Tamás Haidegger

Earlier today, I had the honor of delivering the keynote lecture at the Hungarian Congress of Radiology in Szeged, where I spoke about a topic that’s not just transforming our profession but redefining it: the Future of AI in Radiology. We are in the midst of a data explosion, one that’s changing the face of healthcare. In 2010, medical data doubled every 3.5 years. In 2020: Every 73 days. Estimates suggest this rate will reach four weeks by 2026. A single radiologist may now review up to 50,000 images in a 12-hour shift compared to just 500 images 15 years ago. Over 6,000 medical journals publish more than a million articles each year. While 90% of healthcare data comes from imaging, an astonishing 97% of it is never retrieved. It’s archived and forgotten. This is not just an issue of volume, this is an issue of value. And this is where AI comes in. #AI has the power to close the widening gap between data creation and data interpretation augmenting radiologists, not replacing them. I shared examples of how AI is already embedded across GE HealthCare‘s portfolio: Chest X-rays flagged for pneumothorax at the point of care, saving precious time in critical cases. Deep-learning reconstruction (AirReconDL) improving MRI image sharpness while reducing scan time. Pocket-sized ultrasound (Vscan Air) enabling doctors even in remote areas to scan, diagnose, and decide. AI-driven scheduling tools that reduce costly no-shows by predicting them before they happen. Enterprise-level command centers that orchestrate hospital operations in real time, using predictive analytics. But technology alone won’t drive transformation. Strategy will. Digital change in healthcare is as much a cultural shift as it is a technical one. Success requires clinicians, technicians, IT, and leadership to come together - and above all, it must make life easier, not harder, for caregivers. AI should be invisible. Seamless. A silent partner in the background. Like I said during my talk: “It’s like an invisible friend helping you run faster.” We must stop asking whether AI will replace radiologists and start asking how we can work with it. Like the autopilot in aviation, AI is a co-pilot, not the captain. My firm belief? AI won’t replace radiologists. But radiologists who use AI may replace those who don’t. Let’s embrace this hybrid future: Physician + AI, not with fear, but with confidence. Because when used right, AI doesn’t dehumanize care. It re-humanizes it freeing up time to focus on what matters most: the human connection. At GE HealthCare, we’re all in on AI. Not because it’s the trend, but because it’s the tool that will help create a world where healthcare has no limits. Thank you, György Erõss and Jutas Foldvari for organizing a great couple of days in Hungary. Simon, you have an awesome team here in Hungary. I‘ll be back for sure. #gehealthcare #AI #radiology #digitalhealth

92

6 hozzászólás

📖 Editor's Highlight New study published in European Radiology Experimental, an official journal of the European Society of Radiology   📢 “Dark-field chest radiography signal characteristics in inspiration and expiration in healthy and emphysematous subjects” By: Theresa Urban, PD Dr. Dr. Florian Gassert, Manuela Frank, Rafael Schick, Henriette Bast, Jannis Bodden, Dr. Alexander W. Marka, Dr. med. Lisa Steinhelfer, Manuel Steinhardt, Andreas Sauter, Alexander A. Fingerle, Gregor Zimmermann, Priv.-Doz. Dr. med., Thomas Koehler, Marcus R. Makowski, Prof. Dr. Daniela Pfeiffer & Franz Pfeiffer From: Technical University of Munich, Garching, Germany and Philips Innovative Technologies, Hamburg, Germany. 🫁 Dark-field chest radiography is sensitive to the lung alveolar structure. The change of dark-field signal between inspiration and expiration was assessed in 88 subjects (mean age of 64, 55 males). Dark-field signal appeared significantly higher in expiration compared to inspiration. The dark-field coefficient ϵ was lower in inspiration compared to expiration all participants. The dark-field coefficient in inspiration and expiration, as well as their ratio, was lower for at least moderate emphysema when compared to the control group. 🔗Link to the article: https://rdcu.be/ehXBB #EuropeanRadiologyExperimental 

44

🇨🇿 AI in healthcare is moving from idea to execution In the new Government Program Statement, the cabinet pledges to create conditions for the implementation of validated AI tools to enter routine care. Across CEE, this shift is already visible. It’s no longer pilot theatre: AI solutions run every day, from small polyclinics to the biggest university hospitals, supporting clinicians and shortening the path from finding to action.

