| Literature DB >> 15189528 |
Raja Kannan Mutharasan1, Ashwin Nagaraj, Andrew J Hamilton, David D McPherson, Saroja Bharati.
Abstract
The human atrioventricular conduction system (AVCS), which includes the AV node and its approaches, AV bundle (penetrating, branching, and bifurcating parts), and the bundle branches, is a curved complex structure that has not been reconstructed in three dimensions using computer technology. Microscopic slides of every 40(th) serial section (cut at 7 micron level) of the AVCS were digitized into 600 dots/inch color images. External outlines of each section were manually segmented using commercially available three-dimensional rendering software (Rhinoceros). The AVCS was traced from light microscopy and superimposed onto the external outlines. To account for inherent errors in histological slide preparation, an optimization procedure was used to align external outlines of all sections. The optimal rotation and translation of each section was established by maximizing area of overlap between adjacent sections. A sequential one-dimensional minimization algorithm was used for optimization. Rotation and translation values were then used to align external outlines and the superimposed conduction system, reconstructing the AVCS in three-dimensions. To validate the method, the algorithm was applied to a digitized image transformed with known translations and rotations. The validation procedure demonstrated that each test image aligned in translations and to within 0.01 degree in rotations. Spatial error determined by resolution of the digitized images was +/-0.5/600 inch (+/-21 microns). Three-dimensional reconstruction of every 40th serial section clearly demonstrated the complex curved shape of the AVCS. Three-dimensional reconstruction of the human and canine AVCS permits accurate pathological and electrophysiological correlation of the conduction system.Entities:
Mesh:
Year: 2004 PMID: 15189528 DOI: 10.1111/j.1540-8159.2004.00522.x
Source DB: PubMed Journal: Pacing Clin Electrophysiol ISSN: 0147-8389 Impact factor: 1.976