A recent study published in Annals of Oncology, a cancer journal, chronicles how skin cancer diagnoses tend to be more accurate when they’re made via artificial intelligence devices rather than by dermatologists.
In order to reach this conclusion, researchers from the University of Heidelberg used a deep learning machine called the convolutional neural network (CNN) to observe moles. In the end, over 100,000 slides of these growths were analyzed alongside their diagnoses.
Each of the lesions were magnified 10 times their original size. Over time, the CNN became better at properly identifying both malignant and benign moles after researchers trained it using the slides.
During the second phase of testing, researchers had the CNN machine process 300 new samples alone. It then was used to test 100 images. Those diagnoses were then compared to the doctors’.
For the final testing phase, 58 dermatologists from 17 different countries participated. The researchers found that the physicians missed diagnosing skin cancer more frequently than the CNN machine did. They also determined that doctors often seem to misdiagnose lesions as melanoma and operate on patients when the moles are actually benign.
Researchers found that the CNN was effective in identifying 95 percent of melanomas whereas dermatologists were only accurate 86.6 percent of the time.
The doctors’ diagnoses improved by a little over 3 percent once they were told where the mole had been removed from, what the gender of the patient was and his or her age. Dermatologists’ diagnoses also improved once they were given better level II images. This added detail wasn’t needed for the CNN to render an accurate diagnosis.
While the researchers don’t see the CNN as replacing dermatologists or pathologists in the future, they do see them being used as an additional tool to confirm a suspected diagnosis.
Many dermatologists have been relying on the use of digital demoscopy devices to aid in the diagnosis of skin cancer in recent years. Until technology becomes more widely available to help doctors test for skin cancer, patients will have to rely on a Chicago medical malpractice attorneys to help them recover compensation for their misdiagnoses.