Announcing new publications from Opto-Electronic Advances. DOI 10.29026/oea.2026.250312.
Bone mineral density (BMD) has long been the “gold standard” for assessing fracture risk. However, clinicians face contradictions when treating patients with type 2 diabetes mellitus (T2DM). Although many patients have normal or above average BMD, they suffer from significantly high fracture rates. This deceptive phenomenon led researchers to realize that actual bone toughness is not just the accumulation of inorganic minerals. It highly depends on the internal microstructure, the distribution of organic substances, and the complex bone cell network. Deterioration of this microstructure compromises bone stability, even though the overall bone mass remains seemingly intact.
Accurately capturing these micrometer- or nanometer-scale lesions using traditional diagnostic tools is extremely difficult. Current histological methods rely on tedious chemical staining and time-consuming demineralization, which are not only labor-intensive but also irreversibly damage the fragile native microenvironment of bone. Additionally, traditional light microscopy often provides a fragmented view, visualizing single components such as collagen fibers or specific minerals, but failing to capture precise 3D spatial interactions between multiple key molecules. This makes it nearly impossible to piece together a complete pathological picture of bone fragility caused by T2DM.
To address this medical challenge, multimodal nonlinear optical (NLO) microscopy emerges as an optical breakthrough to unravel the mysteries of “invisible” diabetic bone damage. By exploiting the molecule’s inherent nonlinear optical effects and vibrational spectral signatures, this technology enables in situ “photobiopsy” of bone tissue without the need for extrinsic labeling or destructive demineralization. By integrating stimulated Raman scattering (SRS), second harmonic generation (SHG), and two-photon excited fluorescence (TPEF) into a single imaging platform, researchers can create high-resolution maps that include proteins, lipids, and collagen fibrils. Nondestructively acquiring this multidimensional optical data establishes a solid technological foundation for comprehensively understanding the micromechanisms of complex bone diseases.
To overcome the diagnostic limitations of single-component imaging, a research team led by Professor Ting Li of the Chinese Academy of Medical Sciences and Peking Union Medical College, in collaboration with Beihang University and Shanghai East Hospital, innovatively combined multichannel nonlinear optical imaging with artificial intelligence (AI). Multichannel images contain high-dimensional microscopic pathological data that is invisible to the human eye (Figure 1). The research team used machine learning algorithms to extract spatial texture features from each channel and build a classification model. The results demonstrated that by fusing information from three core optical channels (proteins, autofluorescent metabolites, and phosphates), the AI model could sharply capture the optical heterogeneity of T2DM bone tissue and achieve an excellent diagnostic accuracy of 93.56%. This overall exceeds the typical accuracy of traditional single-channel optical diagnostics, which is approximately 70% (Figure 2).
Through explainable AI analysis, the research team for the first time identified spatial deterioration signatures unique to human T2DM bone tissue. Proteins are typically clustered with high contrast and fine detail within healthy bone cell networks, but in T2DM patients pathological reorganization occurs, resulting in an abnormally uniform and smooth spatial optical distribution. The researchers defined this unique degradation of optical texture as “spatial homogenization of proteins.” This phenomenon is thought to reflect disruption of bone cell communication networks and loss of structural gradients, and serves as a powerful “optical pathological label” for diabetic bone deterioration (Figure 3).
Naturally, further exploration is required to translate this technology into broader clinical applications. Future studies may expand the sample size and incorporate multicenter clinical data to increase the generalizability of the model across diverse populations. Additionally, combining this approach with immunohistochemistry, proteomics, or biomechanical testing can help validate the molecular mechanisms underlying “spatial homogenization of proteins.” Ultimately, this study demonstrates the great potential of combining multimodal NLO microscopy and AI in biomedical research. This not only provides a new optical imaging tool to identify T2DM-related bone quality changes, but also opens new avenues for investigating micropathological changes in complex diseases using advanced optical imaging.
This research was supported by the National Key Research and Development Program of China (Nos. 2025YFE0204500, 2025YFE0218400, 2025YFC2427800, 2025ZD0548504), the Key Research Program of the National Natural Science Foundation of China (No. 92570204), and the PUMC Excellent Talents Program. (2025-I2M-XHJC-045).
Keywords: label-free nonlinear optical imaging, type 2 diabetes, bone quality disorders, explainable AI, multimodal integration
/Open to the public. This material from the original organization/author may be of a contemporary nature and has been edited for clarity, style, and length. Mirage.News does not take any institutional position or position, and all views, positions, and conclusions expressed herein are those of the authors alone. Read the full text here.
