“O-PTIR imaging overcomes the infrared diffraction limit and enables to obtain compositional imaging at an unprecedented submicron resolution…This represents a spatial resolution unattainable by conventional infrared imaging modalities, opening new avenues for analysis of structure-composition associations in bone tissue.”
FULL PUBLICATION >
Reporting in the Journal of Structural Biology: X, researchers at Temple University introduce a novel approach utilizing O-PTIR spectroscopy and imaging coupled with machine learning analysis to assess bone tissue composition at 500 nm spatial resolution. Understanding bone composition at the submicron level is crucial to elucidate factors contributing to bone disease and fragility. Conventional FTIR imaging is limited to spatial resolution of several micrometers, which cannot adequately capture properties of tissue features at the submicron scale of the building blocks of bone tissue quality. The team evaluated thick bone samples embedded in PMMA blocks from wildtype mice and from a mouse model of osteogenesis imperfecta, eliminating the need for cumbersome thin sectioning.
The researchers demonstrated that O-PTIR spectroscopy successfully assessed the distribution of bone tissue mineral and protein, as well as the structure-composition relationship surrounding microporosity at submicron resolution. Spectral imaging highlighted the distribution of PMMA and bone mineral and protein components, with the mineral/PMMA ratio providing detailed views of bone microporosity that showed spatial correlation with scanning electron microscopy observations. The team collected 100 spectra within a 50 μm linear region across a single pore at 500 nm step size, showing clear differences in spectra from bone tissue and pore space. Linescans across cortical bone at 500 nm intervals enabled quantification of mineral content, crystallinity, and carbonate content in spatially defined regions, revealing region-dependent differences between wildtype and osteogenesis imperfecta bones.
Machine learning analysis using support vector machine was successful in identifying bone phenotypes based on input of spectral data. Using approximately 8,300 spectra from wildtype and osteogenesis imperfecta samples, the analysis showed that 86% of wildtype and 90% of osteogenesis imperfecta samples were correctly identified when using the collagen spectral range. In contrast, only 60% of wildtype and 81% of osteogenesis imperfecta samples were identified correctly using the apatite range, highlighting that changes in collagen are more prominent markers underlying the osteogenesis imperfecta phenotype. The findings corroborate the importance of collagen underlying osteogenesis imperfecta pathogenesis and demonstrate the application of machine learning to distinguish typical and fragile bone phenotypes based on their tissue-level spectral data.
The findings highlight the potential of O-PTIR spectroscopy and imaging as valuable tools for exploring bone submicron composition. The approach enables data collection at 500 nm spatial resolution with minimal sample preparation in non-contact reflection mode, and allows analysis of thick and intact bone samples without the need for thin sectioning of calcified bone. The submicron resolution achieved using O-PTIR imaging provides unprecedented detail for assessing tissue features such as microporosity associated with osteocyte lacuna, offering an approach to gain insights into the compositional features of the perilacunar tissue and how it is associated with bone disease and fragility. The method may be of value for analysis of samples from pre-clinical studies on therapeutic interventions for skeletal diseases and bone biopsies from individuals under examination for diseases such as osteoporosis and osteomalacia.
Authors: Isha Dev, Sofia Mehmood, Nancy Pleshko, Iyad Obeid, William Querido
Temple University
