Researchers at Stanford University have just created a machine that can detect skin cancer almost like an expert doctor. With their invention, the Stanford researchers have just made it easier for doctors to determine skin cancer on the spot without further testing. Early detection would also mean a lot for skin cancer patients in preventing the spread of the deadly skin disease.
Early detection of skin cancer using AI
The machine-learning AI was created by a team of researchers from the Stanford Artificial Intelligence Laboratory, CNET learned. Using an AI algorithm, the machine can identify skin cancer by "looking" at images.
The AI machine uses the deep learning technique wherein it was trained to use almost 130,000 images of skin lesions, rashes and even moles to identify the possibility of skin cancer. When tested against human dermatologists, the machine proved to provide an accuracy that is 91 percent as good as an expert doctor's diagnosis.
Early skin cancer detection: better survival rates
According to The Verge, there are roughly 5.4 million cases of skin cancer diagnosed in the United States each year. The process of identifying the deadly skin disease usually involves a series of visual examination by a dermatologist.
Early detection of skin cancer means better chances of survival for the patient. For example, early detection of melanoma makes for a 97 percent chance of survival. But if a patient gets diagnosed too late, there's only around 14 percent of survival rate or even less.
By teaching their AI machine to detect skin cancer, patients can now be easily diagnosed if their conditions are malignant or benign. One member of the research noted the lack of dataset for skin cancer, which pushed them to make their own by gathering images from the Internet. Later on, they worked a nice taxonomy out of 129,450 images with the help of the medical school. All the images portray more than 2,000 of skin diseases, including numerous types of skin cancer.
Not to replace dermatologists
The team added that while the machine is almost as accurate as dermatologists, its aim is not to replace them. Instead, the AI machine can provide people with early screening that is inexpensive yet still effective. The Stanford team also suggested that a mobile app can be created to spot skin cancer even right at home.