The impetus for the work stems from a persistent gap in grain-quality research. Endosperm cavities are known to serve as a phenotypic indicator for assessing maize yield and quality, yet studies of these structures remain scarce. The reasons are practical: a maize kernel is small, opaque, and internally complex, and traditional approaches such as manual sectioning require physically slicing the grain into thin layers, a destructive process that collapses the very spatial relationships researchers want to understand. Sectioning also samples only a handful of two-dimensional planes, making it impossible to know whether a cavity seen in one slice connects to structures elsewhere in the kernel. The Nanjing team, led by corresponding author Dong Jiang and including Xuebo Yin, Yu Jiang, Shengxi Hou, and colleagues, recognized that X-ray micro-computed tomography could overcome these limitations. The technique works by rotating a sample within an X-ray beam and capturing hundreds to thousands of projection images from different angles; computational reconstruction algorithms then combine these projections into a three-dimensional volumetric map in which each voxel, the three-dimensional analogue of a pixel, carries an attenuation value proportional to local material density. Air-filled cavities, being far less dense than the surrounding endosperm tissue, appear as sharply defined low-attenuation regions that can be segmented and measured.

Using this approach, the researchers extracted morphological parameters and spatial configurations of endosperm cavities from kernels of multiple maize varieties, transforming what had been an anecdotal observation into a dataset that can be compared across genotypes. The results uncovered two distinct cavity types with strikingly different geometries and locations. Embryo-adjacent cavities, abbreviated EACs, are distributed in a conical pattern around the embryo, wrapping around the germ in a configuration that suggests they form systematically during kernel development rather than arising as random defects. Internal endosperm cavities, or IECs, by contrast, are located in the floury endosperm at the tip region of the kernel and exhibit a distinctive boat-shaped morphology. The two types are not equally abundant: the volume ratio of EACs to IECs is approximately 5 to 1, meaning the conical cavities surrounding the embryo account for the overwhelming majority of the air volume inside a typical kernel.

To describe where these cavities sit within the grain, the team established a coordinate system with the kernel length axis perpendicular to the horizontal plane, allowing each cavity population to be assigned a precise three-dimensional address. Under this framework, the internal endosperm cavities cluster at coordinates of approximately x = 3.5 millimeters, y = 2.1 millimeters, and z = 1.1 millimeters, while the embryo-adjacent cavities center around x = 2.5 millimeters, y = 2.3 millimeters, and z = 7.1 millimeters, placing the two populations at opposite ends of the kernel along its long axis. This spatial segregation is more than a curiosity. It implies that the cavities originate from different developmental processes or tissue environments, one linked to the region surrounding the embryo and the other to the loosely packed floury endosperm at the kernel tip, and it provides breeders and food scientists with a map of where to look when the internal structure of the grain matters for a given trait.

Quantitatively, the study found that the average volume of endosperm cavities across the kernels examined was 4.1 cubic millimeters, with kernel porosity, the fraction of the kernel volume occupied by air, ranging from 0.4 percent to 3.3 percent across varieties. That range is itself significant, because it demonstrates that cavity content is a heritable, varietal characteristic rather than a fixed property of maize. Indeed, the analysis revealed significant differences in endosperm cavity characteristics among the different varieties tested, and these parameters exhibited highly significant positive correlations with kernel volume, kernel thickness, and cavity surface density, among other traits. In practical terms, bigger and thicker kernels tend to harbor larger and more extensive cavity systems, meaning that breeders selecting for plumpness may be indirectly selecting for the internal air architecture that the CT scans reveal.

The correlations point toward functional consequences. Kernel hardness and plumpness are central to maize quality, affecting everything from milling behavior and breakage resistance during handling to cooking properties and susceptibility to storage pests and fungal infection. Since the cavities represent discontinuities within the endosperm, their size, shape, and distribution plausibly influence how stress propagates through the grain when it is struck, dried, or ground. The new dataset gives researchers, for the first time, a way to test such structure-function hypotheses directly, by comparing cavity metrics from CT reconstructions with mechanical and processing measurements on the same varieties. It also opens the door to digital phenotyping, in which a maize breeding program could screen hundreds or thousands of genotypes for internal grain architecture without destructive sampling, accelerating the selection of lines with optimal combinations of external morphology and internal structure.

Importantly, the authors did not dismiss the older, simpler method. Manual sectioning, for all its inability to capture the three-dimensional features of endosperm cavities, retains genuine advantages: it is operationally simple, requires inexpensive equipment, and allows rapid data extraction. The study confirms that the two-dimensional information gleaned from sectioning reflects, to some extent, the cavity characteristics of different maize varieties, validating its continued use as a rough screening tool. But the comparison also clarifies what sectioning cannot do. It cannot reveal the conical arrangement of EACs around the embryo, the boat shape of IECs at the kernel tip, or the precise volumetric relationships between the two cavity populations. For that level of detail, micro-CT is now the reference standard.

The technical achievement rests on the broader maturation of X-ray imaging in plant science. Micro-computed tomography has increasingly found applications in agriculture, from mapping root systems in soil to measuring structural parameters related to kernel breakage, and the new study extends this toolkit to a structure that had largely escaped quantitative attention. The workflow the researchers developed, involving three-dimensional reconstruction, segmentation of air spaces, extraction of morphological parameters, and statistical comparison across varieties, is documented in enough detail to be adopted by other laboratories, and the paper is accompanied by supplementary videos that allow readers to tour the internal landscape of a maize kernel directly. Funding for the work came from the Collaborative Innovation Center for Modern Crop Production, co-sponsored by Province and Ministry, and the National Natural Science Foundation of China.

The findings lay the groundwork for what the authors describe as research into maize grain digital characterization and the relationship between grain structure and function. In the near term, that could mean new breeding targets: if a specific cavity configuration correlates with resistance to breakage or superior milling yield, breeders could incorporate cavity metrics into selection pipelines alongside traditional traits. In the longer term, understanding why the embryo is encircled by a conical air system, and why the floury tip develops boat-shaped voids, may illuminate the developmental biology of endosperm formation itself, including how drying, filling, and maturation leave their signatures in the mature kernel. Because the technique is non-destructive, the same kernels scanned for cavity analysis can subsequently be used for germination tests, chemical assays, or mechanical testing, enabling fully integrated studies of a single grain.

For a crop grown on more than 190 million hectares worldwide and central to food, feed, and biofuel systems, even modest improvements in quality assessment can translate into substantial economic value. The Nanjing team’s achievement demonstrates that a structure invisible to the naked eye, hidden within an ordinary-looking kernel, can now be measured with millimeter-scale precision and correlated with the traits that farmers, millers, and consumers ultimately care about. As micro-CT scanning becomes faster and more affordable, the internal anatomy of the maize kernel, once accessible only through the destructive blade of a microtome, is set to become a routine dimension of crop improvement, adding a genuinely three-dimensional chapter to the science of grain quality.