Scientists created a Tetris-like AI program for breast cancer diagnosis | Robotics
More than 1.5 million women each year are affected by breast cancer – it is one of the leading causes of cancer deaths for women with more than half a million women dying every year from this disease. It can be treated successfully, but only if diagnosed early. Now scientists from the University of Adelaide are developing a fully-automated medical image analysis program to detect breast tumours and it is kind of like Tetris.
You may remember Tetris – this electronic retro game has little cubes moving around that you had to navigate in certain ways to win the game. The new image analysis program for breast cancer diagnosis is much more complex –it uses artificial intelligence and employs the traversal movement. How is it similar to Tetris? Well, it pretty much uses a cube, which is moving through the breast image. It is typically green, but once it finds a lesion it turns red. The advantage is speed – this method is approximately 1.78 times faster in finding a lesion than existing techniques.
Lesions in breast tissue indicate tumours, which may be quite small. They have to be found relatively quickly, in order to keep the disease in check. The good news is that while being much faster, this method is just as accurate as currently used ones. Scientists used deep reinforcement learning methods in an artificial intelligence system in order to create this program. It allows the algorithm to learn how to do complex tasks without being programmed by humans. And now the program can analyse breast tissue, which is not a simple task at all. That allowed scientists to create this program using a relatively small pool of data, which sped up the process. But of course, the program is still in development.
Researchers are satisfied with AI performance in this task. It intuitively locates lesions quickly and accurately without an extensive pre-programming. Associate Professor Gustavo Carneiro, one of the creators of the program, said: “More research is needed before the program could be used clinically. Our ultimate aim is for this detection method to be used by radiologists to complement, support and assist their important work in making a precise and quick prognosis. Artificial intelligence has an important role to play in the imaging medical field, the potential to use AI in this field is boundless”.
Artificial intelligence is a new frontier for medical diagnostics. Learning algorithms could diagnose certain conditions quicker and more accurately. They could also adapt to the changing conditions and log data much easier, which would allow following the progression of the disease much easier.
Source: University of Adelaide