X-Ray Imaging Could Improve Hemp Seed Germination

Industrial hemp can produce food, textiles, construction materials, cosmetics, and biocomposites. Before any of those products can reach the market, however, growers need seeds that can establish a healthy, reasonably uniform crop.

That is not always straightforward. Hemp seed lots can have low or inconsistent germination rates, while seeds from the same field may mature at different times. Exterior appearance also provides only limited information about whether the embryo and other tissues inside the seed developed properly.

A study¹ published in Industrial Crops and Products investigated whether X-ray microtomography could expose internal characteristics associated with germination. Researchers scanned seeds from four industrial hemp varieties, reconstructed their interiors in three dimensions, and then tracked what happened when each seed was planted.

The results suggest that germination potential is connected more closely to the condition of the internal seed tissue than to basic measurements such as overall width, height, or volume. This could eventually help hemp breeders and seed producers identify poor-quality seeds before planting, although the technology and predictive model remain at an early stage.

Why Hemp Seed Quality Is Difficult To Assess

A hemp seed is more complicated than its familiar outer shell suggests. Botanically, the structure planted by growers is an achene, a dry fruit containing the true seed. Its outer pericarp surrounds the embryo and endosperm, which supply the living tissue and stored energy needed for germination.

A seed can therefore appear intact from the outside while containing irregularities, empty spaces, or underdeveloped tissue internally. Conventional inspection may identify obvious cracks, discoloration, or physical damage, but it cannot fully reveal the structure enclosed by the pericarp.

External conditions add another layer of uncertainty. Temperature, moisture, growing substrate, storage conditions, and seed treatments can all affect germination. The new study focused instead on intrinsic characteristics, meaning structural properties already present inside the seed before it was planted.

This distinction matters commercially. If poor emergence is caused partly by detectable structural deficiencies, growers could potentially remove unsuitable seeds rather than compensating with higher seeding rates. Breeders could also select seed lots more efficiently and obtain more consistent results from field trials.

How Researchers Looked Inside Hemp Seeds

The researchers examined four certified Italian industrial hemp varieties: Carmaleonte, Codimono, CS, and Fibranova. They selected all four textile hemp varieties partly because their seed lots had relatively low germination rates. The reported germination rates for the original lots ranged from 30% for Carmaleonte to 54% for Codimono.

From these lots, the researchers randomly selected 25 seeds per variety, producing a sample of 100 seeds. Each seed was weighed and assigned an identification code so its X-ray measurements could later be matched with its individual germination result.

The team scanned ten seeds at a time using a desktop X-ray microtomography system. Unlike an ordinary X-ray image, which compresses structures into a flat view, micro-CT collects projections from multiple angles. Software combines those projections into a three-dimensional representation of the object.

The scans were captured at a voxel size of 6.5 micrometres, providing enough detail to separate the pericarp from the seed tissue. Researchers measured characteristics including volume, surface area, porosity, pericarp thickness, tissue solidity, mean pore size, and fractal dimension.

After imaging, all 100 seeds underwent a standardized laboratory germination test. The researchers classified each result as a normal seedling, an abnormal seedling, or an ungerminated seed.

What Happened When The Hemp Seeds Were Planted

The test produced a nearly even split between normal seedlings and seeds that failed to germinate. Of the 100 scanned seeds, 42 developed into normal seedlings, 17 produced abnormal seedlings, and 41 remained ungerminated.

Several measurements differed among the outcomes, but most of the meaningful differences were found within the seed tissue rather than in the dimensions of the whole achene. Three findings were especially informative:

• Normally germinating seeds had greater tissue solidity and better pericarp integrity.
• Ungerminated seeds had greater porosity, larger average pores, and more irregular tissue.
• Basic exterior measurements generally did not separate the three outcomes.

Solidity was the strongest discriminator between normal seedlings and ungerminated seeds. In simple terms, solidity describes how completely and regularly the internal tissue occupies its expected three-dimensional shape. Seeds producing normal seedlings had an average tissue solidity of 0.84, compared with 0.72 among ungerminated seeds.

Fractal dimension was the second most informative characteristic. This measurement reflects structural complexity and irregularity. Ungerminated seeds had a higher average fractal dimension than seeds producing normal seedlings, suggesting that a more uneven internal tissue surface was associated with failure to germinate.

Porosity followed the same general pattern. Normal seedlings came from seeds with an average tissue porosity of 0.12, while ungerminated seeds averaged 0.20. The average pore size was also substantially higher among ungerminated seeds, at 154.61 micrometres, compared with 109.51 micrometres among normal seedlings.

Internal Hemp Seed Structure Matters More Than Size

One of the study’s most useful conclusions is that a larger seed is not automatically a better seed. The researchers measured multiple aspects of overall size and shape, including width, height, diameter, surface area, and volume. Most did not differ significantly among germination outcomes.

