Surface from Contour

METHODS FOR VISUALIZING AND MEASURING NEURONS IN THREE DIMENSIONS USING ELECTRON MICROSCOPY.

INTRODUCTION

Microscopic analysis of biological structures can be significantly enhanced by representing the object of study as a three-dimensional entity. Because the cellular and subcellular structures of interest often are physically sectioned during examination, reassembly across multiple tissue slices is required to reconstruct the original shape. We have used three types of data entry to reconstruct and analyze the distribution and structure of neurons and their processes in three dimensional space.

Serial electron microscope photo-micrographs are manually digitized
Neuronal dendrite and axon branching patterns from thick (100-500micron) tissue sections are digitized using a three-dimensional camera lucida system
Sequential images (0 .1-10microns)are captured using a scanning laser confocal microscope.

Data files from these entry procedures have been used to assemble "solid" structures which are subsequently displayed for viewing using surface construction algorithms. Morphometric parameters are analyzed in a quantitative fashion.

RECONSTRUCTING SOLIDS FROM SERIAL SECTIONS

Fig1D Using a digitizing tablet, contours (profiles) are traced from electron microscope (EM) photomicrographs of an axonal bouton from a physiologically identified thalamucortical projection neuron that innervates visual cerebral cortex layer 4. Profiles of a bouton, its mitochondria and synapses are marked on the photo above. Anatomical landmarks entered during digitization (vertical microtubules labeled A,B,C) are used later to automatically stack the profiles into correct alignment.

Fig2D

The aligned stack of profiles may then be viewed and rotated interactively on a computer screen. (Bouton profiles are pink, mitochondria are blue, target cell body is white, postsynaptic densities are yellow).




		Using computer graphics techniques, the triangular mesh appears to be a solid surface. (Bouton is red, target spines are blue, synaptic densities are yellow).


		A contour stack can be made to appear as a solid object by automatically fitting a surface over the contour frame. Triangular patches are fitted between pairs of adjacent contorus, creating a surface mesh.

The image to the left shows some of the techniques used to represent the surface as a solid surface. Here, location and orientation of the surface have been calculated using a Zucker-Hummel 3D operator (the object is a femur). Local surface position is calculated using local itensity changes only (pink patches). Surface normals (white) in A represent the orientation of each patch and are directed to the volume’s “outside”.

All software used for these projects was written by Jill Gemmill during 1985-1996.

Click the button at left to learn more about Neurobiology Research at UAB


	[Jill Gemmill] [Biosketch] [Funded Research] [Visualization] [3D Camera Lucida] [Confocal Microscopy] [Movies] [Surface from Contour] [SciViz Neuron] [Publications] [Presentations] [Awards] [Technical Citations]