Technology can help save lives and a team of Stanford researchers have taken this to the next level. The team have successfully developed image recognition technologies and programmed a computer to analyse and potentially identify cancerous moles and skin lesions.
Experts actively promote the preventive measures and check-ups necessary to ensure that melanomas are detected as early as possible. The five-year melanoma survival rates are 99% when detected early. The figure dramatically decreases to 14% if the cancer is detected at a later stage.
The project was led by Sebastian Thrun, founder of research and development lab Google X and adjunt professor at Stanford University. He explained to CNN that their goal is to make melanoma detection more accessible on a global level. “Our objective is to bring the expertise of top-level dermatologists to places where the dermatologist is not available.”
The Stanford team used an algorithm-based technique to coach a computer to pick up on pattern recognition. This technology is known as “deep learning” and essentially gives the computer the capability of applying basic rules when analysing digital images of moles and lesions and determining the risk of cancer.
The process to “teach” the computer was complex and began by teaching the device to recognise day-to-day objects like animals, chairs, tables, etc. The next step was to slowly program the algorithm to enable detection of different types of skin conditions and eventually healthy skin lesions and moles vs potentially cancerous ones.
Naturally, no two skin aberrations are the same and therefore it was necessary for the team to feed over 129,000 different images representing upwards of 2,000 skin diseases in order to make the devise more artificially intelligent and have a higher detection score.
The accuracy that Thrun’s team was able to obtain was truly outstanding. Dermatologists regularly use a dermatoscope to magnify the area of the skin and help with the diagnosis process. However as they had fed the computer such an enormous variety of images of skin, its performance equalled the accuracy of 21 dermatologists analysis – without the need of the tool.
Although the ultimate goal is not to outweigh the important work carried out by medical professionals, but to help raise awareness on a global scale about skin cancer prevention and provide diagnostic assistance. The natural evolution of this image recognition technology it to apply it to mobile devices in order to increase use and impact.
According to the mobile giant Ericsson, there will be 6.3 billion smartphone subscriptions by the year 2021. New technologies like the skin scanner screening algorithm combined with the wonders of the mobile world and apps will actively help promote universal access to diagnostic care.
If you have any dermatological questions, we invite you to contact the Best Doctors team at askbestdoctors.com.