Earning Justice For The Injured 

  1. Home
  2.  » 
  3. Blog
  4.  » Artificial intelligence and melanoma misdiagnosis

Artificial intelligence and melanoma misdiagnosis

Melanoma is the most dangerous type of skin cancer. Over 6,000 Americans die of melanoma annually, which is why it’s important to be diagnosed properly. People in Ohio should be aware of the issues that can negatively affect diagnostic outcomes when it comes to cancer.

Artificial intelligence and dermatology

Convolutional neural networks, or CNNs, are a kind of artificial intelligence used to interpret visual images. Recently, they’ve been used in dermatology to classify images of growths on skin. Some of these are harmless moles. Others are dangerous melanomas. If CNNs could do this accurately, it would be great, but the problem is that researchers are finding that CNNs aren’t precise enough.

In a letter to the editor of the Journal of Investigative Dermatology, researchers noted some serious discrepancies. They compared the problems they were seeing with CNN identifications to misidentifying a bear as a primate. At the present time, CNNs are just not as good as the trained eye of a dermatologist. Even the color balance of the camera used to make the images can affect the accuracy of the reading.

In the future, it’s expected that convolutional neural networks will improve. But for now, they are not the gold standard for diagnosing skin cancers. They still need to develop further before becoming truly reliable. They can miss melanomas or misidentify harmless nevi as cancerous growths.

If you or a loved one have suffered because of a late melanoma diagnosis, you might want to contact a lawyer. A lawyer may be able to help you determine if you have a good medical malpractice claim. If you do, it may be possible to collect damages for pain and suffering.