Spotlight on AI Manufacturer Siemens Healthineers



Headshot of Peter Shen from Siemens Healthineers

We sat down with Peter Shen, Head of Digital and Automation – North America for Siemens Healthineers, to learn more about Siemens Healthineers' mission, products and their thoughts on transparency in medical imaging artificial intelligence.  Shen is responsible for the overall Digital Health business with a focus on innovation, business development, emerging markets, strategy & commercialization, business management, sales operations, sales support and customer excellence. Shen and his team are focused on the introduction of new and emerging technologies in the healthcare market, including artificial intelligence. 


Q: Can you tell us about Siemens Healthineers and its mission? 

Peter Shen: Siemens Healthineers is a leading medical technology company with more than 120 years of experience and history, bringing a lot of breakthroughs and innovation to market that enable healthcare professionals to really deliver the best care possible for their patients. Our portfolio includes both imaging diagnostics and therapy solutions that are all augmented by digital technologies and artificial intelligence. We’re a global entity with more than 71,000 employees worldwide and about 17,000 employees here in the U.S., and certainly we have a number of AI offerings and touchpoints with patients as well, close to 200,000 patients every hour are engaging with some sort of Siemens Healthineers medical device. 

Q: Your presentation slide [see video] mentions 21,000 patient touchpoints in underserved countries. Can you tell us a little more about that? 

PS: As a global entity, our focus is really trying to bring innovation to all patients including those in rural and underserved population as well, so that’s not just the country level but also within the U.S., trying to make sure we bring enabling technologies like AI to those patients who definitely could benefit from them. 

Q: What is your key product and how can it help radiologists?  

PS: As I mentioned, we have a bunch of different innovations and we’ve been doing this for a long time here. In particular as it comes to artificial intelligence, this is an area where we’ve been working on AI and medical technology for more than 20 years here. As you can see by some of these different statistics, we have an AI Office of Big Data that’s based here in the United States, where we’ve created and maintained one of the most powerful supercomputing infrastructures that are dedicated to artificial intelligence. This allows our research scientists to collect, prepare, organize and even secure deidentified data that we use to train and deliver accurate AI algorithms. And certainly from its inception we try to create and maintain a transparent quality assurance process which involves clinical validation to guarantee the data that we’re using to train those AI algorithms is accurate for diagnosis and treating disease and is really based on a balanced cohort of people of different ages, genders, ethnicities so that we develop reliable and accurate AI algorithms that are unbiased and really reflective of the patient population that they’re going to be applied towards. 

"We have an AI Office of Big Data [...] where we’ve created and maintained one of the most powerful supercomputing infrastructures that are dedicated to AI. This allows our research scientists to collect, prepare, organize and even secure deidentified data that we use to train and deliver accurate AI algorithms."

Q: What makes your product or your company unique?  

PS: For us at Siemens Healthineers, all of our experience around artificial intelligence has really allowed us to create a large portfolio of AI applications and we’re developing those applications not as just standalone AI algorithms, but many of these different AI solutions are integrated into those imaging and diagnostic and therapeutic devices that we create here at Siemens Healthineers. So everything from using AI to help with patient positioning when they’re first interacting with an imaging device, like a CT or MRI scanner, utilizing AI to improve image quality during image acquisition during those exams, and then of course using AI to help the radiologist make a more informed diagnostic decision by providing both qualitative and quantitative data coming from these medical devices, these AI algorithms that are generating this data that helps them visualize or calculate or characterize abnormalities in the images that they see. 

Q: What are your thoughts on the Transparent-AI program?  

PS: We’re really excited about what the ACR is doing around transparency and Transparent-AI in particular. It’s one of the reasons we wanted to be one of the founding members of the Transparent-AI program and transparency activity at ACR, because it really aligns with the AI principles that we’ve established here at Siemens Healthineers around artificial intelligence. Certainly, I think from the decades of experience that we've had with AI, we've come to realize the importance of making sure that there are certain responsibilities that we have as an innovator in artificial intelligence and making sure that those responsibilities or those guardrails are in place there.  We’re really focusing on acknowledging the relationship that the patient has with the clinician and the value of that relationship, and how a technology like AI can actually augment that relationship by really backing up that physician in the clinical decisions that he or she is making to help that patient as well. As I mentioned earlier, we’re really trying to make sure that our AI algorithms are reflective of the patient populations that they're going to be applied towards, so that really is making sure that as we develop these AI algorithms, train these AI algorithms, that they really are utilizing data that that a balanced cohort of different patients, different ages, genders, ethnicities, that really try to minimize the biases that are inherently built into training these AI algorithms. So, it's important for us to really make sure that aspect is happening. 
On the topic of transparency, what's very important for us is also to make sure it's very clear and we make very clear to our end users of our AI solutions, the intended use that's associated with these particular AI algorithms. So, the rationale as to why to use the AI algorithm in the particular clinical scenario and then also really educating our end users, not just on how to use the AI algorithm, but why the algorithm has made the particular clinical recommendations or clinical findings. 

So, really trying to help educate our end users on why the AI has made the determination that it's made — this for us is a really important aspect of transparency. It's not just educating folks on how the AI was created and what data sets were used in everything, but really helping them understand the clinical decision that's behind these different AI algorithms so that they can use that information to have a better understanding of when's the appropriate use, or intended use for these AI algorithms in their routine clinical workflow. 

Q: What can we expect from your company in the next few years? Are there any exciting developments on the horizon?  

