A new development in artificial intelligence showed that such technology can be used to identify cancerous skin spots in Australia and South Korea, said ABC and Korea Biomedical Review.
A trial site for skin cancer detection was established in north Queensland, Australia which has the most dangerous ultraviolet ratings in the globe. The trial will study the potential of AI in detecting skin cancer to save lives.
Doctor Jeremy Hudson, clinical director at the North Queensland Skin Centre, said, “We have some of the highest rates of [skin cancer] in the world.” He added, “There are more deaths from skin cancers than car accidents up here.”
The facility has only begun integrating AI in its routine skin checks. It uses a computer-sized machine made in Germany to take high-resolution photos of the patient’s skin, which are then transmitted to a database.
These images are run through an algorithm that analyzes the photos to look for any skin spots that may be cancerous.
The machine taps into global databases of photos depicting skin cancer, which is used to improve the evaluation of the condition. The images sent to the database also contribute to the dataset.
Dr. Hudson remarked, “This will go an incredibly long way to our goal, which is which we want no one to die from melanoma.”
Regarding technological progress in the field of medicine, he said, “We’re at the point where skin melanoma technology is advancing as rapidly as it ever has in the course of medicine.”
North Queensland Skin Centre’s is touted as having similar skill levels compared to international specialists.
Meanwhile, South Korea’s Asan Medical Center (AMC) also uses an AI model that is at par with the level of dermatologists. According to the Korea Biomedical Review report, the tech is able to detect around 43 types of skin cancer and other diseases.
The facility reported that the model exhibited 66.9% sensitivity, 87.4% specificity, which is comparable to the 65.8% and 85.7% showed by dermatologists.
This technology aims to help curb the misdiagnosis of skin tumors, resulting in a delayed diagnosis. This allows them to get treatment earlier. This misdiagnosis was from the use of existing studies using AI.
However, this model uses images depicting all kinds of malignant and benign skin cancer data from a study by Severance Hospital. This study showed 70.2% sensitivity and 95.6% specificity.