"VA Artificial Intelligence Can Help Diagnose Cancer"

2019-12-09

There have been previous examples of using artificial intelligence (AI) to detect cancer. Recently, pathologists conducted a research study on Apple Create ML and Google AutoML, with the aim of determining whether these programs can differentiate between three different types of cancer. AI can almost perform all tasks that computers can do, or in other words, it is currently a more powerful computer. Researchers wrote in their experimental records, "We attempted to teach computers to differentiate between cancer images and benign tissue images, and conducted research on the two most common types of cancer - colon cancer and lung cancer." The team used a microscope with an attached camera to capture 1250 different tissue samples, including benign tissue and three types of cancerous tissue - colon adenocarcinoma, lung squamous cell carcinoma, and lung adenocarcinoma. They then inputted the photos into the AI program in the form of slides and obtained satisfactory results. They conducted a total of six experiments, all of which achieved an accuracy rate of over 90%. They also tested whether the system could identify specific mutations in colon cancer slides, with a success rate of 70%. This means that with just two computers and a regular physician, the work that pathologists typically struggle with in the laboratory can be completed. Differentiating between different types of cancer is crucial because each type requires a different treatment plan. The AI tests conducted by the research team are still ongoing. On the other hand, they are currently training the system with images, including those of intracranial hemorrhage, to improve its sensitivity and emergency processing capabilities, further studying its practicality. "Now, we will run a model to train the AI and see if it can diagnose intracranial hemorrhage. The significance of this project lies in the assumption that a patient arrives at the emergency room without a radiologist. With this system, the patient's risk can be greatly reduced. The accuracy of AI diagnosis is very good, but its purpose is not to ultimately replace doctors, but to serve as a supplement and emergency aid, as life is priceless." - Experimental diary. The ultimate goal of this team is to create a project that can be widely implemented throughout the VA system. This will greatly benefit pathologists who work independently - their workload will be reduced and their efficiency will be greatly improved.

RECENT AI NEWS

RECENT AI TOOLS