Varian and Google to co-develop AI-based cancer diagnostic platform
Often viewed as a crucial and laborious step, organ segmentation in radiation oncology can constrict treatment workflow.
The Varian-Google partnership intends to address this problem by using neural architecture search (NAS) technology via the Google Cloud AI Platform to determine which organs and tissues in diagnostic images must be targeted or protected during a course of radiation therapy.
Additionally, Varian will leverage Google’s NAS technology to create an AI segmentation engine that generates customised auto-segmentation models for the body’s organs. Medical professionals can use the segmented images to design treatment plans that detail where and how much radiation they’ll deposit during treatment.
The AI models for organ segmentation will be trained using Varian’s proprietary treatment planning image data. According to reports, Varian intends to incorporate these models in its treatment planning software tools for use by cancer centres worldwide.
“At Varian, we are working towards a world without fear of cancer, where high-quality cancer care – personalised and optimised for each patient – is available everywhere,” said Corey Zankowski, senior vice president of technology and innovation office at Varian.
Zankowski continued: “To that end, we have committed ourselves to Intelligent Cancer Care, which seeks to automate routine or repetitive tasks in the radiation oncology workflow through the use of smart algorithms, machine learning, and AI. This collaboration with Google Cloud will turbocharge our efforts in this area.”
A fully managed and end-to-end platform for data science, Google’s AI platform also features a built-in data-labeling service that’s ideal for classification, object detection, entity extraction, and more.
“At Google Cloud, we believe AI technology has the power to impact a wide variety of industries, which is especially true in the field of medicine,” comments Craig Wiley, director of product management at Google Cloud.
© Dennis Publishing