Seed weight showed some relationship with germination, but internal tissue organization provided more useful information. This means visual grading based largely on seed dimensions may overlook defects that influence whether germination can begin and continue normally.

The finding also helps explain why two seeds that appear similar can behave differently after planting. Germination depends on the embryo having access to sufficient, properly organized tissue. Large pores and irregular internal surfaces may indicate incomplete development, deterioration, or structural damage that is hidden beneath an apparently normal shell.

For growers, this creates a distinction between seed quantity and viable planting value. A seed lot may contain the expected number or mass of seeds while delivering fewer healthy plants than anticipated. Technology that can examine internal quality could eventually allow sellers and buyers to evaluate lots by likely field performance rather than appearance alone.

X-Ray Screening Could Support Hemp Breeding

The immediate value of this research is not a ready-made commercial sorting machine. It is identifying measurable traits that future screening systems could target.

Breeding programs often need to evaluate many seeds and plants across successive generations. Low-quality seeds consume greenhouse space, field area, labour, and time before their weaknesses become apparent. Nondestructive imaging could allow breeders to identify structurally promising seeds while preserving them for germination and further breeding.

That nondestructive aspect separates imaging from tests that require cutting seeds open. Each scanned seed in this experiment could be planted afterward, allowing researchers to connect its three-dimensional anatomy directly to the seedling it produced.

In a mature system, internal imaging could complement existing germination tests rather than replace them. Conventional tests measure the actual performance of a representative sample under controlled conditions. Imaging could add information about why a lot is underperforming and potentially help sort individual seeds within that lot.

Why This Is Not Yet A Commercial Hemp Seed Sorter

The study also revealed important limitations. Its predictive model correctly classified 66% of the results overall, but that number overstated its reliability because the three outcome groups were not equally represented. Balanced accuracy, which gives the groups equal importance, was only 55.24%.

The system was particularly weak at identifying seeds that would produce abnormal seedlings. It correctly classified only two of the 17 abnormal outcomes, mislabeling most of them as normal. A screening system that classifies unhealthy seedlings as normal would have limited value in applications requiring dependable crop establishment.

The sample was also small and limited to four related industrial hemp varieties. Differences among cultivars, production regions, maturity levels, storage conditions, and seed ages could affect internal morphology. A model trained on 100 seeds cannot yet establish thresholds that apply across the global hemp industry.

Speed presents another obstacle. Acquiring the projections for each group of ten seeds took approximately one hour and 40 minutes. That is suitable for detailed laboratory research, but not for processors handling thousands or millions of seeds.

The researchers suggest investigating faster two-dimensional X-ray imaging in future studies. If the most important three-dimensional traits have recognizable two-dimensional signatures, conventional radiography could provide a more practical route to high-throughput sorting.

X-Ray Exposure Did Not Reduce Germination

Using X-rays on living seeds raises an obvious concern: the inspection itself could alter the result. To test this, the researchers compared the scanned seeds with control and separately irradiated groups.

They found no statistically significant difference in germination outcomes between the micro-CT seeds and the controls. Results also did not differ significantly between the micro-CT group and seeds receiving the same absorbed dose at a higher dose rate.

This does not prove that every X-ray protocol is harmless to hemp seeds. Biological effects can vary with total dose, dose rate, equipment, and exposure time. It does show that the protocol used in this experiment did not measurably reduce germination, supporting the feasibility of nondestructive X-ray assessment.

The higher-dose-rate group also germinated slightly faster on average, although the difference was not sufficient to establish a priming effect. Further research would be needed before X-ray exposure could be considered a potential seed treatment rather than only an imaging method.

The Future Of Hemp Seed Quality Control

This research points toward a future in which hemp seed quality is evaluated from the inside out. Instead of relying primarily on exterior appearance, weight, and sample-based germination rates, producers could eventually use imaging to detect the internal signatures associated with failure.

The strongest near-term opportunity may be a hybrid system. Fast X-ray imaging could capture internal structures, automated image analysis could measure solidity and irregularity, and a classification model could flag the seeds most likely to be nonviable. Human technicians and conventional germination tests would remain part of the quality-control process while larger datasets improve the models.

More research is needed across a wider range of hemp varieties and growing conditions. Future studies will also need to determine whether laboratory classifications translate into reliable field emergence, where temperature, soil moisture, disease, and planting depth add further uncertainty.

Even with those limitations, the study establishes an important principle: the exterior of a hemp seed does not tell its full story. Internal tissue solidity, porosity, and structural complexity contain information about whether that seed is likely to become a healthy plant. Turning those signals into fast commercial tools could help breeders improve seed lots, help growers establish more uniform crops, and reduce resources spent planting seeds unlikely to succeed.

References:

^{1. Marsala, S. M., Gargiulo, L., Formisano, G., & Mele, G. (2026). The role of seed structure on hemp germination: A study based on X-ray microtomography and 3D image analysis. Industrial Crops and Products, 251, 124233. https://doi.org/10.1016/j.indcrop.2026.124233}