According to a story from New Scientist, a team of scientists at the Institute of Cancer Research in London were able to develop an AI tool that was able to accurately identify differences between the shapes of different tumor cells. In a test of the AI tool, it was able to predict how aggressive an ovarian cancer tumor would be based on the shape of the cells within it.
About Ovarian Cancer
Ovarian cancer can appear on or within the ovary. Ovarian cancer rarely causes distinctive symptoms in its early stages, so many patients are often diagnosed with advanced disease. The risk of getting ovarian cancer is connected to how long a woman has ovulated during her life; women who ovulate for longer periods are at greater risk. Late menopause or early puberty are risk factors, as are not having children, fertility medication, certain genetic variants and mutations (such as BRCA mutations), and exposure to talc, herbicides, and pesticides. Some symptoms of ovarian cancer include fatigue, bloating, a feeling of fullness, loss of appetite, indigestion, abdominal swelling, and pelvic pain. Treatment can include chemo, radiation, surgery, hormone therapy, and immunotherapy. There are many different kinds of ovarian cancer. Five year survival rate is 45 percent in the US. To learn more about ovarian cancer, click here.
Predicting Tumor Aggressiveness
The AI tool looked at samples of ovarian cancer tumor cells from 514 patients with the disease and determined that tumor cells that had misshapen nuclei were more likely to be aggressive. In fact, the form of the disease with misshapen nuclei had a dramatically worse five year survival rate compared to ovarian cancer overall, at just 15 percent. Normally, ovarian cancer has a five year survival of around 45 to 50 percent.
While it is possible for human scientists to observe cells and make such determinations, it is long and extensive process, which is why the use of a faster moving AI tool makes sense.
Still, at this juncture, the AI tool is of limited utility because it is unclear what treatment approaches are best for more aggressive forms of ovarian cancer. Telling a patient that their disease is more aggressive isn’t that helpful if their aren’t corresponding approaches to treatment. However, AI tools such as this could be valuable in the development of more targeted treatments in the future.
