Google Cloud launched a new product suite on Tuesday that promises to make medical imaging data more interoperable and useful by leveraging artificial intelligence.

Imaging data makes up the vast majority of all healthcare data — about 90%. Physicians use medical imaging data every day to diagnose and treat their patients, and billions of studies that rely on this data have been conducted globally, Alissa Hsu Lynch, who leads Google’s medtech strategy and solutions, said in an interview.

As medical imaging technology has advanced, the size and complexity of these images has also continued to grow. That growth has increased the workload for researchers and radiologists who are responsible for having to interpret these images in order to provide patient care. Google Cloud’s new imaging suite addresses this problem, according to Hsu Lynch.

The new suite addresses the common pain point that healthcare organizations face when developing image-based AI models — the labeling and annotation of images. To prepare images for model training, AI developers need to label the areas of concern on every image used in the data set, which is an incredibly manual and repetitive process. To combat this, Google Cloud partnered with artificial intelligence companies Nvidia and Monai to offer annotation tools “that just make the process much faster for organizations,” Hsu Lynch said. 

“The size and complexity of these images is huge, and often, images stay sitting in data silos across an organization,” she said. “In order to make imaging data useful for AI, we have to address interoperability and standardization. This suite is designed to help healthcare organizations accelerate the development of AI so that they can enable faster, more accurate diagnosis and ease the burden for radiologists.”

Hackensack Meridian Health, a New Jersey-based health system, has begun using Google Cloud’s suite for image-based research to help with the earlier detection of metastasis in prostate cancer patients. They are currently de-identifying petabytes of medical images, preparing them to be useful for AI, Hsu Lynch said.

The health system is building its AI capabilities so that it can eventually give image-based clinical diagnoses across a range of imaging, Chief Data and Analytics Officer Sameer Sethi said in a statement. 

Global medical device company Hologic is another early adopter of the imaging suite. The company is using Google Cloud’s offering to strengthen its diagnostic platform that screens women for cervical cancer. Hologic will store its images using the suite, and it will develop an AI model with Google Cloud to improve diagnostic accuracy for those cancer images, according to Hsu Lynch. 

In addition to image storage and assisted annotation, Google Cloud’s new offering also includes imaging datasets and dashboards, tools to build AI data pipelines, and tools to help meet data security and privacy requirements.

But Google isn’t the only tech company to jump into medical imaging. AWS offers an AI-based medical imaging platform on the cloud, and so does Merative (the healthcare analytics company formerly known as IBM Watson). Arterys, a company wholly focused on improving medical imaging with AI, is also another competitor in the space.

Google Cloud’s offering differentiates itself because it is a “comprehensive suite that really helps organizations from start to finish,” according to Hsu Lynch. She also declared that the company is taking a more open, standardized approach than its competitors.

“We’re standardizing around DICOM as the international standard, because in order to make data useful for AI, first it needs to be standardized,” Hsu Lynch said. “We have a managed DICOM web service and DICOM native viewer integration, so everything is built around this standardized approach that helps make it easier for organizations to manage their data and make it useful.”

Photo: Darunechka, Getty Images