Entertainment

AI designed to distinguish between types of pastries identifies cancer cells with 99% accuracy 


Artificial intelligence designed to recognize different type of pastries could be a vital tool in the medical world.

BakeryScan, developed by Japan-based Brain Co., scans baked good on a tray with a camera and uploads the official name of each to a system for easy checkout at a bakery – but scientists found it can also identify cancer.

A doctor from the Louise Pasteur Center for Medical Research in Kyoto had the the system revised to spot cancerous cells on a microscope slide with 99 percent accuracy.

Instead of investigating donut holes and bread ridges, the redesigned system, called Cyto-AisCAN, analyzes a urinary cell to identify and measure its nucleus to determine if it is diseased.

Scroll down for video 

BakeryScan, developed by Japan-based Brain Co., scans baked good on a tray with a camera and uploads the official name of each to a system for easy checkout at a bakery – but scientists found it can also identify cancer.

BakeryScan, first released in 2013, was designed by computer system engineer Hisashi Kambe who sold the innovation to Brain Co.

It is currently used by more than 400 retail shops across Japan and each unit costs $20,000.

BakeryScan works through a camera that is mounted above a backlit checkout tray.

Customers place their selections on the tray and then the camera analyzes the bread or pastries, cataloging their size, shape and color to match them with one of up to 100 different types stored in the checkout system.

A doctor from the Louise Pasteur Center for Medical Research in Kyoto had the the system revised to spot cancerous cells on a microscope slide with 99 percent accuracy

A doctor from the Louise Pasteur Center for Medical Research in Kyoto had the the system revised to spot cancerous cells on a microscope slide with 99 percent accuracy

The cashier confirms the match via a touchscreen display, and then the customer pays – an entire process that takes place in seconds.

Four years after BakeryScan was assisting retail shops, a doctor spotted the technology during a television show and pondered if it could do the same for cancer – he realized cancer cells look similar to bread when under a microscope, The New Yorker reports.

The system uses deep learning for object recognition and instead of differentiating baked goods, the doctor hoped the technology could save lives.

Identifying cancer cells to determine whether tumors are benign or malignant can be labor intensive. 

But having an AI assistant would dramatically speed up the process and lead to earlier diagnoses and more effective treatment for patients.

Brain Co revised BakeryScan for medical purposes to scan small microscope slides instead of puffy pastries.

Cyto-AiscAN was then on its way to two major hospitals in Kobe and Kyoto, where doctors tested and trained the system with cancerous cells.

Overtime, the AI was able to analyze an entire slide at once and not just each cell individually.

James Somers, the writer of The New York piece, shared: ‘The system was apparently working at ninety-nine percent accuracy.’

‘I asked Kambe how it worked—did it use deep learning? ‘Original way,’ he said. Then, with a huge smile, ‘Same as bread.’

AI has come a long way from identifying faces to now assisting doctors to help save lives.

Last year, a computer algorithm developed by British and US scientists has found AI was able to display a 1.2 percent reduction in the number of false positives and a 2.7 per cent reduction in false negatives.

Instead of investigating donut holes and bread ridges, the redesigned system, called Cyto-AISCAN, analyzes a urinary cell to identify and measure its nucleus to determine if it is diseased

Instead of investigating donut holes and bread ridges, the redesigned system, called Cyto-AISCAN, analyzes a urinary cell to identify and measure its nucleus to determine if it is diseased

The breakthrough has been likened to ‘a spell-check for writing email’ and could reduce the number of ‘false negatives’ that can lead to life threatening delays in treatment.

The technology has also taken off amid the coronavirus pandemic with many medical experts turning to the system for help.

University of Copenhagen researchers designed software that can tell whether you are likely to die from the virus using health data.

The team used a computer program with health data from 3,944 Danish COVID-19 patients, as well as any underlying conditions. 

They then trained it to look for patterns in a patients’ prior illness to determine the risk factors and potential outcome from Covid-19 and found that BMI, age and being male were the highest risk factors when it came to the likelihood of dying. 



Source link

Related Articles

Back to top button