Tag

AIpowered

Browsing

Applying artificial intelligence to medical images can be beneficial to clinicians and patients, but developing the tools to do so can be challenging. Google announced on Tuesday that it is ready to take on that challenge with its new medical imaging suite.

“Google pioneered the use of AI and computer vision in Google Photos, Google Image Search, and Google Lens, and we are now making our imaging expertise, tools and technology available to healthcare and life science enterprises,” said Alisa Sou. Lynch, global lead of Google Cloud MedTech Strategy and Solutions, said in a statement.

Jeff Cribbs, Gartner’s vice president and distinguished analyst, explained that health care providers who are looking to AI for diagnostic imaging solutions are typically forced into one of two choices.

“They can purchase software from a device manufacturer, image store vendor or a third party, or they can build their own algorithms with industry agnostic image classification tools,” he told TechNewsWorld.

“With this release,” he continued, “Google is taking their low-code AI development tooling and adding substantial healthcare-specific acceleration.”

“This Google product provides a platform for AI developers and also facilitates image exchange,” said Ginny Torno, administrative director of innovation and IT clinical, assistant and research systems at Houston Methodist in Houston.

“It is not unique to this market, but can provide opportunities for interoperability that a smaller provider is not capable of,” she told TechNewsWorld.

strong component

According to Google, the medical imaging suite addresses some common pain points when developing AI and machine learning models. Components in the suite include:

  • Cloud Healthcare API, which allows easy and secure data exchange using DICOMweb, an international standard for imaging. API provides a fully managed, scalable, enterprise-grade development environment with automated DICOM de-detection. Imaging technology partners include NetApp for seamless on-premises cloud data management and cloud-native enterprise imaging PACS Change Healthcare in clinical use by radiologists.
  • AI-assisted annotation tools from Nvidia and Monae to automate the highly manual and repetitive task of labeling medical images, as well as native integration with any DICOMWeb viewer.
  • Access to BigQuery and Looker to view and search petabytes of imaging data to perform advanced analysis and create training datasets with zero operational overhead.
  • Using Vertex AI to accelerate the development of AI pipelines to build scalable machine learning models with up to 80% fewer lines of code required for custom modeling.
  • Flexible options for cloud, on-premises, or edge deployment to allow organizations to meet diverse sovereignty, data security, and privacy needs – while providing centralized management and policy enforcement with Google Distributed Cloud, enabled by Anthos.

full deck of tech

“One key difference to the medical imaging suite is that we are offering a comprehensive suite of technologies that support the process of delivering AI from start to finish,” Lynch told TechNewsWorld.

The suite offers everything from imaging data ingestion and storage to AI-assisted annotation tools to flexible model deployment options on the edge or in the cloud, she explained.

“We are providing solutions that will make this process easier and more efficient for health care organizations,” she said.

Lynch said the suite takes an open, standardized approach to medical imaging.

“Our integrated Google Cloud services work with a DICOM-standard approach, allowing customers to seamlessly leverage Vertex AI for machine learning and BigQuery for data discovery and analytics,” he added.

“By building everything around this standardized approach, we’re making it easier for organizations to manage their data and make it useful.”

image classification solution

The increasing use of medical imaging, coupled with manpower issues, has made the field ready for solutions based on artificial intelligence and machine learning.

Torno said, “As imaging systems get faster, offering higher resolution and capabilities like functional MRI, it is harder for the infrastructure to maintain those systems and, ideally, stay ahead of what is needed.” “

“In addition, there is a reduction in the radiology workforce that complicates the personnel side of the workload,” she said.

Google Cloud Medical Imaging Suite

Google Cloud aims to make health care imaging data more accessible, interoperable and useful with its medical imaging suite (Image Credit: Google)


She explained that AI can identify issues found in an image from a learned set of images. “It may recommend a diagnosis that then only needs interpretation and confirmation,” she said.

“If the image detects a potentially life-threatening situation, it can also project the images to the top of a task queue,” she continued. “AI can also streamline workflows by reading images.”

Machine learning does for medical imaging what it did for facial recognition and image-based search. “Instead of identifying a dog, Frisbee or chair in a photograph, AI is identifying the extent of a tumor, bone fracture or lung lesion in a diagnostic image,” Cribbs explained.

tools, not substitutes

Michael Arrigo, managing partner of No World Borders, a national network of expert witnesses on health care issues in Newport Beach, Calif., agreed that AI could help some overworked radiologists, but only if it be reliable.

“Data should be structured in ways that are usable and consumable by AI,” he told TechNewsWorld. “AI doesn’t work well with highly variable unstructured data in unpredictable formats.”

Torno said that many studies around AI accuracy have been done and will be done further.

“While there are examples of AI being ‘just as good’ as a human didn’t have, or being ‘just as good’ as a human being, there are also examples where an AI misses something important, or isn’t sure.” That’s what to interpret because there may be many problems with the patient,” she observed.

“AI should be seen as an efficiency tool to accelerate image interpretation and assist in emergent cases, but should not completely replace the human element,” she said.

large splash capacity

With its resources, Google can have a significant impact on the medical imaging market. “Having a major player like Google in this area could facilitate synergy with other Google products already in place in healthcare organizations, potentially enabling more seamless connectivity to other systems,” Torno said.

“If Google focuses on this market segment, they have the resources to make a splash,” she continued. “There are already many players in this area. It will be interesting to see how this product can take advantage of other Google functionality and pipelines and become a differentiator.”

Lynch pointed out that with the launch of the medical imaging suite, Google hopes to help accelerate the development and adoption of AI for imaging by the health care industry.

