Imaging Informatics Unplugged

What happens when a radiologist, a former hacker, and a cybersecurity leader live in the same person? You get Dr. Benoit Desjardins, and this conversation is one of the most eye-opening we've had on Imaging Informatics Unplugged. Benoit is a Professor of Radiology at the University of Montreal and the first Chief Medical Cybersecurity Officer in Canada at CHUM. He's a consultant for the FBI, a 2025 CEH Hall of Fame inductee, and a past NSA Codebreaker Challenge winner.

If you manage a PACS, run an imaging department, or support radiology IT, hit play. This one will change how you think about your environment.

Here's some of what we dig into. Benoit shares how a packed DEF CON room in 2017 pulled him back into hacking, and why he frames healthcare cybersecurity as a patient safety issue rather than an IT one. We talk about what really happens to care when imaging goes dark, including the loss of priors and delayed diagnoses. We also compare the US and Canadian approaches to ransomware, why paying a ransom is a risky bet, and how double extortion works when you have good backups.Then we get into the imaging-specific stuff. Benoit explains how he put PHP code into the 128-byte DICOM preamble and took over a test PACS, proving that an image can be part of your attack surface. We cover practical steps your PACS team can take on Monday morning: treat incoming DICOM as untrusted, validate and sanitize inputs, patch the apps that parse images, segment older infrastructure, and monitor for odd behavior. We also get honest about DICOM security, why encryption is often off by default, and the key management problem nobody wants to own.We wrap with what secure by design means for a province-wide PACS in Quebec, what it takes to train a true cyber physician, and Benoit's warning about AI-powered hacking tools and what hospitals should expect over the next six to eight months. It touches AI in radiology, enterprise imaging, VNA strategy, and vendor procurement along the way.If conversations like this make you want to go deeper on the fundamentals, check out the Imaging Informatics Primer course, the CIIP Foundations Program, and the hands-on DICOM Learning Lab at nagelsconsulting.com. They're built for people like you who live in PACS, DICOM, HL7 and Radiology IT every day.Learn more at nagelsconsulting.com#ImagingInformatics #PACS #DICOM #HealthcareIT #RadiologyIT #Cybersecurity #HealthcareCybersecurity #Ransomware #EnterpriseImaging #MedicalImaging #PatientSafety #RadiologyAI #InfoSec #SecureByDesign #CHUM

Dr. Les Folio has one of the most unusual resumes you'll hear on this show. Before he ever read a scan, he was a volunteer firefighter who got burned on his first field fire, a drummerwho hauled his gear around in a hearse, and an X-ray tech who got hooked watching a CT scanner run its first head scan back in 1977. He spent 20 years in the Air Force as a colonel, deploying to some of the busiest combat hospitals in Iraq and serving as a NASA liaison, before becoming a chest radiologist and imaging informatics section chief at the James A. Haley VA Hospital, a professor at USF Morsani College of Medicine, and a longtime board member of the Society for Imaging Informatics in Medicine. He's published more than 150 peer-reviewed papers and remains one of the most active mentors in theSIIM community.

Les and Jason start with reading room ergonomics: research Les has led comparing radiologist workstation setups to aircraft cockpits and NASA mission control, and why four monitors, notthree, became the standard in so many reading rooms. From there they dig into a naming problem Les sees nearly every day, CT orders labeled “high resolution” that actually route patients to roughly six times the radiation of a standard low dose scan, and the SIIM hackathon team that tackled it when two competing groups, Rad Guide and Les's own VA team, merged instead of competing.

They also cover PACS Man, the crowdsourced lesion-segmentation game that turned a SIIM showcase into aleaderboard-fueled frenzy, and what CIIP candidates consistently underestimate about the gap between passing the exam and being genuinely useful on the job.

Les shares a story from a combat hospital in Iraq where point-of-care ultrasound caught a pneumothorax a chest X-ray would have missed, plus an update on SIIM's POCUS work group and the push to formalize POCUS manager roles. He closes with his radiology pre-processing course, which flips traditional shadowing on its head, and a preview of the space medicine work he's doing with NASA flight surgeon Jeff Jones.

If PACS, DICOM, HL7, or enterprise imaging AI are part of your world, this one is worth the full listen.

Jason also has an update from nagelsconsulting.com: the Imaging Informatics Primer course is a great startingpoint if you're new to the field, the CIIP Foundations material is built for anyone prepping for certification, and the hands-on DICOM Learning Lab gives you live, practical reps with the standard itself.

