Artificial dermatooncologist

A new AI at the level of an experienced doctor determines the severity of melanoma

Svetlana Maslova, Hi-tech+

To choose a treatment strategy, it is very important to determine the degree of cancer damage in order to accurately plan surgical intervention. Quite often, an atypical mole on the skin can quickly grow into metastases far beyond the tumor focus. Swedish scientists have developed an algorithm that evaluates these probable risks at the level of an experienced dermatologist.

Article by Gillstedt et al. Discrimination between Invasive and In situ Melanomas Using a Convolutional Neural Network published in the Journal of the American Academy of Dermatology – VM.

Assessment of the degree of cancer damage is very important for doctors who need to know exactly before surgery whether the melanoma is invasive with developing metastases or the tumor has just begun to grow and is still limited to the epidermis.

Usually, these risks are assessed by a dermatologist using a dermatoscope, but it is still quite difficult to determine the stage of melanoma before surgery.

Artificial intelligence developed by scientists from the University of Gothenburg can change the diagnostic process preceding surgery.

The new algorithm was trained on 937 images of melanoma, and then tested on 200 cases of cancer, which were previously evaluated by seven experienced doctors.

AI at the level of doctors assessed the severity of lesions, while none of the specialists could significantly surpass the algorithm in the effectiveness of diagnosis.

Despite the results obtained, so far the authors plan to use AI only as an auxiliary tool for doctors. Currently, the team is finalizing the tool in order to use it in clinical practice in the future.

Portal "Eternal youth" http://vechnayamolodost.ru