Accurate computer reconstruction and graphics display of complex neurons utilizing state-of-the-art interactive techniques.

Abstract This paper reports some advances in reconstruction techniques motivated by the need for accurate analysis of the structure of many complex neurons. The existing neuron reconstruction system required too much time to enter trees, limited the complexity of traced neurons to that which could be held in computer memory, required too much time to correct for inaccurate focus axis readings, and was limited in realistic display capabilities. Also, since the study required accurate measurement of very fine processes, projecting the neuron image onto a data tablet in order to speed up tracing was not possible. An operator controls the microscope stage and focus movement with a 3D joystick in a “positional mode with clutching.” He also controls the diameter of a circle which is displayed on the computer graphics CRT and mixed with the neuron image by a camera lucida. He moves the stage and focus to pass the neuron processes under the circle and indicates to the computer when to capture a coordinate. The most common point type is captured automatically; the next three most common types are indicated by voice command. The remaining types are entered with pushbuttons. Once all pieces of all trees from all tissue sections have been entered, a “match the squares” merging technique is used to attach truncated tree pieces to their continuations in adjacent tissue sections. An automatic scaling program corrects for tissue shrinkage and wrinkling and for incorrect focus axis readings. Finally, 3D dynamic displays, realistic plots, and statistical summaries of the neurons are generated. Reconstruction time has been halved, but an order of magnitude improvement is still needed.