Physics-Informed Fourier Thermography Transformer for Non-Destructive Detection and Quantification of Braeburn Browning Disorder in Apple

Inside a crate of apples, a hidden crisis often unfolds. The fruit may look perfect on the outside, with smooth skin and a vibrant blush, yet deep within the flesh, a disorder known as Braeburn browning is silently spreading. This condition, which turns the crisp white tissue into a diffuse brown mush, is a nightmare for growers and retailers because it cannot be seen by the naked eye. Traditional sorting machines, which rely on cameras to spot surface blemishes or color changes, are useless here; they simply pass the damaged fruit along, only for the defect to be discovered later by a consumer or a packer who cuts the apple open. The challenge is not just finding the bad apples, but understanding why they are bad without destroying the good ones. To solve this, scientists have turned to the physics of heat. Every object holds thermal energy, and when that energy moves, it tells a story about what lies beneath the surface. If you warm or cool an object quickly, the way its surface temperature returns to normal depends on the material inside. A solid, healthy piece of tissue conducts heat differently than a collapsed, damaged one. By watching how an apple cools down and then slowly warms back up, it is possible to see the invisible damage before it ever becomes visible.

A team of researchers set out to build a system that could read this thermal story for Braeburn apples. They gathered 640 apples from eight different orchards, harvested at two different times, and subjected them to various storage conditions designed to create a wide range of internal damage. Some apples were stored in normal air, while others were kept in special atmospheres with high carbon dioxide or low oxygen, conditions known to trigger the browning disorder. The goal was to create a population of fruit that ranged from perfectly healthy to severely damaged, ensuring the test would be rigorous. Once the storage period ended, each apple was placed in a controlled environment and given a brief, gentle blast of cool air. This cooling pulse was mild enough to avoid injuring the fruit but strong enough to create a temporary temperature difference. As the apples sat there, a special infrared camera recorded their surface temperature every tenth of a second for ninety seconds. This created a video of the apples slowly recovering their warmth.

The researchers did not just look at the raw video. Instead, they translated the changing temperature of every single point on the apple's skin into a mathematical language that describes how the heat moved over time. They focused on two specific aspects of this movement: the strength of the temperature change and the timing of that change. Timing, or phase, proved to be the most powerful clue. Just as a clock hand moves at a steady pace regardless of how bright the room is, the timing of the heat recovery remained consistent even if the cooling blast was slightly uneven or the apple's skin had a different shine. This timing shifted noticeably for damaged fruit; the heat took longer to return to normal in areas where the tissue had collapsed. To make sense of this complex data, the team built a computer model that acted like a highly trained observer. This model looked at the timing maps alongside standard color photos of the apple's skin. It learned to ignore the superficial details, like the red blush or the size of the apple, and focused entirely on the subtle delays in heat flow that signaled internal rot.

The results were striking. The system could distinguish between four levels of damage with an accuracy of nearly 92 percent. It correctly identified apples that were healthy, those with trace amounts of browning, those with moderate damage, and those that were severely ruined. More importantly, it could estimate exactly how much of the apple's flesh was affected, with an error margin of less than three percentage points. When the researchers tested the system on apples from orchards and harvest dates it had never seen before, it performed just as well, proving that it had learned the physics of the damage rather than just memorizing the look of specific trees. The system also proved robust against real-world imperfections. Even when the cooling air was slightly weaker or stronger than intended, or when the apple's skin was slightly damp, the model maintained its high accuracy. This stability is crucial because a commercial sorting line is never perfectly uniform.

The study also clarified what this technology is not. It is not a magic wand that can see through the apple in three dimensions like a medical scan, nor does it replace the need for careful storage management. The researchers found that the system works best because it detects the specific way damaged tissue slows down heat flow, a physical change that happens before the color changes. While other methods like X-rays can see inside fruit, they are expensive and slow. This thermal method is passive, safe, and fast enough to be integrated into a sorting line, provided the fruit can be cooled and imaged in a batch. The researchers emphasized that the system does not rely on the apple's color or the specific storage conditions it was kept in; it relies on the fruit's own physical response to a gentle thermal probe.

In the end, this work offers a practical bridge between the destructive testing that currently dominates quality control and the need for a non-destructive way to screen every single apple. By converting a brief cooling pulse into a calibrated measure of internal health, the system turns a hidden physiological failure into a visible signal. It allows growers to sort fruit based on its true internal condition rather than its deceptive exterior. While the technology still needs to be tested across more seasons and different apple varieties to be fully ready for widespread use, the core finding is clear: the way an apple recovers from a chill reveals its secrets, and with the right tools, those secrets can be read before the first bite is taken.