JERUSALEM, Feb. 13 (Xinhua) -- Israeli researchers have developed a new method to determine which cancerous cells are most likely to become metastases, as reported on Wednesday by the southern Ben-Gurion University (BGU).

The researchers used a deep learning method to extract quantitative data about the appearance and behavior of melanoma skin cancer cells from video clips taken with an optical microscope.

The method studies characteristics that cannot be identified by the human eye and led to several new diagnoses.

For example, the method makes it possible to distinguish the cells of different patients based on the videos.

In addition, it was found that common cell cultures for the study of melanoma are similar in behavior and very different from tumor cells taken from patients.

Therefore, the researchers say that the use of a cell culture as a functional model for melanoma may be clinically irrelevant.

The researchers were also able to predict the worsening of stage 3 melanoma, in which metastases are in the lymphatic system, to stage 4 where they spread in the patient's body.

The researchers hope that the method will be used to diagnose cancer cells directly from tumors in melanoma patients - to enable diagnostics, treatment adaptation and drug development.

The new program has potential for breakthrough uses in personalized medicine and expanding the method to other diseases.