In the world, melanoma is also the enemy of public health. In Australia, there are more than10.3 million new melanoma patients every year, and the number of deaths reaches10.6 million every year. It is estimated that 65,438,000 people will die of skin cancer in the United States in 20 16 years. At present, researchers are trying to find new ways to find this disease. According to new research, this new way is to take photos with the convenient and ubiquitous iPhone.
Noel Codella, a scientist at IBM Academy of Sciences and the author of this study, said: "iPhone photography will provide doctors with images of patients' skin and analyze whether it is melanoma. "In fact, it is through a smart phone: when someone finds something suspicious on their skin, they can use the camera of the mobile phone to take photos of the diseased part and submit the images for evaluation through the analysis service. These images can identify and reliably identify the characteristics of the disease. But in fact, it is much more complicated than this.
In 20 13, IBM announced that it had reached a cooperation with new york Si Long Kettering Company to apply image recognition technology and machine learning system to skin cancer screening. Codella's team analyzed more than 3,000 cases of melanoma and other skin lesions. The accuracy of the system to distinguish negative and positive skin cancers is over 95%, and the time to analyze each picture is less than one second. The key to the success of this project depends on two factors. The first is the widespread use of dermatoscope, which is a photo analysis device that can be connected to the camera of smart phone, which can optimize the lesions. The second (more important) factor is the development of a huge database containing cancer spot images. The database adopts IBM's machine learning, computer vision and cloud computing capabilities and development means, and is accessed under the condition of continuous identification of melanoma through technology.
In 20 15, the team of IBM released the preliminary research report of this diagnosis method. In the report, IBM described the advantages of computer vision diagnosis. With the help of MSK and the international skin imaging cooperation agency, IBM has obtained the initial data set. In the research, IBM has made great achievements, but at present, this algorithm is still not smart enough, and medical experts still need to draw the outline of skin lesions in pictures. As for the accuracy of the diagnosis, IBM said that there was no direct comparison with human medical experts. After the publication of the research report, IBM has been promoting the progress of computer vision diagnosis technology. Its goal is to make the computer automatically identify the skin lesion area and improve the analysis efficiency. At the same time, they also invited eight experts to compete for the diagnosis effect. The results show that the performance of IBM's computer vision diagnosis method is three times better than the previous diagnosis method, and the accuracy rate has reached the level of eight industry experts. The online version of this paper has been provided and will be published in 2017 IBM R&D Journal.
However, it should be noted that this diagnostic method is still in the research stage. Codella said that it may take some time to optimize this technology before it is widely used by dermatologists. Of course, this means that IBM's research experiments may also fail. However, if this tool is finally successfully used by doctors, it will definitely make a great contribution to defeating skin cancer and will also be the gospel for skin cancer patients. (Science fiction Galaxy Constantine/Wen)
If you need to reprint this article, please contact WeChat: khxx-wk.
Science fiction Galaxy official WeChat official account: kehuanxx