“AI has the potential to help reduce the burden for health care workers and improve and even save people’s lives,” she said.

“By offering our imaging tools, products and expertise to healthcare organizations, we are confident that the market and patients will benefit,” he added.

A new service powered by artificial intelligence that can turn portraits into talking heads was announced by D-ID on Monday.

Called Creative Reality Studio, the self-service application can convert a facial image into video, complete with speech.

The service is aimed at professional content creators – learning and development units, human resources departments, marketers, advertisers and sales teams – but anyone can try out the technique on the D-ID website.

Creative Reality Studio
Video by John P. Mello Jr.


The platform reduces the cost and hassle of creating corporate video content and provides an unlimited variety of presenters – versus limited avatars – that include users’ own photos or any images that the company has the right to use, according to the company. Gained notoriety when its technology was used in an app called Deep Nostalgia. The software was introduced as a way to animate old pictures.

The company said the technology enables customers and users to choose a presenter’s identity, including their ethnicity, gender, age and even their language, accent and tone. “It provides greater representation and diversity, creating a stronger sense of inclusion and belonging, which drives further engagement and interaction with the businesses that use it,” it said in a news release.

Matthew Kershaw, D-ID Marketing Vice President, told TechNewsWorld, “The use cases include empowering professional content creators to seamlessly integrate video into the digital space and presentations with specialized PowerPoint plug-ins, the use of customized corporate video narrators.” Generating more engaging content.

impressive services

The quality of these services is impressive, and continues to get better, maintained Daniel Castro, vice president of the Information Technology and Innovation Foundation, a research and public policy organization in Washington DC.

“The service isn’t at a level where it’s completely replacing a presenter, but there’s no reason not to expect it to be there relatively soon,” he told TechNewsWorld.

D-ID explained that the use of video by businesses has increased dramatically and more of them are integrating it into their training, communication and marketing strategies.

Accelerating this trend, it continued, are the rapidly evolving worlds of avatars and the metaverse, both of which demand a more creative, immersive and interactive content approach from digital creators. Production budgeting, however, can be prohibitively expensive and requires significant allocation of time and talent.

“The service is an evolution of the avatars and emoji people use today, but can be used in lengthy discussions or presentations,” said Ross Rubin, principal analyst at Reticle Research, a consumer technology consulting firm in New York City.

“The idea is to save time, especially if you were going to read a script,” he told TechNewsWorld. “It can be more engaging to an audience than simply watching audio or slides.”

democratizing AI

D-ID CEO and co-founder Gil Perry noted in a news release that the company’s technology, which is limited to the enterprise, has been used to create 100 million videos.

“Now that we are offering our self-service Creative Reality platform, the potential is enormous,” he continued. “It enables both large enterprises, small companies and freelancers to create personalized videos for multiple purposes on a large scale.”

Kershaw said D-ID’s technology will further democratize creativity. “I say ‘forward’ because technology has really been democratizing the arts for decades,” he said.

“From the installation of synthesizers, samplers and sequencers in music to Photoshop and Illustrator in photography and illustration, and premiere and desktop editing in film production and motion graphics, the ability to create high-quality productions outside of specialist high-end studios It’s been happening since the 1980s,” he said. “This is the latest episode of that long-running series.”

“This is certainly a step forward towards democratizing AI,” agreed Aviva Litton, a security and privacy analyst at Gartner. “It has great use cases in education, healthcare and retail,” she told TechNewsWorld. “It’s a better way to communicate with people. We’re becoming a more visual society. Nobody has time to read anything.”

deepfake concerns

With growing concern over the use of “deepfakes” to spread misinformation and take social engineering to new heights, there is always the potential for misuse of new synthetic media solutions such as D-ID.

“As with any technology, it can be used for the ill by our bad actors, but our platform is aimed at legitimate businesses that would have no interest in that kind of use,” Kershaw said.

“Plus,” he continued, “we’re not deepfakes. We don’t put someone else’s face on someone else’s body, and we’re not trying to tell anyone something they didn’t say.”

“Within D-ID’s platform, we have put in place a number of security measures to ensure that our technology is not used in this manner,” he said. “We do not repeat the voices of celebrities or those without permission from any person.”

The company also filters abusive and racist comments, and prohibits the platform from being used to make political videos.

“D-ID is putting railings on their platforms, but we all know that railings are never perfect,” Litton said.

“It is a good tool to spread misinformation because these social media sites are not ready for deepfakes,” she said. “Even if social media sites are good at detecting deepfakes, they will never be enough. It’s like spam. Spam always gets through. It will happen too, but the consequences There will be worse.”

need for origin

Detecting deepfakes is a losing proposition in the long run, Litton said. Even today, detection algorithms typically cannot detect more than 70% of deep fakes.

He added that determined adversaries will keep pace with deepfake detection using generative adversarial networks so that the detection rate is eventually reduced to 50%.

She predicts that in 2023, 20% of successful account takeover attacks will use deepfakes to turn over sensitive data to socially engineered users or transfer funds to criminal accounts.

“Many safeguards need to be implemented industry-wide, which is why we are also working with industry bodies and regulators to implement legal safeguards that will make the industry more secure and reliable in general ,” said Kershaw. “We think that having an industry-wide system for watermarking content invisibly through the use of steganography, in particular, would get rid of almost all potential issues.”

“You will be able to see a section of media and click a button to see where it came from and what’s in it,” he said. “Transparency is the solution.”

“There are many ways to deal with counterfeiting, but the most important is to know the origin and authenticity of the media,” Castro said.