Learn more at nagelsconsulting.com

#ImagingInformatics #PACS #DICOM#HealthcareIT #RadiologyIT #CIIP #SIIM #POCUS #RadiologyAI #MedicalImaging#HealthIT #MedicalEducation #EnterpriseImaging #CombatRadiology

4. Hashtags

Quick one this week: no guest, just me and a question that's been rattling around my head for a while. Does Shazam actually hear the song it's identifying? Short answer: no, not even a little. And once you understand what it's actually doing instead, you won't look at your MRI scanner or your ultrasound machine the same way again.This is a solo lecture episode where I trace one piece of math, the Fast Fourier Transform, from a guy studying heat in the early 1800s, through Carl Gauss casually stumbling onto the same trick, to a 1965 breakthrough that made it fast enough to matter, and straight into the guts of how your MRI and Doppler ultrasound actually work.Here's the thread I pull on in this episode. Music is just air pressure moving over time, a messy waveform with no obvious structure. Shazam runs that waveform through the Fast Fourier Transform, or FFT, which breaks it apart into pure frequencies, the mathematical DNA of the sound. That's the whole trick behind catching a song in a loud room.Now swap out the song for a body. MRI doesn't collect an image the way a camera does. It collects raw frequency data in something called K space, and if you've ever looked at raw K space data, it looks like static, not anatomy. FFT is what turns that static into an actual image you can read clinically. No FFT, no MRI image, period.Doppler ultrasound works the same way from a different angle. It's looking at frequency shifts caused by moving blood, and FFT is what converts those shifts into velocity, direction, and flow, so you're watching physiology happen in real time instead of just staring at vessels.I also get into the history, because it's a fun one for anyone who likes the origin story behind the tools they use every day: Joseph Fourier's work on heat transfer in the early 1800s, Gauss quietly beating everyone to the punch, and the 1965 Cooley-Tukey algorithm that finally made this fast enough for real-time medical imaging, and eventually, for your phone in a grocery store.If you work anywhere near PACS, radiology IT, or imaging informatics, this is one of those under-the-hood concepts (DICOM, HL7, and enterprise imaging systems all sit downstream of it) that's worth actually understanding rather than just accepting as a black box.If you want to go deeper on the fundamentals behind the imaging tech you work with every day, that's exactly what we built at nagelsconsulting.com. The Imaging Informatics Primer course is a great starting point if you're looking to build a solid foundation, the CIIP Foundations material will help if you're working toward certification, and the hands-on DICOM Learning Lab lets you get in and actually work with the data instead of just reading about it.Learn more at nagelsconsulting.com#ImagingInformatics #FFT #FastFourierTransform #MRI #Ultrasound #DopplerUltrasound #RadiologyIT #MedicalImaging #HealthcareIT #EnterpriseImaging #PACS #DICOM #RadiologyAI #SignalProcessing #MedicalPhysics

Michael Rosenberg has spent the last decade obsessed with a problem most of healthcare IT would rather not think about: how do you get a medical image from one institution to another, safely and without a human babysitting every step? An industrial and systems engineer by training, Michael cofounded Medicom Technologies in 2015 out of North Carolina State University, and as CEO he and his team built the company around what they call Enterprise Imaging Interoperability. Their software now runs across every VA medical center in the country along with some of the nation's leading IDNs and academic medical centers. In this conversation, Jason sits down with Michael to dig into why some of the most well-intentioned efforts in image sharing, the ACR's Ditch the Disc campaign among them, never fully solved the problem, and what it will actually take to get there.They start by grading the ACR's Ditch the Disc initiative: strong on messaging, short on the policy and mandates needed to change workflow. From there, Michael explains why three-point patient matching (name, date of birth, and one other identifier) is still common in DICOM image exchange, and why he considers it unsafe. He walks through Medicom's FHIR-based approach for pulling additional demographics from the EHR before a match is made, and why the study instance UID deserves treatment as a universal identifier across systems.Jason and Michael also dig into what a national scale rollout looks like, drawing on Medicom's work across the VA network, and how that compares to centralized models already in place in Canada and parts of Europe. From there they cover patient-driven image sharing, including a surprisingly high adoption rate for patient upload tools, the privacy risks buried in supposedly de-identified imaging data, and what happens when interoperability moves beyond radiology into pathology, cardiology, and oncology, where the workflows barely resemble each other. They close with a look at where imaging regulation is headed, including proposed ONC certification requirements that could finally make interoperability mandatory rather than optional.If PACS, DICOM, HL7, VNA, or enterprise imaging AI are part of your world, this one's worth the full listen.Jason also has a quick update from nagelsconsulting.com: the Imaging Informatics Primer course is a great starting point if you're new to the field, the CIIP Foundations material is built for anyone prepping for certification, and the hands-on DICOM Learning Lab gives you live, practical reps with the standard itself.Learn more at nagelsconsulting.com#ImagingInformatics #PACS #DICOM #HealthcareIT #RadiologyIT #EnterpriseImaging #VNA #HL7 #RadiologyAI #MedicalImaging #Interoperability #PatientMatching #FHIR #HealthIT #DitchTheDisk5. Chapter List (Timestamps)Note: chapter times are offset by 1:23 to account for the intro script

