00:0095% of cases of lung cancer in Malaysia are affected in a relatively late stage.
00:14I am Shamsuddin from the cancer AI team.
00:18We are actually a project that has been set up specifically to help solve the problems of new cancers
00:25with a more early stage.
00:28With the AI system that we have built with more than 900,000 x-ray images with labels,
00:36we have trained it.
00:38And now, if someone gives us an empty x-ray image,
00:43we put it in the system and the cancer AI can do detection.
00:48The results of the study can be detected whether there is a problem or not.
00:53So it can be used as an early sample.
00:56And with this process, we can see that the high-risk patients
01:01can be given priority for further detection.
01:06Usually, after the x-ray, we will get the x-ray film or image.
01:12And for our system, we only need the soft copy of the x-ray in PNG format, for example,
01:20which we can interpret or upload to the cancer AI system.
01:25And for now, we make this system using what we host in the cloud.
01:30So it can only be uploaded to our cancer AI system.
01:35And we click submit.
01:37And behind the engine, it will run.
01:39And in three seconds, the cancer AI can do detection.
01:42And it will be shown on the right side of the screen later
01:46what are the risks if there are.
01:49And if we want to see where the high-risk locations are,
01:53we can click on the heat map and it will show in the middle of the x-ray image
01:58where the risk areas are
02:02for us to give to radiologists for more specific attention.
02:09With the existence of this AI process,
02:11we can speed up the detection process more objectively.
02:15In those three seconds, we can determine where the cancer is located.
02:20Actually, we want to speed up the initial process.
02:23When we upload the image, we can detect that there is a potential cancer.
02:28Then we can proceed to the radiologists.
02:31So from here, we can reduce the burden of radiologists
02:36and speed up the process.
02:45Microsoft Mechanics
02:46www.microsoft.com.com
02:47www.microsoft.com.com
Comments