Abstract: Fusarium head blight (FHB), caused primarily by Fusarium graminearum, is a major fungal disease threatening wheat production and food safety worldwide. In addition to causing substantial yield losses, FHB can lead to the accumulation of deoxynivalenol (DON), a toxic mycotoxin that poses a potential risk to the quality and safety of wheat and its derived products. During FHB development, pathogen invasion, DON accumulation, kernel structural deterioration, and nutrient depletion occur simultaneously and are closely interconnected. However, the relationships among these processes, particularly their spatial characteristics at the single-kernel level, remain insufficiently understood, limiting both accurate evaluation of disease resistance and rapid monitoring of DON contamination. In this study, Fhb7 near-isogenic lines (NILs) with contrasting resistance backgrounds were employed as a genetic framework to investigate the mechanisms underlying FHB resistance and DON formation. An integrated multi-scale analytical strategy combining near-infrared hyperspectral imaging (NIR-HSI), scanning electron microscopy (SEM), and synchrotron radiation Fourier-transform infrared microspectroscopy (SR-FTIR) was established to characterize infection-induced changes from spectral, structural, and biochemical perspectives. This approach enabled the relationships among pathogen infection, kernel structural damage, nutrient changes, and DON accumulation to be examined across complementary analytical scales. The results demonstrated that the presence of the Fhb7 allele reduced diseased spikelet incidence and the proportion of highly DON-contaminated kernels, indicating a stable capacity to restrict pathogen spread and suppress excessive toxin accumulation. NIR-HSI analysis revealed distinct spectral differences between Fhb7⁺ NILs and Fhb7^- NILs, reflecting their differential responses to pathogen infection and DON accumulation. Among the spectral features, 1013, 1195, and 1463 nm were identified as key sensitive wavelengths associated with pathogen infection and DON accumulation. These characteristic spectral responses provided important information for differentiating wheat materials with contrasting Fhb7 genetic backgrounds. Based on these identified spectral signatures, NIR-HSI-based predictive models were further developed for Fhb7 genotype discrimination and DON quantification. The classification model using shallow wavelet transformation achieved excellent discrimination performance, with classification accuracies of 0.98 and 0.99 for the validation and cross-validation sets, respectively, demonstrating the feasibility of rapid and non-destructive identification of Fhb7 genetic backgrounds. Meanwhile, the DON prediction model optimized by Box-Cox transformation showed reliable performance, achieving a validation coefficient of determination of 0.81 and a ratio of performance to deviation of 2.18, indicating reliable predictive performance and highlighting the potential of NIR-HSI for rapid DON screening and quantitative evaluation at the single-kernel level. Multi-scale structural and biochemical analyses further clarified the biological basis of the spectral and toxin-related differences. SEM observations revealed a progressive deterioration of kernel microstructure during pathogen invasion and DON accumulation, characterized by surface degradation, increased tissue porosity, extensive hyphal colonization, and progressive destruction of internal kernel structures, ultimately accompanied by severe embryo structural collapse. These structural alterations provided direct microscopic evidence of infection-induced damage and were closely associated with the spectral changes detected by NIR-HSI. SR-FTIR chemical imaging further demonstrated that fungal infection caused systematic depletion of major nutritional components, including lipids at 2926 cm⁻¹, proteins at 1650 cm⁻¹, and polysaccharides at 1080 cm⁻¹, with pronounced spatial heterogeneity across kernel tissues. Compared with Fhb7^- NILs, Fhb7⁺ NILs maintained higher levels of nutrient-associated spectral signals, suggesting that Fhb7 contributed to the preservation of host tissue integrity and reduced infection-induced nutrient loss. Notably, a characteristic absorption peak at 1710 cm⁻¹ was exclusively detected in Fhb7⁺ NILs, providing in situ spectroscopic evidence supporting DON-glutathione (DON-GSH) detoxification mediated by the glutathione transferase activity associated with Fhb7. Taken together, these findings establish a cross-scale framework linking pathogen invasion, DON accumulation, nutrient depletion, structural damage, and spectral responses during FHB development. The results indicate that Fhb7 mediated resistance involves coordinated restriction of pathogen expansion, preservation of kernel structural integrity, protection of nutritional components, and activation of DON detoxification pathways. This study provides mechanistic insights into the relationship between FHB resistance and DON accumulation and presents a non-destructive strategy for Fhb7 resistance screening and DON contamination monitoring, with potential applications in disease-resistant wheat breeding and grain safety management.