Can Computers Diagnose Melanoma?

With so many advances in technology and computer learning, is it possible that one day computers could replace doctors? Robots already assist in surgeries and 3D bio-printers can create synthetic body parts. But can computers reliably make a medical diagnosis?


Medical researchers in California think so – in a collaboration between Stanford’s Department of Electrical Engineering, the Department of Dermatology and the Department of Pathology, among others, scientists have developed a computer algorithm that can diagnose melanoma from a typical photo of a mole taken by any smartphone.



Image: Source


The researchers programmed a computer learning algorithm called a “convolutional neural network” or “CNN,” by using 129,450 clinical images showing 2,032 different diseases to “teach” the CNN what a specific carcinoma looks like. The authors then put the CNN to the test against 21 board-certified dermatologists in a challenge to accurately diagnose the most common and most deadly skin cancers. The authors of the study report that their method performs with a similar success rate as the board-certified dermatologists when it comes to distinguishing malignant melanoma and keratinocyte carcinoma from benign lesions.


Current apps in the U.S. provide information and education about skin cancer and allow users to save pictures of any skin abnormalities, but do not suggest a diagnosis. However, in countries like Australia, Canada and the U.K, you can already download an app onto your phone that can check your mole photo for common melanoma markers and give you a probability that you might be looking at a malignant cancer.


This new publication might provide the evidence needed to push for a diagnostic app to be approved for use in the U.S. These services could make a huge difference to patients in rural areas where the closest dermatologist might be hundreds of miles away. While all current services are careful to warn consumers that it is not a substitute for seeing a real, live dermatologist, apps might provide the information needed to encourage patients with concerns to go ahead and make an appointment to have their moles checked at the doctor.



Image: Source


While you’re waiting for policymakers to catch up with technology, check out these bizarre skin conditions that you never want to get.


Featured Image: Source

5352 Total Views 11 Views Today

Laurie Breen

Laurie Breen is a freelance writer well-versed in research communications and grant writing. She received her Bachelors Degree in Psychology from Smith College and has worked previously at the University of Queensland's Centre for Clinical Research in Brisbane, Australia. Her favorite conversational topic is "antibiotic-resistant bacteria," making her a big hit at parties.