딥러닝으로 영상의학 정확도를 높이다
Deep learning is a burgeoning field of artificial intelligence where algorithms learn from stored data to perform human-like tasks.
It's very useful for medical use,... but it's even more so... in the field of radiology.
Our Park Hee-jun explains how deep learning is being used in the field.
The modern workplace requires people to be more productive than ever.
The same is true for doctors and radiologists who are asked to study an immense amount of data,... and make judgements that can save life or take it away.
But new cognitive technologies are rapidly improving the precision and efficiency of radiology imagery -- one being deep learning.
(standup)
“Radiology is a significant application for deep learning,... which can even outperform humans in analyzing radiological images. For example, this center has developed a system that can analyze hundreds of thousands of patients’ images in a matter of seconds before making a diagnosis.”
After patients get their medical images taken at the hospital,
the data is sent to the system, which identifies their conditions through its accumulated data.
(Korean)
"Deep learning picks up details from big data that human beings can easily miss. This technology can do things that are hard for people,... and combined with radiology, it can make the most out of big data."
One disease where deep learning can make a huge difference is the respiratory disease pneumoconiosis, which is caused by inhaling dust and often affects manual laborers.
Confirming whether or not a patient has it takes a lot of time and work,... and misdiagnoses are common.
In Korea, the government offers compensation for people who contract the disease from industrial accidents,... so accurate screening is important.
A research team at Seoul St. Mary's Hospital has developed a system using deep learning to improve diagnoses.
After examining more than 12-hundred images of patients with pneumoconiosis,... the team succeeded in creating a system with an accuracy rate of 95 percent.
But 95 percent still leaves room for improvement.
To make sure that speed doesn't cost lives, experts in the field say they need more access to sample images.
(Korean)
"Korea has an extremely strict policy on protecting personal information. But the precision of deep learning systems depends on the amount of real radiological images available. I think providing this data will help take Korea's deep learning technology to the next level."
More data would improve deep learning and expand the practical uses of the Fourth Industrial Revolution.
And by making the technology available to more radiologists, their workflow will improve, allowing doctors more facetime with their patients.
Park Hee-jun, Arirang News.
Arirang News Facebook: [ Ссылка ]
------------------------------------------------------------
[Subscribe Arirang Official YouTube]
ARIRANG TV: [ Ссылка ]
ARIRANG RADIO: [ Ссылка ]
ARIRANG NEWS: [ Ссылка ]
ARIRANG K-POP: [ Ссылка ]
ARIRANG ISSUE: [ Ссылка ]
ARIRANG CULTURE: [ Ссылка ]
------------------------------------------------------------
[Visit Arirang TV Official Pages]
Facebook: [ Ссылка ]
Twitter: [ Ссылка ]
Instagram: [ Ссылка ]
Homepage: [ Ссылка ]
------------------------------------------------------------
[Arirang K-Pop]
YouTube: [ Ссылка ]
Facebook: [ Ссылка ]
Google+: [ Ссылка ]
