 |
| Artificial intelligence accurately detects skin cancer |
Research into evidence supporting the use of AI solutions to diagnose skin cancer has spurred the development of algorithms at the University of Gothenburg to determine the severity of skin cancer as precisely as a dermatologist.
Scientists at the University of Gothenburg have shown that the technology can reach the same level as dermatologists in assessing the severity of skin cancer.
The system is designed to help doctors determine the stage of skin cancer. The results have been published in the American Journal of Dermatology.
Although patients usually discover melanoma independently by discovering new moles or changing existing ones, it is difficult even for a dermatologist to determine whether it is harmful.
Researchers suspect artificial intelligence could help them with this task. They used convolutional neural networks to classify melanomas. Convolutional neural network is an effective image analysis method that has been shown to identify various skin lesions.
The research was conducted at Saar Glenska University Hospital in Gothenburg. Researchers trained and validated the system using 937 melanoma images captured with a handheld dermatoscope scanning tool for melanoma skin.
They tested the results of the algorithm evaluation in 200 cases diagnosed by dermatologists and compared the performance of the system with the test results of 7 independent dermatologists to arrive at a conclusion.
"The results of this study are very exciting," said the study author, a researcher at the University of Gothenburg and a medical expert at the University Hospital of Sar Groenska, Sam Polisi.
He added, "I hope that this algorithm will be used to help make clinical decisions in the future as no dermatologist can significantly outpace this algorithm."
The researchers acknowledge that the algorithm needs to be improved and evaluated in the clinical environment and in future long-term research.
However, research shows that AI can be useful in assessing the severity of melanoma before surgery, which may affect the completeness of the procedure.