PS: We've spent a lot of time discussing AI that's built into image acquisition devices. A lot of the work today, in the field of artificial intelligence, especially within radiology, has been focused on diagnosis and recommended actions by trying to find certain abnormalities and characterize and measure them. Where we're excited is on the further steps where there are opportunities to use artificial intelligence to really focus on the patient and use AI to help in terms of prognosis, or to serve as a predictive kind of analytics related to a disease or abnormality to see if there's a more favorable treatment and then using AI to be able to not just help the individual patient, but then look at a cohort of patients as well. So really looking at the patient population. See if there are patients that are similar where we could be more effective in terms of diagnosis or disease 

Where we see a lot of promise for AI and where we at Siemens Healthineers are doing a lot of work, is really looking at what I would call "multimodal AI.” 

So really looking at utilizing AI not just within one discipline like radiology, but actually combining clinical data that's coming from radiology and from other disciplines like pathology or lab results or even genomic information as well, and then overlaying AI on top of all that clinical data to find correlations within all that data to then help that physician make a more informed diagnostic decision or a more personalized treatment decision for the patient. 

By being able to not only see their radiology images, but also putting in that in context with a biopsy or pathology results and putting that in context with the lab results that they get. We're hoping that the radiologists and other clinicians could actually then make a more informed diagnostic decision on that patient because they have more access to that clinical information.


"Where we see a lot of promise for AI and where we at Siemens Healthineers are doing a lot of work, is really looking at what I would call "multimodal AI.” So really looking at utilizing AI not just within one discipline [...], but actually combining clinical data that's coming from [...] other disciplines as well, and then overlaying AI on top of all that clinical data to find correlations [...] to then help that physician make a more informed diagnostic decision or a more personalized treatment decision for the patient."

Q: Is there anything about Siemens Healthineers that we haven't touched on that you would like to?  

PS: From a Siemens Healthineers perspective, we've had a long history around artificial intelligence where we're excited that our AI solutions are visible not only in the standalone AI products that we offer, like our AI-Rad Companion solutions, but they're also integrated into many of our other platforms as well.  A lot of our imaging devices have AI that's built into them and utilize AI to improve image quality, and our post-processing platforms like syngo.via also integrate a lot of our AI solutions to help the clinician make that more informed diagnostic decision. We’re excited to have those offerings available to clinicians so they can utilize these different technologies regardless of the platform that they have. 

We're also very excited to partner with the ACR here to really make sure that the end user is educated in terms of what's available from an AI perspective. And again, as we talked about the intended use of these AI algorithms, how they're created, and what's the clinical decision behind a lot of these different AI algorithms, these aspects of being transparent around AI are important for us as well.

Q: What do you envision for the future of AI in radiology?  

PS: It's an exciting time to be in radiology, particularly with artificial intelligence. Certainly, a lot of buzz has also been around generative AI and the ability to use generative AI to help within the imaging value chain. 

We see great opportunities as the patient goes through their imaging procedures here and everything from the history of that patient, scheduling that patient, acquiring the images, processing those images, interpreting them, reporting them, and getting the results back to the patient. We see a lot of different opportunities where generative AI could actually play a particular role. 

A great example is from a radiologist’s perspective, being able to use large language models to help really find the specific relevant clinical values in that patient's clinical history that they have, and bringing those to light so that the radiologist just can see those in context with the imaging that they're looking at. 

A great example here is really looking at longitudinal data with different volume measurements that they've made on previous imaging exams so that the radiologist really understands the disease progression for that particular abnormality. And then also putting that in context with other relevant and disease-specific information about the patient that they've done from other departments or other disciplines within the health system as well. 

Putting that all together in the same visualization platform, having a large language model, or summarizing a lot of the findings and some of the other clinical data around that patient — having that accessible for that radiologist — we believe that's going to create a more reliable reading and reporting environment for that radiologist. We see an opportunity with generative AI and AI in general to really make a significant impact within imaging and in particular for the radiologist. 

We're challenged right now with staffing, both from the physicians’ standpoint and also a clinical standpoint here, and unfortunately imaging still continues to grow at a phenomenal rate. We have an aging population that is going to need more imaging services as we try to determine, some of the health care challenges that they have. So there's going to be a lot more demand that's put on our radiologists and also our technologists as well to support all this increased imaging. 

We see then artificial intelligence and generative AI playing a great role here in terms of helping relieve some of the operational burden that the radiology staff has to deal with, so simple things in terms of making quantitative measurements and being able to position patients, quickly and, and to do exams quickly. 

A lot of those operational benefits are there, but then a bunch of different clinical benefits are as well. So really, being able to quantify a lot of abnormalities, automatically finding them for the radiologists so that the radiologists can quickly start to assess what's going on when they're doing their image and interpretation. 

And all of this really leads for both the radiologists and for the technologists to be able to spend more time with the patient and really focus their attention on the patient rather than having to spend all their time dealing with IT solutions or trying to find clinical data about their patient or look for the patient history amongst all the IT systems that are there. 
The goal, at least from our perspective, is to use a technology like AI to try to summarize a lot of those clinical findings, present them in a way that aligns with the clinical workflow that these radiologists have so that they can quickly make those diagnostic decisions and then really focus on the patient going forward. 

To see their full range of medical imaging AI products, visit Siemens Healthineers on AI Central.

This interview is part of a series interviewing medical imaging AI manufacturers on AI Central to help our members better understand the imaging AI marketplace. Since its inception in 2018, the ACR Data Science Institute® AI Central database has evolved from a short online list of FDA-cleared imaging AI products to the most complete and up-to-date online, searchable directory of commercially available imaging AI products in the United States. More than 200 software as a medical device (SaMD) FDA-cleared products have been curated by more than 100 manufacturers, and thousands of radiologists per month access the site in search of suitable AI solutions. Learn more on