This new computer program will detect skin cancer in humans better than doctors
People love basking in the sun in their backyards or on the beach in order to get that perfect skin tan. Some people also take use of artificial tanning beds in order to do so in the comfort of their homes, but what these people don’t consider is the risk of contracting melanoma from all the radiation they absorb by exposing their bare skin in sunlight or under tanning beds’ UV rays. Skin cancer or Melanoma is one of the fastest spreading cancers in US and one which is hard to detect. So researchers have come up with an algorithm to use a computer to do this work faster and more accurately.
1 What is skin cancer or Melanoma?
Melanoma is the most dangerous form of skin cancer. It happens when cancerous growths lead to skin cells to multiply rapidly and form tumors. These malignant tumors originate from the pigment creating melanocytes, which are present in the epidermis’ basal layer. In looks, melanomas resemble moles and some even develop from moles already present on the body. Majority of melanomas are black or brown, but they can even be blue, purple, skin colored or red. Main cause of melanoma is continuous and prolonged exposure to the UV rays from the sun. An estimated 10,000 plus people are killed every year in the US alone due to melanomas.
The major problem with treating melanomas are detecting them early enough to start an effective regimen of treatment. If the melanomas are detected late, they usually have spread to different parts of the body by then and it becomes too late to do anything. Researchers at the Stanford University have come up with a computer program that will help detecting melanomas more accurately in patients and saving their lives. Click next to learn more.
2 The new computer program
Researchers at Stanford University published their findings in the Nature Journal and it said that by` using deep learning and visual processing, scientists managed to use a computer in diagnosing possible cancer. It did so by referencing more than 130,000 images of skin cancer and it worked on an algorithm that the researchers developed on their own, which was inspired by the one written by Google, which detects 1.28 million images from more than 1,000 object categories. “There’s no huge dataset of skin cancer that we can just train our algorithms on, so we had to make our own,” co-lead author Brett Kuprel said.
3 Tests and performance of the algorithm
The algorithm was tested by the researchers against a 21 point guideline set by leading dermatologists in the country regarding each image. Several tests showed that the algorithm performed in sync with the Dermatologists’ guidelines. Currently, the algorithm is used on only one computer, but researchers are working on a mobile app, so that people can diagnose skin cancer on their own. According to the American Academy of Dermatology, about 8,500 Americans are diagnosed with skin cancer every day.