Norman Young has spent 25 years building software at the intersection of imaging and medicine, but the story behind his work starts with his own diagnosis. In 1992, as a third-year engineering student at the University of Toronto, Norman was diagnosed with Hodgkin's lymphoma, an experience that reshaped the rest of his career. After working on eFilm, one of the first free PACS viewers, at University Health Network, he founded ClearCanvas in 2005, an open-source medical imaging platform still used by clinicians around the world. In 2021, the American College of Radiology recognized his work with their Global Humanitarian Award. Today Norman is building something new: CareChorus, a nonprofit aimed at helping patients and caregivers make sense of their own medical records.Jason and co-host Mohannad Hussain sit down with Norman to trace that whole arc. He explains why ClearCanvas went beyond free software into full open source, what it took to build a professional-grade PACS and RIS largely on his own before landing the hospital contract that got the company off the ground, and what it was like watching download counts climb from every continent, including, improbably, Antarctica. They also get into how AI coding tools like Claude Code and Codex are changing what a small team can ship, and why Norman still insists on reviewing every architectural decision even when the AI writes the first draft.The back half turns personal. Norman walks through the two stories, one about a friend's husband facing a serious diagnosis, the other about a colleague who passed away after struggling to get coordinated care, that led him to found CareChorus. He breaks down the four problems the platform is built to solve, from organizing scattered medical records to helping patients actually understand what their reports mean, and explains why he built it as a nonprofit with data that stays on the user's own device rather than in the cloud.If PACS, DICOM, open-source imaging informatics, or healthcare IT are part of your world, this conversation is worth the full listen.Jason also has an update from nagelsconsulting.com: the Imaging Informatics Primer course is a great starting point if you're new to the field, the CIIP Foundations material is built for anyone prepping for certification, and the hands-on DICOM Learning Lab gives you live, practical reps with the standard itself.Learn more at nagelsconsulting.com#ImagingInformatics #PACS #DICOM #HealthcareIT #RadiologyIT #OpenSource #EnterpriseImaging #HealthIT #MedicalImaging #PatientAdvocacy #AIinHealthcare #CIIP #DigitalHealth #HealthTech

What happens when a remote Arctic community with zero medical radiation technologists needs a chest x-ray — and the nearest one is a plane ride away? In this special crossover episode with SIIMCast, Jason and Mohannad sit down with Greg Toffner, a PhD researcher training local Inuit community members across 25 Nunavut communities to become Basic Radiological Technicians (BRTs). Greg breaks down the concept of task shifting, how it’s reshaping radiology IT access in one of the most geographically isolated regions on earth, and what six years of hands-on curriculum design, mobile x-ray equipment, and government policy work have taught him about building a sustainable imaging informatics workforce from the ground up.You’ll hear real numbers on health disparities in the North, how a 60-procedure competency framework turns a health center janitor into a trusted member of the care team, and where AI might eventually fit into enterprise imaging and radiology workflows in low-resource settings. It’s a conversation about access, dignity, and what imaging informatics looks like when PACS, DICOM, and specialist radiologists aren’t around the corner.If you’re building your own credential in imaging informatics, check out the CIIP Foundations Program at nagelsconsulting.com, and keep an eye out for our upcoming DICOM training program featuring hands-on, live imaging learning labs: Learn more at nagelsconsulting.comKey Topics Covered• What “task shifting” means and how it’s applied to solve radiology workforce shortages in remote Arctic communities• Life inside Nunavut’s 25 flying communities — the geography, climate, and health disparities driving the program (life expectancy, respiratory disease, and smoking rates)• How Greg’s team built a simplified radiology curriculum for learners with no formal healthcare background• The three-phase, hands-on training model — from image critique to a 60-procedure competency audit• Why government policy recognition, not just curriculum quality, is what makes a task-shifted role sustainable• Early findings from Greg’s PhD research: pride, community trust, and the ripple effects on local health outcomes• Where the program is headed next — expanded scope into lab work and ECGs, and where AI might fit inhealthcare IT, medical imaging, radiology technology, healthcare interoperability, PACS administration, DICOM standard, HL7 integration, radiology informatics, imaging workflows, vendor neutral archive, enterprise imaging, radiology AI, medical imaging data, PACS migration, rural health access

