AI in cancer care and research

AI is creeping into every aspect of life. The same is true for medicine and interventional radiology! With transformative technology, it can be hard to differentiate what is over-stated or over-hyped from what is practical or coming soon to your IR suite, even if you do nothing, buy nothing, or learn nothing. AI is here to stay and is already working behind the scenes in the shadows with image quality or IR toolkits. To fully apply AI to IR problems and workflows, we need to ethically embrace AI tools and AI nuts and bolts. IR and AI both uniquely mix data and computer science with image processing (segmentation), physical sciences (planned treatment zones and registration), and biological sciences (what cellular, inflammatory, and immunological pathways are we modulating?).

Hard problems need high tech solutions. Enter AI in IR.

IR and medicine relies upon too many human factors, which keeps us from being standardized, reproducible, and optimized. AI has the potential to change that. If you feel like we don’t yet know what we are doing, remember that at the root of science is the fact that we learn as we go. Enter deep learning and AI. You do not need to code or program to know how AI will be applied in IR. The AI landscape is immense and radiology is already the medical field most applying AI, by far. Why should IR be different?

How might IR be transformed by AI?

To better play a more impactful role in future medicine, we need to embrace AI, such that pharma, devices, and other disciplines fully appreciate the potential power of image guided local therapy when powered by AI tools. We are not there yet, but there is so much potential if we answer questions as a team. In the rapidly evolving landscape of minimally invasive, image-guided therapies, we often lack the ability to reproduce results or uniformly treat specific situations in a standardized fashion that enables higher level evidence, effective use of guidelines, and avoiding “reinventing the wheel” with every tumor board decision tree. Enter Large Language Model AI.

What will we discuss at the AI for IR session?

CIRSE 2024 will provide an immersive team-oriented exploration into the current status of this AI space, as well as views on what is yet to come in the future and key questions to answer in order to get us there. Help define what we become! You will learn the basic language of AI for IR. We will cover AI applications and their impact on IR. How is radiomics different? How can we apply AI to treatment planning and patient selection? Can AI help with follow up and decision making or guideline implementation?

Imagine an AI tumor board selects a specific treatment plan. An AI-IR biopsy defines pathology + transcriptomics + radiomics images that determine a personalized combination of therapies. AI suggests a few consults based upon medications and the electronic health record. The AI tools segment, register, and confirm adequacy or the AI-predefined treatment plan. AI tools pick which catheter and wire and fluoro angle, and when you reach a procedural endpoint. AI follow up models predict when to image, auto-detect secondary findings, and helps determine when to select what next therapy. AI auto-populates the report and wearable patient sensors detect post-procedural problems before they surface.

This session complements: SPHAIRE – Spotlight on AI and emerging technologies in IR

CIRSE 2024 provides a fertile forum to engage and network and to be a part of the multi-disciplinary excitement by fostering partnerships between academics, industry, and scientists to define and address IR challenges and problems with AI solutions. IR teams thrive by embracing novelty and innovation. A new age is upon us. All aboard!

Last, but not least, I would like to give a quick shout-out to the new CVIR editorial team in AI: CVIR: The new editors for AI