A Gauss-Seidel iteration scheme for reference-free 3-D histological image reconstruction

Area of Science:

• Histology
• Medical Imaging
• Computational Biology

Background:

• Three-dimensional (3-D) reconstruction of histological slices offers high-resolution morphological investigation.
• Slice preparation introduces deformations, necessitating image unwarping for accurate 3-D reconstruction.
• Existing unwarping methods often rely on external references or conservative criteria, potentially altering natural morphology.

Purpose of the Study:

• To develop an unwarping method that separates anatomical structures from deformation artifacts in histological image stacks.
• To present an iterative scheme applicable to various unwarping methods without external references.
• To preserve the natural curvature of biological tissues during 3-D reconstruction.

Main Methods:

• An iterative scheme inspired by the Gauss-Seidel method is proposed for image stacks.
• The method treats unwarping as a separation of low-frequency anatomy and high-frequency deformation errors.
• The approach is validated on synthetic, simulated (CT-derived), and real histology data.

Main Results:

• The iterative scheme effectively reduces deformation artifacts in 3-D histology volumes.
• Natural anatomical curvature is successfully preserved post-unwarping.
• On simulated data, the mean Target Registration Error (TRE) was reduced to under 1 pixel after six iterations.

Conclusions:

• The proposed iterative unwarping method accurately reconstructs 3-D histology volumes by separating anatomy from deformation.
• The technique is universally applicable, requires no external references, and preserves tissue morphology.
• This method significantly improves the fidelity of 3-D histological reconstructions.