BigReg: an efficient registration pipeline for high-resolution X-ray and light-sheet fluorescence microscopy

Area of Science:

• Biomedical Imaging
• Preclinical Research
• Bone Biology

Background:

• Multimodal imaging, combining X-ray microscopy (XRM) and light-sheet fluorescence microscopy (LSFM), is crucial for studying bone remodeling diseases like osteoporosis.
• Accurate registration of XRM and LSFM data is challenging due to signal inconsistencies and initial misalignment, hindering detailed structural and functional analysis.
• Conventional registration methods often fail in reference-free scenarios or with significant misalignments.

Purpose of the Study:

• To develop an automated, reliable registration pipeline (BigReg) for multimodal mouse bone imaging using XRM and LSFM.
• To overcome limitations of existing methods in handling inconsistent signals and large misalignments.
• To enable micrometer-level structural correspondence for enhanced preclinical bone research.

Main Methods:

• Introduced BigReg, a two-stage automated registration pipeline for high-resolution XRM and LSFM volumes.
• Stage 1: Coarse alignment using surface feature extraction and successive global-to-local point-cloud-based methods.
• Stage 2: Fine alignment via a modified cross-correlation technique in the 3D Fourier domain.

Main Results:

• BigReg achieved accuracy comparable to manual landmark-based registration (LMD: 8.36 ± 0.12 μm, LM fitness: 85.71 ± 1.02%).
• The pipeline successfully initialized otherwise failing mutual information-based methods, further improving accuracy (LMD: 7.24 ± 0.11 μm, LM fitness: 93.90 ± 0.77%).
• Demonstrated precise alignment of fine structures like lacunae and osteocytes.

Conclusions:

• BigReg is the first automated method for registering XRM and LSFM volumes without manual intervention.
• The pipeline facilitates quantitative, multimodal analysis of bone microarchitecture and disease pathology.
• Opens new avenues for osteoporosis research by enabling accurate alignment of critical bone structures.