Siemens uses Intel's Xeon to Speed Cancer Care Amid Staff Shortages

Healthcare is at a critical inflection point. There’s a demographic shift toward an older population as people live longer, fewer babies are born and a large generation enters its senior years. As people age and cancer screening improves, the rate of cancer diagnoses increases.

Combine those factors with a massive shortage in radiology techs – there’s an 18.1% vacancy rate now, up from 6.2% in 2021, according to the American Society of Radiologic Technologists – and it’s the perfect storm for cancer patients to experience delayed diagnoses, longer treatment plans or the possibility of cancer progression – all packaged with increased anxiety.

But artificial intelligence (AI) is quickly becoming a solution to alleviate the burden on radiologists and speed up treatment for patients. With help from powerful Intel^® Xeon^® Scalable processors, Siemens Healthineers uses artificial intelligence to more accurately spot abnormalities, helping patients receive diagnoses faster and begin targeted treatments sooner.



An image shows an AI scan of a spinal column. The height of each vertebra is measured and compared. Siemens Healthineers’ AI-Rad Companion Chest CT automatically processes these measurements and marks abnormalities with easy-to-understand color-coding. Deviations are highlighted in red and yellow, and bone density is reported. (Credit: Siemens Healthineers via Klinikum Nürnberg, Nuremberg, Germany)


AI Helps Patients Start Treatment Faster
One of Intel’s leading healthcare industry customers, Siemens Healthineers is a global medical technology company focused on bringing innovation to both healthcare professionals and patients. At the Intel Vision event starting March 31, Siemens Healthineers is displaying its AI-Rad Companion software powered by Intel Xeon Scalable processors optimized for AI at the edge.

AI-Rad Companion is a set of AI algorithms that helps clinicians speed up diagnosis and therapy. Using a CT or MRI scan of a patient about to undergo cancer treatment, the AI almost instantly can visually segment out anatomical structures and color-contour, or “shade,” the cancerous cells of a tumor to be targeted for radiation, leaving healthy cells untouched.

“Manually segmenting out these different anatomical structures can take hours or maybe even days for the clinician to do. Now with the AI-Rad Companion, we can do that in a matter of seconds,” says Peter Shen, head of Digital and Automation for Siemens Healthineers. “The impact is that now patients who are anxiously waiting to start their radiation treatment for cancer don’t have to wait days or weeks. Those radiation plans can be put together almost immediately.”

Using Intel’s OpenVINO™ technology, an open source toolkit that can be used to accelerate AI inferencing, AI-Rad Companion’s algorithms have been trained to recognize organs and anatomical data. With that information, the AI can recognize even the smallest abnormality and alert the physician to where and what it is. It can help the doctor diagnose certain diseases in the lungs, brain and prostate, and neurodegenerative diseases like Alzheimer’s. This can reduce the cost of care because it’s speeding up the treatment planning for patients.

The AI-Rad Companion software assists the clinician in identifying, measuring, characterizing and quantifying abnormalities, which can reduce a clinician’s diagnostic interpretation time when reviewing complex imaging cases. And because Xeon Scalable processors are designed for both cloud and server-based applications, medical facilities using AI-Rad Companion can choose which is the most efficient way to access the software.



Two images show AI scans of lungs. In the top image, Siemens Healthineers’ AI-Rad Companion Chest CT detects and highlights abnormal lung nodules. After color-coded segmentation of the lung nodules, the volume, diameters and tumor burden are calculated. In the bottom image, the location of a detected lung nodule is marked with the red dot. This makes it easy for the physician to get a clear picture of the nodule location within the lung. (Credit: Siemens Healthineers via Klinikum Nürnberg, Nuremberg, Germany)

AI Assists Clinicians, Does Not Replace Them
It’s important to note, AI is not taking the place of human thought. A doctor is making the diagnosis, and a clinician signs off on the AI recommendation. But because the AI is so thorough, contour scan results are much more consistent than those from human technicians.

About 95% of the AI-generated contours are deemed clinically acceptable. And the use of AI is becoming more acceptable to the public as well.

A 2023 survey of 1,027 people across four generations in the U.S. found 64% would trust a diagnosis made by AI over a human doctor. That number rose to 82% of Gen Z respondents. And people were most comfortable with AI in the medical imaging analysis aspect.

And the massive time savings provided by AI-Rad Companion frees up radiology technicians to do other critical tasks and move more patients through with the same number of staff.



An image shows an AI scan of a brain. AI-Rad Companion segments out the different brain structures using colors and provides volumetric analysis of each. It then compares the volumes to a normative database and generates a list of the abnormalities or deviations to alert clinicians. (Credit: Siemens Healthineers via Centre d’imagerie diagnostique, Lausanne, Switzerland)

Why It’s Critical Clinicians Get Comfortable with AI
Siemens Healthineers says over 2 million imaging exams have been processed by AI-Rad Companion in hospitals and clinics around the globe1. Increasing adoption is critical to the industry.

Today, more than 1,000 AI algorithms are approved by the FDA2, but few are reimbursed by Medicare or Medicaid. And private insurance companies currently do not pay for AI assistance in treatment at all.

If hospitals and clinics can’t get reimbursed for AI-driven treatments or diagnostics, they’re less likely to invest in these technologies, even though they are proven to help speed diagnosis and treatment. These benefits are then restricted to an exclusive group of providers and patients who can afford them.

That’s why Shen says it’s important to get clinicians comfortable with this kind of technology and educate them on why an AI algorithm has come to the clinical recommendations it has made. If medical workers are comfortable using it, they can properly demonstrate its clinical and economic value through data and patient outcomes, strengthening the case for reimbursement and potentially enacting change.

Shen testified before the Senate Committee on Finance in 2024 and before the House of Representatives Committee on Energy and Commerce in 2023 about the need for a consistent reimbursement approach to incentivize AI adoption in healthcare.

“We're taking the extra steps to make sure that the federal government and different payers recognize the impact of artificial intelligence on patient care,” Shen says. “We've been actively trying to support different ways, including through reimbursement, to drive adoption of AI in healthcare and ensure the accessibility to this innovative technology that helps clinicians in both their diagnosis and treatment.”



An image shows an AI scan of a prostate. The AI-Rad Companion Prostate MR performs an automated segmentation of the prostate to differentiate it from other organs. The radiologist manually marks and characterizes lesions and other targets the AI has found. In this image, the purple line identifies the prostate while the yellow line contours an abnormality, or lesion, that could be cancerous. (Credit: Siemens Healthineers via Clinica Universidad de Navarra, Spain)

1 Data as of April 2024 (Siemens internal metrics)
2 Data as of December 2024 (Read the latest information here).

www.intel.com