CT detection of new coronary pneumonia with AI

Recently, China ’s leading host cloud sharing platform Mochichi Cloud used neural network models to train 20 normal case CTs and pneumonia cases CT respectively, and successfully completed CT image detection of pneumonia . Although this case is only a case of CT identification of common pneumonia, it also has a certain reference for new coronary pneumonia. At present, the "New Crown Pneumonia Prevention and Control-CT Detection of Pneumonia" case mirroring has been officially launched on the Juchi Cloud platform. Developers can use the official website of Juchi Cloud to "machine lease"-"I want to lease"-"choose mirror"-"Jupyter tutorial Demo ".

WeChat picture_20200214162645

It is understood that Juchi Cloud was launched in October 2019 and is a leasing platform specializing in GPU servers, providing domestic users with high-performance, low-cost, and cost-effective GPU leasing services. Facing the pneumonia epidemic of new-type coronavirus infection, Juchiyun actively assumed social responsibility and refueled Wuhan in its own unique way.

Babbitt quickly contacted the product operation staff of Juchi Cloud, focusing on what the thinking behind the “New Crown Pneumonia Prevention and Control-CT Detection of Pneumonia” case on the launch of Juchi Cloud, what can be used for, what effect is expected to achieve, and how artificial intelligence Interviews with issues such as integration with blockchain.

Babbitt: What's the starting point of the case of "Ji Chuyun Pneumonia Prevention and Control-CT Detection of Pneumonia"?

Moment cloud:

The current epidemic situation of new coronavirus affects everyone's heart. I have seen the use of AI model tests to predict the X-rays of patients with pneumonia, so we also want to take this opportunity to use a simple model to let AI lovers and friends who like AI understand AI in the medical field Application. For this purpose, we made this demo. In simple terms, it is through the neural network model to predict whether X-rays of patients with suspected pneumonia are pneumonia. The ability to quickly complete this demo is mainly due to the stability and ease of use of Mochi Cloud Platform.

Babbitt: What brainstorms have you experienced?

Moment cloud:

At the moment of the epidemic, we are out of social responsibility and a sense of mission. It took 4 days to complete the development of the idea. It takes about 20 minutes to train 5 epochs on the 2070 card of the Mochi cloud platform.

Babbitt: Is it possible for clinical use after development?

Moment cloud:

This demo proves that AI image technology can be used in clinical medicine to screen chest radiographs of patients with new coronary pneumonia. Since it is only a demo, it cannot be used in clinical yet, mainly for AI enthusiasts to learn how to use it.

Babbitt: Why is Mochiyun focusing on AI?

Moment cloud:

In the future, artificial intelligence may become the main and main direction of technological development; with more and more practitioners and learners, the demand and use of machines in the industry will also increase; compared to users who build machines themselves High cost and high threshold, GPU leasing can greatly reduce the use and entry barriers for users; we hope that through the platform of Moments Cloud, it can help some practitioners and enthusiasts; so that they can be faster and lower Learn and test.

Babbitt: Who are the users of Juchi Cloud?

Moment cloud:

The user population is biased towards machine deep learning users, practitioners, medical practitioners, scientific research practitioners, graphic algorithm practitioners, and some AI enthusiasts. At present, our users are relatively young, most of them are mainly student groups.

Babbitt: Facing big platforms like Google, Alibaba Cloud, Tencent Cloud, and Baidu Cloud, what are the advantages of Moments Cloud?

Moment cloud:

Compared with big factories and those big platforms, in addition to the price, our platform has the advantages of open and transparent machine information, flexible and diverse billing, and simplified rental process.

Babbitt: What are the latest developments in Moments?

Moment cloud:

We have been online for only 5 months, and it is still a relatively basic and junior platform. In order to better serve the majority of users and meet user needs, we are constantly working hard to improve. Although, we currently support some mainstream application scenarios, we will explore more use environments to meet the needs of different users.

Babbitt: AI and blockchain are currently the most promising technologies. How will the two be combined?

Moment cloud:

Blockchain is to ensure accurate recording, authentication, and execution, while AI is to make decisions, evaluate, and understand certain models and data sets, and ultimately generate autonomous interaction and learning. Both artificial intelligence and blockchain have the following characteristics:

1. Both AI and blockchain need data sharing. A distributed database emphasizes the importance of sharing data between multiple clients on a specific network, while artificial intelligence relies on big data and an external open data sharing platform. The more data provided, the more prediction and evaluation of the machine will The more accurate, the more reliable the generated algorithm is.

2. Both AI and blockchain require a highly secure network. When conducting high-value transactions on the blockchain network, we have great requirements for the security of the network. Compared to AI, autonomous learning of machines also requires high security to reduce the possibility of catastrophic events.

3. Both AI and blockchain require trust in the platform and data. For the progress of any widely accepted technology, there is no greater threat than lack of trust.