Figure 7. a) Raman spectroscopic imaging of enamel surfaces. b) Raman analysis of enamel sagittal plane of molars.[105]Copyright 2021, Wiley-VCH.
Bone mineralization
The analysis of bones by Raman spectroscopy can reveal much more detailed information on relevant biomineralization. For example, the structural information of the femoral shaft, the structural information at the joints, the spatial distribution of the skull, the trabecular structure could be readily obtained, etc.[52,165,216,217] The effects of different diseases on bone physicochemical properties were analyzed by Raman spectroscopy (Figure 6 ). It is observed that Raman spectroscopy can provide specific mechanical information about bones, predict the physiological age of organisms and the risk of fracture, etc., through the combination of logical operation and algorithm.
Raman spectroscopy was also used to study the effect of the severity of osteoarthritis on the biological composition of the human tibial plateau osteochondral junction.[182] Through multivariate cluster analysis, calcified cartilage, subchondral bone plate, calcified cartilage, and non-calcified cartilage were identified (Figure 8 ).[182] The unsmooth bones lead to the difficulty of Raman imaging. Anders et al. solve the challenge brought by the inherent topology of this unique biological system by using the real-time focusing and tracking technology of continuous closed-loop feedback to optimize laser focusing.[218] In situ, analysis of organic and inorganic components of the biomineralization is possible despite surface height deviations of more than 100 μm in the femur. There are also some Raman spectroscopic studies focusing on bone diseases and the analysis of bone complications caused by other conditions, such as osteoporosis and type-Ⅱ diabetes.
The mineral apatite of cortical bone tissue and bone in healthy and ovariectomized (OVX) - induced osteoporosis in female mice was studied.[219] It was found that the lesion did not undergo significant amorphization, but the relative content of organic matter changed. Pankaj et al. show that type-Ⅱ diabetes and related therapeutic drugs can harm bone quality through Raman spectroscopy.[220,221] Some molecular biologists have studied the effects of different gene expressions on bone quality in mice by Raman spectroscopy, such as low-density lipoprotein receptor (LDLr) and plastin3 (PLS3).[48,222] From the level of disease to molecular biology, Raman spectroscopy is more commonly utilized to obtain the distinct structural information of the bone.
Brittleness, toughness, and mechanical strength of bone are important evaluation criteria for bone quality. Ozan et al. found that while the elastic deformation ability decreased, the increase of mineralization, crystallinity, and substitution degree of B-type carbonate were significantly related to the decrease of elastic deformation ability with age.[175] Osteogenesis imperfecta (OI) is a genetic disorder that manifests on a macroscopic scale as an increase in bone fragility. The Raman spectroscopy results showed a higher mineral-matrix ratio and lower crystallinity in OI samples, suggesting that OI samples have smaller but more abundant mineral crystals that can lead to increased bone fragility.[223] Besides, it was found that a decrease in the low-frequency component of the amide III band and an increase in the high-frequency component of the amide I band were found, indicating the rupture of collagen crosslinks.[224] This breakage of collagen’s secondary structure is affecting the mechanical properties of the bone.[49] It has been validated in mouse, rabbit, and human disease models.[49,148,225–229]Interestingly, based on the difference in bone quality caused by disease, some scholars took the Raman spectrum as the input of linear discriminant analysis (LDA) and evaluated that the linear support vector machine (LSVM) algorithm can successfully identify renal bone dystrophy.[230]
Although Raman spectroscopy can achieve bone quality assessment and mechanical performance analysis through deep learning and classification algorithms, the calculation results are unreliable due to the small sample size. Secondly, the current research on deep learning and classification algorithms for diagnosing minerals is relatively simple. To improve the accuracy of Raman spectroscopy in evaluating bone quality, more accurate deep learning and classification algorithms need to be developed.