23

2 hozzászólás

📢 Exciting news from #Zagreb #Croatia 🇭🇷 at the EFOMP School for Interventional Radiological Practices! Today, Virginia Tsapaki (International Atomic Energy Agency (IAEA)) presented the important topic “Dose Management Systems in Interventional Radiology”, along with the brand-new IAEA publication “Dose Management Systems: From Setting Up to Quality Assurance”, to which European Federation Of Organizations For Medical Physics - EFOMP is proud to have contributed. Special thanks to Roberto Sanchez, Scientific Chair of the School in #Zagreb and all European Medical Physics Experts who contributed and was among the experts involved in drafting and reviewing the publication 👉 🔗 https://lnkd.in/du2BjK-g  #medicalphysics #ESMPE #DoseManagementSystems

80

Doctors 🩺 could greatly benefit from using specialized AI models 🤖 to extract and reason about the information included in medical case summaries 📃, when they need to recommend diagnostic tools 🩻 or treatment lines 💊 for their patients. In our recent article, "A Large-Scale Benchmark for Evaluating Large Language Models on Medical Question Answering in Romanian", accepted in the prestigious npj Digital Medicine journal (impact factor = 17.0), we introduce MedQARo, an evaluation benchmark that enables the development and evaluation of AI models to answer clinical questions in Romanian. MedQARo is the result of the concerted efforts of both AI (Ana R., Radu Ionescu) and medical (Alexandra-Valentina Anghel, Ionuț-Lucian Antone-Iordache, Simona Coniac, Andreea Iuliana Ionescu) teams. 📖 Check out our preprint at: https://lnkd.in/d3cMHJfd #medicalAI #LLM #benchmark #medicalQA #QuestionAnswering

67

2 hozzászólás

The new #DICOM standard has been released for 2025. I am posting this because DICOM is the hugely successful result of years of international cooperation and such cooperation should be celebrated. This release also corresponds to the anniversary of my small contribution to the standard 20 years ago: the Spatial Registration Module. For the DICOM nerds, this is C.20.2 on page 1611 (yes this is a huge standard) of DICOM PS3.3 2025b Information Object Definitions. What is spatial registration? It is the ability for a radiologist to view different medical images within a common reference frame, for example two images of the same patient before and after treatment to see if a tumour has shrunk. Having the mathematics to describe the location of the same anatomy in the different images allows the tumours to be super-imposed on each other for ease of comparison. In this way you can think of registration like “alignment” where two or more images are rotated, translated and even stretched until they overlap neatly with each other. Below is an example of MRI and ultrasound of the spine registered together to help guide surgery using the advantages of both images [Behnami  et al. Model-based registration of preprocedure MR and intraprocedure US of the lumbar spine. Int J CARS 12, 2017]. Who did this? Back around 2001-2005, working group 17 was formed with reps from ALI Technologies (now McKesson), Philips, ADAC Laboratories, Vital Images, Kodak, I3 Dimensions, Toshiba, Agilent and others. Spanning multiple countries, industry verticals, expertise and languages, we agreed on a common standard to represent registration.  My contribution was the mathematics: using a single matrix approach inspired from my time studying robotics. This approach can be used for different medical images and for different types of registrations: rigid, scaled, and affine. Why is this important? From the DICOM website “DICOM® — Digital Imaging and Communications in Medicine — defines the formats for medical images that can be exchanged with the data and quality necessary for clinical use. DICOM® is implemented in almost every radiology, cardiology imaging, and radiotherapy device (X-ray, CT, MRI, ultrasound, etc.). With hundreds of thousands of medical imaging devices in use, DICOM® is one of the most widely deployed healthcare messaging Standards in the world. There are literally billions of DICOM® images currently in use for clinical care. From the emergency department, to cardiac stress testing, to breast cancer detection, DICOM® is the standard that makes medical imaging work — for doctors and for patients.” How is this relevant today? We should take pride that our diverse group of people from different/competing companies managed to find a common standard that has helped patient care for more than 20 years. In general, DICOM has been a huge success and a benefited healthcare around the world. This type of collaboration around a common goal should be celebrated.

