Our skin is the largest organ in the body. Skin cancer is the most common form of cancer, it globally account for at least 40% of cases. There are three major type of skin cancer basal-cell skin cancer (BCC), squamous-cell skin cancer (SCC) and melanoma. Around 4.3 million cases of BCC are diagnosed in the U.S. each year, resulting in more than 3,000 deaths.
Now a team of researchers have created an artificial intelligence system that can identify skin cancer more accurately then humans. The study is conducted by an international team of researchers in Germany, the USA and France. The study has published in the leading cancer journal Annals of Oncology.
The AI system is build around convolutional neural network (CNN). A CNN is an artificial neural network inspired by the biological processes that occur when the brain’s neurons are connected to each other and respond to what the eye sees.
Researchers found that the CNN was more reliable at identifying melanoma, an aggressive form of skin cancer, and benign moles (called nevi) than even experienced dermatologists.
“I expected only a performance on an even level with the physicians. The outperformance even of the average experienced and trained dermatologists was a major surprise,” first author Dr. Holger Haenssle, a senior physician at the department of dermatology, University of Heidelberg, Germany, told Healthline.
The team has taught the AI system to distinguish dangerous skin lesions from benign ones. The system was trained over more than 100,000 images of malignant melanomas and benign moles. Researcher compare it’s performance with dermatologists.
For this test fifty-eight dermatologists from 17 countries around the world participated. More than half of the doctors were considered expert level with more than five years’ experience. Nineteen percent said they had between two to five years’ experience, and 29 percent had less than two years’ experience.
The dermatologists were shown 100 images of skin lesions and asked to make a diagnosis, whether it was a malignant melanoma or benign mole. They were then asked to indicate their decision on how to manage the condition, such as surgery, short-term follow-up, or no action needed
Four weeks later, the researchers gave dermatologists clinical information about the patient, including age, sex, and the position of the lesion, and close-up images of the same cases.
The study authors then showed the CNN a set of 300 images of skin lesions. CNN is capable of machine learning, or teaching itself from what it has “seen,” so it can keep improving its performance.
At the end of the test, researcher found that the dermatologists correctly detected an average of 87 percent of melanomas, and accurately identified an average of 73 percent of lesions that were not malignant. Conversely, the CNN correctly detected 95 percent of melanomas.
“The CNN missed fewer melanomas, meaning it had a higher sensitivity than the dermatologists, and it misdiagnosed fewer benign moles as malignant melanoma, which means it had a higher specificity; this would result in less unnecessary surgery,” said Professor Haenssle.
“When dermatologists received more clinical information and images at level II, their diagnostic performance improved. However, the CNN, which was still working solely from the dermoscopic images with no additional clinical information, continued to out-perform the physicians’ diagnostic abilities.”
The expert dermatologists performed better in the initial round of diagnoses than the less-experienced doctors at identifying malignant melanomas. But their average of correct diagnoses was still worse than the AI system’s.
“These findings show that deep learning convolutional neural networks are capable of out-performing dermatologists, including extensively trained experts, in the task of detecting melanomas,” Haenssle said.
Dr. Suzanne M. Olbricht, FAAD, president of the American Academy of Dermatology, told Healthline in a statement that the group “welcomes technological advancements” but said that human doctors are not about to be replaced with a laptop.
“Although artificial intelligence may be a useful tool in skin cancer diagnosis, no machine can replicate the high-quality, comprehensive skin, hair, and nail care provided by a board-certified dermatologist,” Olbricht said in the statement.
“While this technology has great promise for the practice of dermatology, more work and research are necessary for that promise to be realized,” she said.
In an accompanying editorial, Dr. Victoria Mar of Monash University in Melbourne, Australia, and Professor H. Peter Soyer of the University of Queensland in Brisbane said “Currently, there is no substitute for a thorough clinical examination,” they conclude. Though there’s “much more work to be done to implement this exciting technology,” they say, “sooner than later, automated diagnosis will change the diagnostic paradigm in dermatology.”
You can read the study in detail form cancer journal Annals of Oncology.
More in AI :