[이대승] What if my x-ray pictures are secretly leaking?

[출처: 코넬 대학교]

[이대승’s 블록체인 헬스케어] The x-ray pictures you see are not medical images of people that actually exist. This is a picture made with the Generative Adversarial Network (GAN) algorithm, which is an artificial intelligence algorithm. Of course, these videos are designed to provide a lot of learning materials so that real deep learning algorithms can better detect diseases. However, if these materials are being presented to the court, or photographs made to obscene certain celebrities, how do you verify them? If we claim that even if we submitted data from the hospital to refute it, we claim that we brought another photograph that was manipulated in a closed room agreement with the hospital.

#How to solve the data forgery problem

It’s not just photos. The same goes for videos. Whether it is a manipulation video or not guideIt will be announced until. The date or location of the video may be announced differently to make it misunderstood as a distorted context, or it may be subtly stitched to convey incorrect content that differs from the entire context. There is a lot of cutting back and forth, which can lead to misunderstandings by sharing only a small part, or there is a way to completely manipulate the video itself. It may be possible to manipulate the image taken for the analysis of gait that is walking normally and change it to look like a patient with severe brain disease is walking, or to manipulate it as if it were normal even with a serious disease. In a quarrel over a problem in the office, if a doctor or patient brings a different video and submits it to the court, it will be necessary to determine what is real.

As such, images and videos are of great value as evidence because they can read even the subject’s mind, unlike other forms of evidence, and they also exhibit greater power than hundreds of documents. Due to its great power, the temptation of forgery is not small, and because it is in the form of electronic records, it is extremely easy to forgery than other forms of evidence. What solution can blockchain really offer to this problem?

#’Can the’Image Chain’ innovation solve the problem?

Until now, there is a limit to the capacity that can be contained in each block of the blockchain, so it has been possible to contain only simple text or the content of essential smart contracts. There have been various efforts to link images or video data with the blockchain, but it was only possible to put the code created by hashing the entire video or image file on the blockchain. The blockchain had to be operated independently of the original file itself. There are also many attempts to capture images and videos using the IPFS (InterPlanetary File System) system, but since IPFS is still in its infancy, there are various limitations, such as having to split each image/video file into pieces of 256 kilobytes countless times. There are various approaches other than how to manage large files in connection with the blockchain, but all of them have obvious limitations to use in a realistic way.

2020Interesting on December 25th, PaperWill be announced. A Polish researcher suggests that this is called an Imagechain, thinking the opposite. It is an argument to put a distributed ledger in the image itself, rather than putting image information in a block of a small blockchain. Like so below.

(Image source: https://www.mdpi.com/1424-8220/21/1/82/htm)

Previously, blocks and blocks were bridged by a’hash’, but Image Chain is to connect each image with information that can be established as a block chain. In addition, blockchain information is inserted into each image like a hidden watermark so that it cannot be easily identified or altered.

The advantage of the image chain asserted in this paper is that each picture file can be used as it is, so the image can be used as an image without any conversion program, while the entire block chain data can be completely copied and transferred. Is that it has the effect of copying. For example, if you make a picture of a child’s growth into an image chain, copy the pictures at once and share them with family members, it has the effect that not only pictures but also blockchain data are distributed and stored among all family members.

(Image source: https://www.mdpi.com/1424-8220/21/1/82/htm)

In addition, if each image or video, like a block, contains the time or necessary information in the image chain at once and is configured as in the photo above, an image chain containing all the time sequence and context of each image will be built. It’s a fun idea that solves many of the misleading factors mentioned above. Through this, it is believed that problems such as random omission of the course of the disease according to the treatment process or inserting missing pictures can be solved. The number of unjust people who are harmed through distorted context or manipulated photos or videos will increase rapidly in the future.

As the value of bitcoin has recently skyrocketed and news of investment from institutions continues, digital bundles made with blockchain seem to be fully established as an asset. And beyond the frame of simply’trustworthy assets’, the blockchain seems to gradually melt into the’part that needs truth’ in society while changing its appearance in various ways. By capturing images and even videos beyond text data through the blockchain, we hope that it will be a great force to prevent unjust people from creating when truth is needed.

Dae-Seung Lee, Ophthalmologist, Hanyang University IAB Advisory Professor