Google Research Transforms Medical Imaging

 In short: 

A team of researchers at Google has developed a new model of artificial intelligence, which they claim has the potential to have a significant impact on medical research and clinical applications. A research group led by Chikové Azizi, a resident at Google's Research Center on Artificial Intelligence, has published a paper titled "Self-Supervised Large Models, Classifying Advanced Medical Images," which outlines a new approach that uses self-supervised learning to train DL models for medical imaging. . The researchers claim that the model could help create a self-supervised deep neural network that can improve the efficiency of clinical diagnosis in a variety of fields, such as ophthalmology, dermatology, mammography, and pathology. Azizi and his team, called Multistate Contrasted Learning (MICLe), have created the new model, which is trained on unlabeled data, thus enabling the application of AI in areas where collecting clearly defined data sets might have previously been difficult - such as as in cancer research.

Why it matters: 

The main conflict that this research attempted to resolve was to make deep neural networks more powerful and efficient in critical medical applications. Currently, in many medical research, practitioners do not always have abundant, clearly articulated data sets. This has made it difficult for medical AI researchers to create effective training models for deep neural networks to identify medical data with a high degree of accuracy.