Rethink Imaging

EP 55 • October 15, 2026

Moving Beyond Compliance: Implementing Medical Physics 3.0 in Patient-Centered Radiology

Featured Guest

Dr. Ehsan Samei, PhD, DABR, FAAPM, FSPIE, FAIMBE, FIOMP, FACR

Medical Physicist • Duke University Health System

Listen Now32 min

Also on:

Let's Talk

Beyond dose

Ask us what your imaging data can actually show you.

Start a Conversation

Share This Episode

Subscribe to Rethink Imaging

Medical imaging operations often structure protocol development and quality assurance around compliance standards and machine specifications. Host Chris St. John sits down with Dr. Ehsan Samei to discuss how medical physics can expand beyond compliance box-checking to serve individual patient needs across clinical operations.

Dr. Samei outlines the framework of Medical Physics 3.0, arguing that clinical physicists, radiologists, and technologists must collaborate as equal partners. He advocates shifting from machine-centric testing to indication-specific protocol design, bridging the historical gap between research physics and clinical practice to ensure images reliably answer specific diagnostic questions.

Host

Chris St. John

Host, Rethink Imaging • Imalogix

Featured Guest

Dr. Ehsan Samei, PhD, DABR, FAAPM, FSPIE, FAIMBE, FIOMP, FACR

Medical Physicist • Duke University Health System

Watch the Episode

Key Takeaways

Medical Physics 3.0 shifts professional focus from machine compliance to patient-centered, indication-specific imaging.
Protocol management should prioritize clinical indication rather than machine models or body parts alone.
Medical physicists serve as translators across scanner parameters, task-generic attributes such as noise and resolution, and task-specific diagnostic requirements.
Bridging the gap between research physics and clinical practice helps relevant scientific advances reach patient care.
Radiologists, technologists, and physicists deliver stronger outcomes when they work as equal partners across the imaging workflow.

Full Transcript

Rethink Imaging Podcast Transcript

Guest: Dr Ehsan Samei

Host: Chris St. John

DR EHSAN SAMEI 00:00:00 I don’t care about radiation dose and image quality and stuff that physicists talk about. I care about when my physician measures the size of the tumor, can I trust it or not? And I thought to myself, I thought that’s what we were doing. I just realized we never translate it from our image quality of space to the actual clinical task basis space that the clinicians are asking for.

CHRIS ST. JOHN 00:01:43 In this more patient-centered imaging model, how would the entire imaging community begin to distill that level of detail when imaging different patients? Obviously, there are going to be discrepancies and disagreements, and people have different opinions; I’m taking that for granted. But how do the roles of the radiologist, the tech, and the physicist come together to distill the necessary protocols within all of these tags? How do you move through the process of figuring out what is working better for who? It just feels like a gargantuan task.

DR. EHSAN SAMEI 00:02:15 It seems overcomplicated, but it’s actually not. Think about when we plan a surgery or when we plan a radiation therapy process. You have a specific need of the patient at hand. The tumor of this patient is in a different location than the tumor of the next patient, so the treatment plan needs to be adjusted accordingly. There is a dosimetrist, a physicist, and a radiation oncologist. The radiation oncologist provides a primary medical prescription that needs to be implemented. The physicist is given the responsibility to make sure the radiation is appropriated in such a way that meets the goal of the radiation oncologist.

Of course, sometimes those goals cannot be met, so we go back to the drawing board and make adjustments accordingly. Then dosimetrists try to do the calculations, and there are a couple of iterations back and forth. The process is not super inefficient, especially when we put automated processes into the workflow, and then we do right by the patient to the best of our ability.

Now, can we adapt something somewhat similar in radiology? I think we could. There is one big problem: the order of magnitude of the patients that we have in radiology is significantly higher than in radiation oncology. So that customization happening in radiation oncology or in surgery is more difficult to do in radiology. But this is what we are called to do, and a mini version of this is already happening when a referring physician talks to a radiologist. A referring physician wants something to be imaged, but they don’t know quite what to order, and the radiologist often looks over their order and makes certain adjustments. Sometimes technologists look over the order, too.

A mini version of this is already happening. I think we need to standardize that process, and in that workflow, the physicist needs to be an integral part because the physicist is the one who supposedly understands the equipment completely inside and out. In the same way that a radiation oncology physicist understands how to appropriate radiation for therapy purposes, a physicist needs to fully understand how to appropriate radiation in radiology imaging for the benefit of the patient. The model is there; we just need to implement that model in radiology as well.

CHRIS ST. JOHN 00:05:02 What about in the interim? What about now, tomorrow, or next week? If I’m a physicist at either a small facility or somewhere large and academic, what steps can I be taking now to get more feedback from my radiologist who is underwater because they are seeing more exams than they’ve seen in their entire career? What steps can physics be taking now within the context of imaging today?

