Cancerous skin lesions are not always easily identified. Currently, visual examinations play a significant role in the diagnosis of skin cancer. However, recent research from the University of Birmingham confirms that visual inspection, even by knowledgeable practitioners, isn’t always a viable way to diagnose skin cancer.

The researchers warn that smartphone apps have also thus far turned out to be disappointingly unreliable in detecting cancerous melanomas. Their conclusion: Methods such as dermoscopy, teledermatology and artificial intelligence techniques are the most reliable ways to determine whether a particular skin lesion is sufficiently suspicious as to warrant a referral to a specialist for further examination.

Artificial intelligence is of particular interest to dermatologists and other healthcare professionals right now. Recent research published in the Annals of Oncology revealed that a convolutional neural network (CNN) could be trained to recognize skin cancer with greater accuracy than human dermatologists are able to.

Researchers at Stanford University created a new dataset featuring skin cancer images and related data. Some translation was necessary, as the available data in its original form was presented in multiple languages. The researchers fed their newly compiled dataset into an algorithm of Google’s that was already able to identify large numbers of images from various categories of different objects.

Then the researchers tested the capabilities of their algorithm as compared against the skills of 21 dermatologists. They were particularly interested in whether the algorithm was able to correctly identify benign or malignant lesions, and what degree of accuracy it was able to achieve in doing so. Their conclusions were positive. The CNN did not miss as many instances of melanoma as the human dermatologists did, and it was more accurate in its assessment of benign moles.

Moles sometimes become melanomas, but they don’t always. There’s a protein that, when present, typically results in a benign lesion rather than a cancerous one.

When a bothersome skin lesion develops, whether or not cancer is suspected, a human practitioner is still actively involved in deciding how to handle the situation.

It can be beneficial to remove even benign moles, particularly those that are unsightly or placed in inconvenient locations on the body. When mole removal is deemed to be the best course of action, the practitioners at Cosmos Clinic recommend using an erbium laser to remove benign moles after applying a local anesthetic.

For cancerous moles where melanoma is present, surgery is usually required. In some cases, a specialist may also recommend additional treatment such as radiotherapy, chemotherapy or drugs.

Skin cancer can be more effectively treated if it is diagnosed as early as possible -- and artificial intelligence techniques are poised to help us improve our efficiencies in recognizing and diagnosing the various skin cancers our patients suffer from. It appears that putting this technology in the hands of capable practitioners could, indeed, take us one giant step closer to beating skin cancer in the future.