Are Artificial Intelligence-Assisted Three-Dimensional Histological Reconstructions Reliable for the Assessment of...

Area of Science:

• Biomaterials Science
• Histology
• Artificial Intelligence

Background:

• Accurate 3D reconstruction of bone microarchitecture is crucial for understanding skeletal diseases and evaluating treatments.
• Current methods for histological reconstruction can be labor-intensive and may not fully capture complex microarchitectural details.

Purpose of the Study:

• To develop a novel AI-assisted technique for creating 3D histological reconstructions from serial sections.
• To validate the accuracy of this technique in replicating bone trabecular microarchitecture by comparing it with microCT reconstructions.

Main Methods:

• A bone biopsy sample was processed for microCT scanning and histological analysis.
• A modified U-Net architecture was employed for AI-assisted tissue classification.
• Detector-free local feature matching with transformers was utilized for section alignment and 3D reconstruction.
• Micromorphometric parameters were calculated using CTAn software for both histological and microCT datasets.

Main Results:

• Strong linear relationships (correlation coefficients ranging from 0.666 to 0.777) were observed between microCT and histological reconstructions for parameters like BV/TV, BS/TV, Tb.Pf, Tb.Th, and Tb.Sp.
• Bland-Altman and mountain plots indicated good agreement for Bone Volume per Total Volume (BV/TV) measurements between the two methods.
• The AI-assisted technique demonstrated high fidelity in capturing trabecular microarchitecture.

Conclusions:

• The developed AI-assisted 3D histological reconstruction method accurately replicates bone trabecular microarchitecture.
• This novel technique offers a valuable tool for simultaneous assessment of histological information and microarchitecture in bone research.