-12 weeks old) rats using succinic dehydrogenase histochemistry and digital image analysis. Our goal was to determine whether some regions of the maturing cortex grow more than others. To this end, we examined (1) the growth of barrels and the surrounding (interbarrel) cortex, (2) the growth of the major somatic representations within SI, and (3) the overall growth of SI compared to the neocortex as a whole. With respect to the first of these issues, SI barrels and barrel-like structures grow more than the intervening cortex. The growth of these elements varies according to region: barrels in the head representation more than double in size, whereas the barrel-like structures in the paw representations increase by only about half this amount. The growth of the major somatic representations within SI is also heterogeneous, the representation of the head enlarging to a greater extent than the representations of the paws. Thus, the ratio of the total area of head representation to the combined paw representation is 15% greater in adults than in juveniles. Finally, the primary somatic sensory cortex grows to a somewhat greater extent than the neocortex as a whole. These observations demonstrate that postnatal cortical growth is not uniform; it varies among cortical barrels and the immediately surrounding (interbarrel) cortex, among the representations of different body parts, and between SI and the rest of the neocortex. As an explanation of this differential growth, we suggest that the neuropil of metabolically (and/or electrically) more active cortical regions grows to a greater extent during maturation than that of less active regions.

Nerve cells with irregular processes: demonstration of anisotropic core geometry of a pyramidal cell.

An analytical, recursive method has been developed to demonstrate the anisotropic electrotonic geometry of nerve cells containing varicose or spiny dendrites. The procedure has been based on the distribution of the core geometry of dendrites into modules which consist of module elements where the physical length is much shorter than the actual space constant. The unambiguous representation of the anisotropic core geometry has been possible by plotting the decomposed geometries separated under the condition of the unidirectional spread of the wave front of dendritic potentials. This decomposition has revealed the bidirectional, "smoothed" core geometries as a function of irregular distribution of varicosities or spines. The shape of decomposed core geometries may change according to the position of the input site. The shaping of core geometry reflects the electrotonic effectiveness of a synaptic site to any arbitrary locations which may lead to considerable savings in computations on synaptic effects. The detailed, computer-reconstructed geometry of the apical dendritic field of the pyramidal cell has been analysed by the proposed method. The frequency-dependence of input impedances has been compared between the original and the transformed core geometries assuming that the current is injected into the soma. The significance of dendritic irregularities in the impedance matching has been studied when the shaping of the core geometry has been induced by laminar inputs. The proposed approach may be useful in comparing the input dependence of the receptive fields of different non-smooth cells. The mismatch of the core geometries induced by the opposite travelling waves from the same anatomical location has also been studied and the possible control of the preferred, direction-sensitive activities will be discussed. The important differences between the compartmental modelings based on the known isotropic treatment of dendrites and the more realistic anisotropic approach will be illustrated.

Approximation of missing sections in computer reconstruction of serial EM pictures.

Replacing missing contours or completing a series of contours of biological structures has been investigated. The paper reports on a method for approximating lateral surfaces between adjacent contours and generating new artificial outlines by slicing the interpolated surface parallel with the original sections. The research method was developed mainly due to the need to fill in damaged or missing ultrathin sections in longer series, since their absence may impede the three-dimensional visualization of selected components. The new program attached to the shape reconstruction program system existing at Second Department of Anatomy, Semmelweis University Medical School, Budapest, Hungary also permits new contours to be generated into the models of single or branching objects. Besides replacing missing contours, the program makes the reconstructed model more precise.

A computer reconstruction system for biological macro- and microstructures traced from serial sections.

A system of programs for building three-dimensional (3D) models of biological macro- and microstructures has been developed. A PDP 11/34 laboratory minicomputer with graphical output devices, and a QUANTIMET image analyzer are used for the reconstruction. The contours of the features can be digitized directly from tissue sections mounted on the stage of an OPTON research light microscope, or indirectly from microphotographs, and from camera lucida drawings. The operator selects contours of features in question from the image by using Quantimet's interactive input devices and commands the Quantimet system to record the image. The 3D models are built from a series of 2D outlines. In the visualization phase the reconstructed objects can be investigated from different viewpoints with their hidden lines removed. Additional calculations (volume and surface area) help in quantitative classification of the reconstructed structures.

A new approach to merging neuronal tree segments traced from serial sections.

The merging process which is one of the most significant problems in computer-aided reconstruction of neuronal trees has been investigated. This paper reports a new approach in merging tree segments traced from serial sections. The research was motivated by the fact that in most cases users can distinguish connectable cut ends of the different tree segments without any interaction, and with this knowledge merging can be easily solved by a computer program added to the neuron reconstruction program system at Semmelweis University Medical School, Budapest. Then, the merged neuron trees can be handled as if they had been digitized from one histological section.