What happens when a 25-year medical imaging veteran retires, then comes back with an AI coding assistant and rebuilds his entire mental model of how PACS gets built? Chris Hafey, founder of Merkalis and longtime fixture in the imaging informatics community, joins Imaging Informatics Unplugged to talk about coding with Claude and other LLMs, why he believes solo founders can now out-build 50-person dev teams, and what that means for PACS administration, Radiology IT, and Enterprise Imaging going forward. We dig into the real story behind his Monday-morning community meetups, the skill-degradation risk facing junior engineers as AI takes over more of the coding, and the Merkle-tree-based architecture Chris built to solve DICOM synchronization — a problem that's plagued Vendor Neutral Archives and Imaging Informatics teams trying to keep on-prem and cloud PACS in sync. Chris also explains why he thinks some legacy PACS vendors would be better off rebuilding from scratch than fighting decades of technical debt, and what AI in Radiology and HL7-connected systems need to actually trust the data they're syncing. If you work anywhere near PACS, DICOM, or healthcare IT and want an unfiltered, practical conversation instead of a vendor pitch, this one's for you.Jason also shares an update on the CIIP Foundations Program for imaging informatics professionals working toward their credential, plus a first look at the new DICOM training program featuring hands-on live imaging learning labs. Learn more at nagelsconsulting.com.Key Topics Covered• How Chris's Monday-morning imaging community meetup started during COVID, went on hiatus, and came back unfiltered• Going from hand-coding for a month to building a full DICOM web server with OHIF integration in about two hours of real thinking time using Claude• Why Chris believes a small domain-expert team using LLMs can now do what used to require a 50-person dev team and tens of millions in funding• The skill-degradation risk for junior and senior engineers as LLMs take over more of the actual coding• Why some legacy PACS vendors might be better off rebuilding from scratch than maintaining decades of technical debt• Applying a Git-like delta model to DICOM so PACS, VNAs, and AI pipelines can stay in sync without re-sending whole studiesTags• healthcare IT• medical imaging• radiology technology• healthcare interoperability• PACS administration• DICOM standard• HL7 integration• radiology informatics• imaging workflows• vendor neutral archive• enterprise imaging• radiology AI• AI coding assistants• PACS migration• healthcare data management

If you've ever fought through a PACS migration, wrestled with malformed DICOM data, or wondered why two systems that supposedly speak the same standard can't talk to each other — this one's for you.

Jason sits down with Larry Sitka, founder of Acuo Technologies and one of the earliest architects of the Vendor Neutral Archive (VNA) concept, to unpack three decades of building enterprise imaging infrastructure from the ground up.

Larry got his start writing network drivers at Bell Labs, helped shape DICOM 3.0 at 3M, and then built Acuo — a company he grew from a sketch on a napkin in 1997 into an enterprise imaging powerhouse acquired twice over.In this episode, Larry and Jason dig into why DICOM interoperability is still a mess, how AI false positives are driving radiologist frustration, and why the next evolution of enterprise imaging isn't about storing data — it's about perceiving it. They also tackle the knowledge gap that's forming as experienced imaging IT professionals retire, what AI governance for radiology actually looks like, and why Larry thinks the industry has been building things for the wrong user all along.

Whether you're a PACS administrator, imaging informatics professional, or just someone who cares about getting radiology AI right, this conversation will give you a lot to chew on.If you're looking to build a stronger foundation in imaging informatics or sharpen your DICOM knowledge, check out the CIIP Foundations Program and the upcoming DICOM training with hands-on live imaging learning labs at nagelsconsulting.com.

Learn more at nagelsconsulting.comKey Topics Covered

• The origin story of Acuo and why the VNA concept emerged in 1997 — before anyone had a name for it
• Why DICOM interoperability remains broken and what a real conformance testing standard would look like
• The shift from data persistence to data perception — and how AI changes what we actually need from a VNA
• How AI false positives are burning out radiologists and what multi-algorithm inference engines could do instead
• The knowledge gap in imaging IT: what gets lost when experienced DICOM engineers retire
• AI governance for enterprise imaging — why recalibrating AI models is the next big challenge in PACS/VNA
• The future of enterprise imaging: running a million AI inferences a night at population scale

