Global food security is threatened by the projected population increase to 9.1 billion by 2050, demanding a 70% increase in cereal production. Post-harvest losses, estimated at 10-20% in developing and 1-2% in developed countries, significantly impact food availability and farmers' livelihoods. These losses encompass both quantitative (dry mass loss, insect/fungal damage) and qualitative (nutritional loss, reduced germination, mycotoxin contamination) aspects. Moisture content is a critical factor influencing storage life; high moisture levels promote fungal growth and spoilage. While macroscale studies have addressed storage optimization, microscale understanding of seed structure and its role in spoilage resistance is crucial. This study aims to address this gap by using high-resolution synchrotron phase contrast X-ray microtomography (SR-µCT) to visualize the internal structures of spring and durum wheat during storage, providing detailed 3D data previously unavailable.

Numerous studies have investigated wheat storage at both macro and micro levels, focusing on safe storage limits, loss reduction, quality assessment, and guideline development. Macro-level studies often involve large-scale storage simulations, examining the interplay of abiotic (moisture, temperature, humidity) and biotic (insects, fungi) factors on grain quality. Micro-level studies utilize destructive and non-destructive methods like X-rays, hyperspectral imaging, and thermal imaging to assess defects. However, limitations exist in conventional X-ray imaging, including low contrast, long scanning times, and non-selective wavelengths. Synchrotron X-ray imaging, offering higher resolution and edge enhancement, proves advantageous in characterizing low-density materials, distinguishing spoilage from moisture and air, which is difficult with absorption X-ray imaging. Previous research has highlighted the importance of crease features in disease susceptibility, making high-resolution imaging particularly relevant.

Six commercially available wheat varieties (three spring, three durum) were used. 500-gram samples of each variety were conditioned to 17% moisture content (wet basis) and stored in airtight jars at 22-25°C for five weeks. Samples were also frozen at -18°C. SR-µCT scanning was performed at the Canadian Light Source using a filtered white beam of approximately 20 keV. 15-20 kernels per variety and replicate were scanned, with images acquired at 0.06° increments for 180°. The sample-detector distance was 5 cm. Image processing involved dark and flat field corrections, phase retrieval using Paganin's method, stitching of image slices, and binning to reduce data size. Segmentation was performed using interactive thresholding and morphological operations to separate seeds from air spaces. Avizo and ORS Dragonfly software were used for 3D image analysis, measuring seed volume, porosity, and the volume of damaged areas due to spoilage.

Visual inspection revealed varying levels of deterioration in both spring and durum wheat varieties after five weeks of storage, with durum showing more significant damage. SR-µCT revealed distinctive features in seed structure, size, and shape between the two wheat classes. Durum wheat varieties exhibited a higher prevalence of cracks in control samples compared to spring wheat. The endosperm was found to be more damaged in durum varieties (especially AAC Spitfire and CDC Defy) than spring varieties. Seed coat damage was also more pronounced in durum wheat. Fungal deterioration was more extensive in durum wheat varieties from three weeks onwards, and this was clearly visible in the images. Freezing samples at five weeks did not alter the existing structure, indicating that the observed changes primarily resulted from spoilage. Image segmentation and 3D analysis quantified the volume of spoilage and air gaps within the seeds. Durum wheat consistently showed a significantly higher volume of segmented deterioration and air space compared to spring wheat. AAC Spitfire exhibited the maximum spoilage volume among all varieties. The germ area was identified as the primary site of fungal infection, leading to deterioration that spread along cracks and damaged areas. Histogram analysis of gray values showed significant shifts towards lower values in spoiled samples, particularly in AAC Spitfire, indicating changes in the seed structure due to fungal infection. The best-performing spring wheat variety, Faller, had a tight closed crease and fewer endosperm imperfections.

The study's findings highlight the variability in storage performance among wheat varieties within the same class. The observed differences in spoilage levels strongly correlated with the initial condition of the control samples. The SR-µCT effectively differentiated between phases with low X-ray density differences, enabling the segmentation of spoilage and air spaces, which are difficult to distinguish using absorption X-ray imaging. The precise measurement of porosity, previously inaccessible with laboratory methods, demonstrates the technique's potential. The identification of the germ area as a primary infection site suggests vulnerabilities that could be targeted in future breeding programs or storage strategies. The findings support the recommendation of using high-resolution imaging of dry seeds prior to storage to predict their behavior and prevent post-harvest losses. The high variability observed emphasizes the need for variety-specific storage guidelines and breeding programs aiming for enhanced storage stability.

This study successfully employed synchrotron phase contrast X-ray microtomography to characterize the internal structure of spring and durum wheat during storage. The results revealed significant differences in spoilage between wheat types and varieties, with durum wheat showing greater susceptibility to deterioration. The high-resolution 3D imaging data provides crucial insights for optimizing post-harvest storage strategies, improving wheat breeding programs, and developing more accurate predictive models. Future research could focus on applying machine learning algorithms to analyze crease characteristics and their impact on storage performance.

The study was conducted under controlled laboratory conditions, which may not fully reflect the complexities of real-world storage environments. The relatively small sample size of wheat kernels per variety might limit the statistical generalizability of some findings. The study focused on specific varieties and might not be directly generalizable to all wheat cultivars. Further investigation is needed to determine the long-term effects of freezing on seed viability and germination.