Seafood lovers know what makes tuna sashimi and sushi so tasty is the fish’s fatty marbling; for the restaurant industry, it’s the level of fattiness, too, that’s used to judge its quality and pricing.
Fujitsu, one of Japan’s largest and oldest tech companies, has developed a machine that can tell which highly prized tuna fish are the tastiest and therefore able to command high prices from competing chefs.
A conveyor belt transports a whole frozen tuna fish, about a metre in size, into a machine dubbed “Sonofai” that beams ultrasound waves.
Sensors pick up the waves to draw a zigzagging diagram on a screen to indicate the fattiness of the fish.
“If there is a lot of fat, then it’s harder for the soundwaves to pass, so the waveforms will be displayed smaller. And for this one, AI analysed it and concluded that it is extremely fatty,” Yasutomi Suguru, a researcher at Fujitsu, said.
Ultrasound and AI
To inspect and accurately evaluate the fat content of tuna, Sonofai uses both ultrasound and artificial intelligence (AI).
It is based on the same technology as medical ultrasound scans, where high-frequency inaudible sound waves are absorbed or bounced back to examine the fattiness of the human liver, which is linked to diabetes and other health problems.
It can also create video-like images of things that aren’t visible, like a foetus inside a mother’s womb.
Fatty meat absorbs fewer sound waves than lean meat, and AI sorts real data from misleading “noise,” or irregularities.
“Since the tuna is frozen, when something is emitted and then returns, it often doesn’t come back correctly. We refer to that as ‘noise.’ Our [AI] technology can remove that noise and accurately analyse the data,” said Hideto Okada, the head of AI Strategy and Business Development at Fujitsu.
The developers say the new tech is safer, more sanitary, and efficient, as wholesalers no longer need to cut the fish open to be assured of its quality.
Tuna is a sought-after ingredient in Japan’s high-end culinary scene. The fat content of the red fish is integral in defining its superiority and pricing.
Ultrasound is different from a CT scan or X-ray as it doesn’t use radiation, which can harm tissue.
New tool for the gastronomy scene
While beef has a grading system for fat and anticipated flavour, gauging the quality of tuna is new, according to Okada.
For sushi chefs like Junichi Kurita, the new tech will allow for experimentation while assuring quality.
“The reason we don’t use albacore tuna – bincho-maguro – at our Edomae sushi restaurants is because the fat content varies way too much from fish to fish,” Kurita, who works as a sushi chef at Tokyo restaurant Kanda Edo-gin, said.
“It’s not typically something that’s highly valued. But if it becomes clear which ones actually have good fat content, I think it will make it easier for us and restaurants to purchase it,” he added.
Usually, several people assess how fatty a tuna is by cutting the tail with a giant saw-like knife, an operation that takes about 60 seconds per fish. Sonofai boasts a processing time of a mere 12 seconds.
The machine is set to go on sale first in Japan in June for around €182,250, with plans to expand to other countries, with the main target market being fish-processing outfits and fishing organisations.
The developers are saying they’ll work on future upgrades to test for freshness, firmness, and other characteristics of tuna and other fish varieties.
“Freshness, texture, and density – being able to understand those internal conditions is, I think, a significant step,” said Hisashi Ishida, the president of Sonofai.
For more on this story, watch the video in the media player above.