352

23 hozzászólás

The European Congress of Radiology has opened in Vienna! In this report from Day 1 of ECR 2025, I discuss the show's message of sustainability in radiology and focus on AI for medical imaging. #ECR2025 #ECR25 #radiology

46

2 hozzászólás

#MicroPort is thrilled to announce the first robotic-assisted procedures with our MicroPort MedBot #Toumai Laparoscopic Surgical Robot in Bulgaria! 🏥 Performed at University Hospital "Alexandrovska" by leading specialists: ✅ Prof. Marincho Georgiev (Urology) ✅ Prof. Atanas Yonkov (General and Hepato-Pancreatic Surgery) Toumai® features: 🔹 Force rendering technology for enhanced precision 🔹 Multi-layer safety features ensuring stability 🔹 Advanced imaging systems for superior visualization This achievement marks a significant milestone for #RoboticSurgery in #Bulgaria, and reinforces our commitment to expanding access to state-of-the-art surgical solutions across Europe for #BetterAndLongerLives. 🔗 Learn more about our surgical robots: https://lnkd.in/gPPzCY34 #MedicalInnovation #HealthcareTechnology #SurgicalRobotics

111

🎉 PARROT has been published in the European Journal of Radiology Artificial Intelligence! 🦜📚 What does PARROT offer? 📝 2,600+ realistic radiology reports 🌐 13 languages, 21 countries 🩻 Multiple modalities: CT, MRI, X-ray, Ultrasound 🧠 Written by radiologists, not generated text 🔓 Fully open, privacy-safe, and reusable 🤖 A unique benchmark for LLMs, and report generation in radiology. 🌱 Perfect for testing generalisability beyond English-only datasets. ⚖️ Open science without privacy barriers. Link to the paper: https://lnkd.in/eJUQMcxp Github link: https://lnkd.in/eN2yFgBf 👏 Big congratulations to Bastien Le Guellec and Keno Bressem for leading this huge effort 🚀✨ Outstanding leadership behind a truly ambitious project💪 👏 Congrats to the entire author team for this massive collaborative effort - proud to be a member of the team!

155

5 hozzászólás

Can AI write abstracts and are reviewers able to detect them? The answer is no! However, AI can detect abstratcs written by AI - thus, we need such tolls in current time to assure that real abstratcs are presented at our meeting - read at https://lnkd.in/erCuCHdE

236

7 hozzászólás

Taehee Lee et al explored the availability of #AI models from development studies published in leading radiology journals in 2022, finding that various steps are needed to improve transparency and relicability in radiology AI research #EuropeanRadiology https://lnkd.in/dNuTKwfa

16

1 hozzászólás

Check out our new journal paper published on: "The gaps and challenges in digital health technology use as perceived by patients: A Scoping Review and Narrative Meta-Synthesis" Authors: Georgia Livieri, Professor Eleni Mangina, Prof. Evangelos Protopapadakis and Dr Andrie Panayiotou, (2025). Frontiers Digital Health Volume 7. 🔗 Read it here: DOI: 10.3389/fdgth.2025.1474956 This scoping review explores the perceived gaps and challenges patients face when using digital health technologies. By synthesizing findings from multiple studies, we identified 25 gaps and 74 challenges, including generational differences in digital propensity, work process deficiencies, and ethical concerns such as trust and data ownership. Our findings underscore the need for ethically aligned technology development that prioritizes human rights within the digital healthcare ecosystem. #DigitalHealth #PatientExperience #HealthTechEthics #PatientCentricCare #HealthTechnology #EthicalTech #HumanRightsInHealthcare

30

🇭🇷 Promising updates from the RadioVal Consortium Meeting in Zagreb! We are integrating AI into the RadioVal clinical decision support system (powered by QP-Insights) with: ▪️ Automatic lesion segmentation ▪️ 3-year pCR and survival prediction ▪️ Continuous monitoring of model drift Quibim also leads the sustainability and exploitation strategy, which includes market analysis and assessment of the technology landscape to ensure long-term impact in #BreastCancer care. #AIinHealthcare #DigitalHealth #MedicalImaging

47

We are pleased to announce that registration is now open for the MultiFact Symposium: Advances in AI and MRI for Fetal Brain Development, which will take place on March 14th in Munich. This free event will bring together experts in AI and magnetic resonance imaging applied to fetal brain development, providing a platform to share knowledge and recent advances in the field. The MULTIFACT project is funded by ERA-Net NEURON, supporting research on fetal brain development through advanced imaging and AI techniques. You can find more information and register at the following link: https://lnkd.in/dYvapP8y  #MULTIFACT #fetalbrain #NeurodevelopmentalDisorders #fetalMR CIBM Centre d'Imagerie Biomedicale Department of Biomedical Sciences - UNIL TU Muenchen UNIVAM BCN-MedTech BCNatal Fetal Medicine Research Center Universitat Pompeu Fabra Institut d’Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS)

96

2 hozzászólás