Optimization method for isodensity surface models obtained with electron microscope tomography data

Electron microscope tomography produces reconstructed biological tissue volumes containing structural data with few-nanometer spatial resolution. Frequently, surface models are constructed to visualize and analyze the structural data. These models are usually constructed to approximate the contrast boundaries of various structural components within the volume. An automatic method is presented for obtaining isodensity surface models that correspond most closely to the steepest contrast boundary at the edge of each component. We validate our method using simulated volumetric data, and demonstrate its application to a reconstructed volume obtained from the neuromuscular junction of the frog.