DR. EHSAN SAMEI 00:05:43 Very good question. I would start with protocols that are already at the facility. Look at the protocols and look at how large the buckets of the protocols are. If you’re a physicist, sit down with your radiologist to see if that bucket needs to be chopped into smaller buckets or not. They might say there are really three different types of patients that fall into that bucket. If they’re annoyed by how one group is imaged poorly while other groups are imaged well, that gives you an indication of a scope for that group. If the images for all the bariatric patients are of poor or inadequate quality, that’s the first thing to address.

Then I would look across different makes and models of devices available to see how much incongruence or congruence exists across different protocols to essentially make the protocols indication-specific—not machine-specific or body-part-specific, but indication-specific. It’s for the sake of that indication. Of course, there might be incidentals that you also detect, but because you don’t know what the incidentals are, you don’t know how to orchestrate your examination accordingly.

That’s where I would start. Then look at metrics of quality that matter, such as noise, resolution, or contrast, and try to orchestrate your protocols to deliver that desired image quality, not to set the or dose to a certain level. Those are means toward an end.

Why does image quality need to be characterized in terms of resolution and noise and not tumor size? Because we don’t know what tumors we are encountering from one exam to another. You can think about those metrics of image quality, such as noise and resolution, as task-generic attributes that reflect the quality of the images available to us, which can then be appropriated toward the clinical task.

You can think about three zones that need to be translated toward one another:

The set of parameters at the scanner level, which ordains task-generic attributes of image quality like noise, resolution, and contrast.

The task-generic attributes of image quality itself.

The task-specific attributes, such as how accurately I can measure lesion sizes or levels of calcification.

The job of a physicist is to translate from that first space to the second space to the third space. You go from English to French to German, if you want. Who is the most qualified person to do translation in that clinical space? That is the job of a physicist. The physicist is a translator between those three different spaces.

CHRIS ST. JOHN 00:09:01 This all tracks for me. I keep coming back to the relationship between radiologists, physicists, and techs. I’ve heard anecdotes about the concept of “good enough” from the perspective of radiologists reading an image. What is your relationship to this concept of “good enough” when it comes to medical imaging? Finding what you need to find theoretically is great and good enough, but in these larger conversations about dose and trying to optimize images, what is your relationship to “good enough”?

DR. EHSAN SAMEI 00:10:00 That’s a very fair question. I’ve thought about this. In medical imaging, at least the way it’s practiced right now, there is a need for some degree of frugality, which is where “good enough” needs to be visited.

By frugality, I mean that we are still somewhat obsessed with radiation dose in that we don’t want to use more than what we need. We are also obsessed with imaging time, for example in the context of MR, because spending too much time costs money and you want to get the patient in and out sooner. There is also the element of motion artifact; rarely can you get an image that is not affected by motion artifacts, so doing the imaging quicker is another factor involved. Not only do we have many degrees of freedom in how we acquire images, but there is also a call to use the least amount of time, radiation, and money to get what we need.

Radiation risk is a big factor for any imaging that uses ionizing radiation. People have been obsessed about radiation exposure for the thirty-plus years I’ve been in this field, and that will never go away, especially when patients are young or need repeated exams. If a patient has had ten exams, even knowing that the eleventh exam carries no higher cumulative risk model than the tenth, you cannot erase that history. The research we’ve done shows patients care about that, and we cannot ignore what patients care about.

So “good enough” is a good concept. I am not advocating for over-engineering the process, and physicists definitely have the tendency to over-engineer and complicate things. But there is a degree of care that needs to go into imaging patients. The trust that exists between the medical establishment and the patients who trust their bodies to us requires scrutiny and care to make sure the resources we use—in this case, radiation—are best appropriated for the care of the individual patient, not an average pediatric patient out there. For a neonate in the clinic, I am obligated to live up to the trust the caregiver has given me to make sure I use the exact amount of radiation needed to best care for that patient, and not an iota more.

CHRIS ST. JOHN 00:13:43 A lot of what we’ve been talking about without identifying it directly is Medical Physics 3.0, which is medical physics for every patient. Do you mind touching on it, explaining the concept and what’s been going on?

DR. EHSAN SAMEI 00:14:11 In the early days of medical physics, a lot of our energy was focused on the engineering aspects of image formation and imaging devices. Honestly, the technologies were bad back then, so they needed a lot of care. The technology has matured significantly since those days when X-ray tubes were not very reliable and we spent a lot of energy making sure their output was accurate and within certain limits. We put regulations in place to ensure devices met those requirements, and we still practice that today.

