Radiology is a gladiator sport
Dr Das[1] is one of the best radiologists I’ve ever worked with. Experienced. Meticulous. Trusted by peers. But his Apple Watch tells a different story. His heart rate spikes at the exact moment he hits ‘submit’ on a report. Not because he’s rushed, but because he’s unsure.
Not unsure of his skill, mind you, but whether he’s caught everything, whether the ambiguous shadow is something or nothing, whether a tiny, secondary detail could become a headline in hindsight. He reviews it again. Sometimes he changes his mind. Sometimes he doesn’t. But the stress remains.
Radiology, he once told me, isn’t just a discipline. It’s a gladiator sport. You step into the arena every day, knowing you’ll be judged on every call you make. You’re expected to be fast, sharp, and unfailing under conditions that are anything but.
The arena is unforgiving
Unlike surgeons, radiologists rarely witness the direct impact of their decisions. Instead, they work behind the scenes, operating on pattern recognition, subtle cues, and probabilities. They don’t see blood. They see shadows. Every day, they separate the insignificant from the critical, the incidental from the life-threatening.
But here’s the catch. One diagnostic miss can trigger a chain reaction: delayed treatment, unnecessary testing, incorrect medications, avoidable complications. It adds cost. It erodes trust. It breaks momentum in patient care. A patient returns with an advanced diagnosis that should have been caught. A hospital receives a letter from Legal. A family asks why no one saw it coming.
Accuracy in diagnosis, then, isn’t just a clinical metric, but something much more. It’s what protects patients from harm, hospitals from liability, and radiologists from burnout and blame. Radiologists are showing up to fight with intelligence and integrity, just like a gladiator would. But even Maximus[2] needed help in the arena.
Brilliant minds, real limits
Every radiologist knows the feeling. You’re ten cases deep into a shift, four hours from your last proper break, and still trying to focus with the same intensity you had at 9 a.m. But no matter how sharp the eye or how deep the intuition, there’s only so much one person can catch in a high-volume, high-pressure environment. Patterns begin to overlap. Cognitive shortcuts creep in. Even the best gladiators eventually lose the fight, not because they weren’t strong, but because they were left to fight alone.
We’ve optimized the human side as much as we can. We have rigorous training, structured reporting, regular auditing, standardized protocols and quality checking. But they’re all forms of retrospective reassurance: basically, hindsight. And they don’t prevent fatigue, bias, or variations from showing up mid-report. The need of the hour is support that’s fast, quiet, and embedded in the moment of decision-making in real time, when it still counts.
Not armor. Backup.
Radiologists aren’t asking for a shield. They’re asking for reinforcements: something that shows up right there in the diagnostic process, reading with them. Most aren’t hoping for some mythical machine to replace them. They’re waiting for a system that understands just how hard this job already is and meets them halfway.
Clinicians need support: the kind that doesn’t demand extra clicks or extra caution. Instead, it makes every decision more confident, more consistent, less alone. Because when you’re 37 reports deep into a 60-report day, and the image in front of you could go either way, what you need is a trained, tireless second reader whispering, “You’ve got this. And if you don’t, I’ve got you.”
AI that works at the right time
In high-stakes arenas like radiology, timing and precision are everything. Our AI autonomously handles about 15 to 17 percent of routine scans, mostly normal CXRs and extremities, thanks to a continuously improving generalised vision model trained on massive, real-world data streams from over 2,000 teleradiology sites. This frees radiologists to focus on complex cases.
The real power comes from partnership. Cases flagged as ambiguous or unusual are quickly escalated to radiologists who review and finalise reports. Their input feeds back rapidly, creating a continuous learning loop where human insight and machine intelligence refine each other. The result? Faster turnaround times, greater accuracy, and reduced fatigue. The AI adapts with every cycle, evolving in sync with clinical demands, not just for automation, but for decision support that earns its place in the room.
Back when AI was real only in research labs, we called it Clinical Decision Support. Now, we call it AI. Call it what you want. Accuracy is key. And within the team responsible for achieving accuracy, AI is now an integral member. It helps radiologists catch what truly matters with speed and precision. It is what patients deserve, what hospitals expect, and what radiologists need to fight the good fight every day.
Call it CDS, call it AI, or like Chris Martin, call it magic.[3] What really matters are outcomes for patients. AI just happens to be the best tool available to deliver it live and at scale.
Because in this arena, labels do not matter. Impact does.
References:
[1] Name changed
[2] Character from the film Gladiator (2000) inspired by real-life gladiators
[3] Referring to ‘Magic’, song by Coldplay, and AI-driven technological innovations used in concerts