If your digital pathology deployment has a brand-new whole-slide scanner, a solid IMS, and petabytes of storage — but your pathologists are reading on consumer monitors — you may have a serious problem you haven't budgeted for yet.In this episode of Imaging Informatics Unplugged, Jason sits down with Tom Kimpe, VP of Technology & Innovation for Healthcare at Barco, for a deep-dive webinar from Canada Health Infoway on the critical role the display plays in the digital pathology imaging chain. Tom unpacks why color gamut matters more in pathology than almost any other imaging domain, how color variability sneaks in at every step of the workflow — from tissue staining through scanner to viewer to display — and what the peer-reviewed science actually says about diagnostic accuracy and reading efficiency when pathologists use medical-grade versus consumer monitors.Spoiler: a 6–8% reduction in reading time. 100% diagnostic concordance on medical displays versus measurable drop-off on consumer hardware. Missed concurrent diseases. These aren't marketing claims — they're published, peer-reviewed findings.Whether you're a PACS admin, imaging informatics specialist, or radiology IT leader expanding into enterprise imaging and digital pathology, this is the kind of workflow and display standardization knowledge that will save your organization from an expensive mid-project scramble. If you're working toward your CIIP credential, the CIIP Foundations Program at nagelsconsulting.com is exactly where this conversation fits into the bigger picture — and keep an eye out for the upcoming DICOM training program with hands-on live imaging learning labs that will make concepts like ICC profiles and DICOM WG-26 click in a whole new way.Learn more at nagelsconsulting.comKey Topics Covered

• Why pathology tissue has a broader colour gamut than sRGB displays can reproduce — and what that means clinically
• How colour variability is introduced at every link in the digital pathology chain: lab prep, scanner, viewer, and display
• The difference between consumer, professional, and medical-grade displays — and why it matters for pathology specifically
• ICC colour profiles: what they are, how they work, and why adoption is accelerating toward an industry standard
• Peer-reviewed evidence showing measurable impact of display quality on diagnostic accuracy, concordance, and reading efficiency
• The DICOM WG-26 ecosystem: how file format standardization is finally catching up to scanner adoption
• Where display procurement falls through the cracks in digital pathology deployments — and how to fix it

If you have ever watched a radiology AI demo hit 98% accuracy in testing and then wonder why nobody is actually using it in the clinic, this episode is for you. Hit subscribe so you never miss a conversation like this one.Jason sits down with Satvik Tripathi, incoming Medical Physics and Imaging Informatics PhD student at the University of Pennsylvania, AI scientist for RAD-AID International, and one of the sharpest voices in the field on the gap between research performance and real clinical value. Satvik has been working at the intersection of AI and radiology since 2019 and brings a perspective that cuts through the noise.They get into the hard questions: why multiple-choice benchmarks are a terrible way to evaluate medical LLMs, what data leakage is quietly doing to published performance numbers, and why a fine-tuned model is not always the winner in a clinical context. Satvik also breaks down what it actually takes to build a benchmark that means something, and shares early findings from his team's head-to-head testing of over 20 models on an internally annotated clinical dataset.The conversation also digs into agentic AI in imaging informatics, global health deployments through RAD-AID in Botswana and India, AI-assisted oncology workflows, and why running smaller open-source models locally might be smarter than everyone thinks. Plus, Satvik makes a case that prompt engineering is not a productivity shortcut but a legitimate scientific method.Whether you are a PACS administrator, imaging analyst, radiology IT professional, or just someone trying to figure out which AI tools are actually worth your time, this is the kind of conversation that helps you cut through the hype and think more clearly about what is coming.If this episode is useful to you, please subscribe, leave a review, and share it with a colleague in the imaging informatics community. It makes a real difference. And if you are working toward your CIIP credential or want to go deeper on the foundations of this field, check out the CIIP Foundations Program and the upcoming DICOM training with hands-on live imaging learning labs at nagelsconsulting.com.Learn more at nagelsconsulting.comKey Topics Covered

• Why AI model performance metrics often fail to predict real-world clinical impact, and the two questions every AI deployment team should be asking before going live
• The flaws in how medical LLMs are benchmarked today, including multiple-choice test limitations, data leakage, and the gap between controlled evaluations and actual clinical usefulness
• How Satvik's team built an internal annotated dataset and tested more than 20 models head-to-head, with results that challenge conventional assumptions about fine-tuned models
• The promise and current limitations of agentic AI in radiology, including what true agentic systems require versus what vendors are actually shipping• Using AI to democratize global healthcare through RAD-AID's work in Botswana and India, including Google-funded foundation model deployments and lessons that translate back to Western healthcare systems
• Why prompt engineering is a scientific method, not just a productivity trick, and how structured prompting can reduce hallucinations and improve reproducibility in clinical AI applications
• The practical case for smaller, on-premises open-source models over large cloud-based generalist models, including cost, privacy, sustainability, and compliance considerations