I call that “physics for machines” or “physics for compliance”—making sure our devices comply with standards we established decades ago. However, some of those compliance requirements are no longer really needed and have become somewhat irrelevant. We need to shift from caring for machines to caring for patients. It’s not just about physics; it’s about physics for medicine, or physics for the patient and accommodating the individual needs of the patient. Personalization means directing our efforts toward the individual needs of patients.

The next wave we are advancing toward is physics with patients. When I say “physics for patients,” it implies physics is the subject of the good being offered to the patient. But the patient also has a voice. How a mother thinks about the dental X-ray her child is going to get should matter in the way we practice medicine, so that needs to be incorporated into the process. At the end of the day, we are here because patients and their caregivers trust us, so we need to listen to what they say.

Medical Physics 3.0 essentially swings our efforts away from focusing primarily on compliance and machines—even though that is still our stronghold—and orients those tools toward patient care.

I remember a conversation I had with a senior administrator at MD Anderson who told me, “Ehsan, I don’t care about radiation dose and image quality and the stuff you physicists talk about. I care about whether I can trust it when my physician measures the size of a tumor.” I thought to myself, “I thought that’s what we were doing!” Then I realized we never translated our work from the image quality space to the actual clinical task space where clinicians operate.

It’s not a huge translation, but the transition needs to happen. Going back to what I said earlier, there are three spaces:

Machine parameters and machine specifications.

Image quality attributes and the reliability of the information gained from images.

How that information is relevant to the care of the individual patient, whether it’s a cardiac, lung, or abdominal patient.

We need to be able to confidently translate from one space to the next. Medical Physics 1.0 and 2.0 focused on the earlier stages, but Medical Physics 3.0 goes all the way across that entire spectrum.

CHRIS ST. JOHN 00:18:29 How does the translation across those spaces happen? How did your thinking change, and how would you teach someone to think in this way?

DR. EHSAN SAMEI 00:18:43 There is a whole science behind this. For different indications, there are different definitions of what we call an imaging task, or task-based imaging. You can say, “I am looking for something that is typically 1 centimeter and has 2%, 3%, or 5% contrast.” That is a reference task, and we never want to miss something as obvious as that. What image quality needs to be present in order to make that task discernment? What noise, resolution, or contrast do I need? Then you back-calculate how to set up the machine to ensure those attributes are delivered.

There is a science behind each of those steps. Books and articles have been written about it, and you can apply those principles. The problem is that most of those articles and methods have been out in the research literature for decades but have never made it into clinical activity. That’s another dichotomy in the field: research medical physicists do research, publish papers, and feel like they’re done, while clinical physicists spend most of their time making sure machines are compliant based on old standards. These two spaces do not meet. Part of Medical Physics 3.0 is bridging the gap between research medical physics and clinical medical physics.

CHRIS ST. JOHN 00:20:32 You’ve even said that the split between research physics and clinical physics is artificial and detrimental.

DR. EHSAN SAMEI 00:20:44 The irony is that as clinical physicists, we have credentials to the extent that what we do is scientific. As a clinical physicist, you have no excuse not to use the highest and best science in medical physics and apply it to your clinical environment. You need to be backward-oriented toward research.

As research medical physicists—and I can speak to this because I am both—we justify getting funding and grants by saying the amazing things we are going to do are clinically relevant and will affect patient care. Research physicists are forward-oriented toward patient care to justify their research. Even though we pay lip service to this professionally, we don’t always practice it.

CHRIS ST. JOHN 00:21:36 How do you think about teaching that distinction? You have trained tons of practicing physicists. A lot of people get into physics thinking it’s going to be one thing, and then it ends up being a lot of compliance work and box-checking. As an educator working with your students, how do you prepare them for this split or the different paths they may choose?

DR. EHSAN SAMEI 00:22:23 I do that all the time. When looking at anything with a research mindset, if a student or trainee wants to do something, I always ask the question: “So what?” Are you doing that just to publish a paper? Of course you will publish a paper along the way, but that’s not the ultimate purpose. The “so what?” needs to be fully answerable for any research activity, and it needs to be directed at a pathway to ultimately reach a patient. Your research work needs to affect the care of individual patients; otherwise, don’t call yourself a medical physicist—just call yourself a physicist.

“So what?” is the question for research physicists. “Why?” is the question for clinical physicists. When clinical physicists perform measurements, they need to ask: Why is this measurement reliable? How do you know? How can you be certain it’s relevant to patient care? They need to draw back on the latest science in that space.

The fact that there is a dichotomy between research physics and clinical physics shouldn’t surprise us. As human minds, we differentiate and specialize. You go to a kidney doctor, ask them about your liver, and they say, “Don’t talk to me about the liver; I only care about your kidney,” because they have to study so much about the kidney that they can’t focus on anything else. It’s the same thing here: research physicists specialize in research, and clinical physicists specialize in the clinical space.

