U.S. Patent No. 10,803,586 details the use of an active learning tool, Annotation Assistant, to increase the speed and accuracy of training AI models for image analysis.
Aiforia, a leading provider of deep learning artificial intelligence (AI) solutions for image analysis across a variety of medical domains, has been granted U.S. Patent No. 10,803,586 by the United States Patent and Trademark Office (USPTO). The patent covers the techniques deployed by Aiforia’s active learning tool, Annotation Assistant, for the accelerated development of AI models for image analysis in pathology.
AI-assisted image analysis is becoming widely used across healthcare, from radiology to pathology across a wide range of areas such as oncology and neuroscience. Artificial intelligence models are proven to speed up and standardize image analysis while providing increasingly accurate quantitative data, enhancing the ability of healthcare professionals in research and clinical labs to make discoveries, decisions and diagnoses.
Aiforia’s cloud-based software, Aiforia Create, enables medical professionals to create AI models, without the need to code or use dedicated hardware, for any image analysis application. This is now made even easier thanks to the newly patented technique employed by Annotation Assistant which increases the efficiency and accuracy of training AI models with Aiforia Create.
“AI is transforming medical image analysis, but the misconception is that it is difficult and time-consuming to get started with it. Aiforia Create has removed these barriers and with the Annotation Assistant tool we have made it even easier and faster for pathologists, scientists, and researchers to train AI models for various types of image analysis applications,” explains Aiforia CTO Tuomas Ropponen.
Annotation Assistant utilizes active learning, a highly sought-after technique in artificial intelligence. The tool works by scanning images for areas that would provide the most useful training data during AI model development, suggesting these to the user. This auto scanning technique therefore decreases training time, improves the quality of the AI model and is particularly useful in analyzing rare cases and large data sets with heavy variation. The same tool also provides an ideal method to rapidly adapt pre-validated AI models to a new laboratory.
“What makes Aiforia unique is the investment we make in our offering. We not only invest in developing software that is very user friendly, simplifying the use of AI in medical image analysis, but also in providing technology that is cutting edge. This patent truly signifies our commitment to bringing powerful, complex AI techniques straight to the hands of medical professionals,” explains Kaisa Helminen, COO of Aiforia.