Even though that’s natural, we have an equal responsibility not to lose the big picture. Medical Physics 3.0 is about looking at the big picture—that both spaces are there to serve the whole person and the whole of medicine.

We have a similar problem in medical practice itself. Integrative medicine is a big topic now. Every time someone uses that term, you should ask why “integrative medicine” needs to be a thing. It implies medicine wasn’t integrated before. It’s like “evidence-based medicine”—the first time I heard that term, I thought, “Are you telling me medicine was practiced without evidence?”

CHRIS ST. JOHN 00:25:32 I quote you all the time when my friends complain about health care and their individual journeys. I always pull out my favorite Dr. Samei quote: “Medicine is just educated guessing.” You told me that the first time we recorded together, and it has stuck with me ever since.

DR. EHSAN SAMEI 00:26:03 We do need to specialize, but we also need a process by which we can cross the boundaries of our specialization. It’s a bit like a left-brain/right-brain scenario: the left brain looks at details, while the right brain looks at the world holistically. You need both. That’s why we have two hemispheres; they aren’t redundant because they have two different focus areas.

CHRIS ST. JOHN 00:26:36 Coming back to the collaboration between radiologists, technologists, and physicists in imaging: from the perspective of a radiologist or technologist, what should they be going to their physicist for? We often frame this around the physicist’s responsibility to be proactive, but what should clinicians and techs be bringing to the physicist?

DR. EHSAN SAMEI 00:27:08 The answer is very straightforward. A radiologist needs to see certain diagnostic details if a patient is afflicted by whatever condition they suspect. There is always an indication that causes an image to be ordered; patients don’t just go to a hospital and say, “Image me.” Something is going on, guesses are involved, and they perform imaging to affirm or deny those guesses (or occasionally perform exploratory imaging).

Radiologists should go to the physicist and say, “Make it so. I want to be able to see that specific sign or lesion if it’s there. I want to see it confidently, and I want this imaging exam to answer my clinical question.” In the same way a radiation oncologist goes to a medical physicist and says, “I want this tumor treated with this amount of radiation, these healthy areas spared, and the dose not to exceed this level—make it so,” a radiologist should go to the physicist and say, “Make it so.”

Now, there are expert radiologists who can answer those technical questions themselves—radiologists who are effectively physicists and radiologists at the same time. If you know how to do it yourself, by all means. But in the same way that I go to a statistician for a complex statistical need, most of the time clinical imaging questions are very complicated, and there is a whole workforce specialized to answer that exact question: “How can I make it so?” Go to that specialist and ask for their help.

The one caveat that scares me is that because physicists haven’t routinely been asked to answer that question, some physicists in the workforce might feel unequipped to answer it. The responsibility is on us as medical physics educators to make sure our curricula equip the next generation of physicists to answer what radiologists and referring physicians need. We are medical technology translators.

In that dance, we are dealing with equal partners working together. Nobody is subservient to another; we are one and the same, serving the patient and working at the mercy of the patient’s needs. Technologists are also vital—there are amazing technologists who know the ins and outs of how equipment attributes change the rendition of an image, and they need to be in that space.

In any human interaction, there is a degree of power dynamics where we bargain, delegate, or exercise power. The best partnerships happen when we share space as equal partners. Radiologists, physicists, and technologists each have an integral role to play. When you build a house, the plumber isn’t better or worse than the electrician or the HVAC technician; without all of them, you don’t have a house. They might not get paid the same amount, but you need them all, and you need to treat them with respect and honor their expertise.

CHRIS ST. JOHN 00:31:55 In that case, this is a beautiful place to end today. Dr. Samei, thank you so much for being back on Rethink Imaging. It is always a pleasure, and we can’t wait to talk again.

DR. EHSAN SAMEI 00:32:08 Thank you so much, Chris. You’re a great person to talk to, and I appreciate the time we had together.

CHRIS ST. JOHN 00:32:11 Thank you so much.

Questions from This Episode

What is Medical Physics 3.0?
Medical Physics 3.0 is an operational framework that shifts medical physics from machine-centered compliance toward patient-centered care. It connects physics expertise to clinical tasks so imaging protocols and image quality attributes support specific diagnostic needs.
How do indication-specific protocols differ from traditional protocol design?
Traditional protocols are often organized by machine model or body part. Indication-specific protocols align scanner settings, dose, noise, contrast, and resolution with the clinical question so teams can pursue consistent diagnostic confidence across patients and devices.
How can radiologists and medical physicists collaborate more effectively?
Radiologists can define the diagnostic details needed for a clinical indication, while physicists translate those needs into scanner settings and measurable image quality requirements. Technologists add practical equipment expertise, making all three equal partners in the workflow.