The radiology job of “reading scans” is actually two different jobs in rapid alternation: pattern recognition (pixels, asymmetry, gestalt) and translation (turning that visual certainty—or uncertainty—into clean prose with a defensible impression).
The vast majority of my work is diagnostic these days, but there have been times in my life when I wore lead, did a lot of procedures, and planned for a career as such. Interventions were much more physically demanding and certainly had moments of emotional intensity that I haven’t felt in years at this point in my career. But I always found diagnostic radiology more mentally exhausting, especially earlier in my career, when a greater fraction of cases required a greater amount of Type 1 thinking.
A lot of that end-of-day exhaustion isn’t from volume alone; it’s from the accumulated switching cost (attention residue) of bouncing between visual cognition and written language, case after case. If you want real efficiency (and fewer mistakes), you don’t just read faster—you reduce forced context switches so your brain can stay in one mode longer.
We all need to work on our own personal version of the iterative loop and efficiency/ergonomics.
This is one of the missing joys of working without a better process—whether that’s macros, pick lists, templates, or even better physical integration with your computer. A streamlined, efficient process minimizes interruptions, and that seamlessness allows you to move through the cases instead of fighting to get each one done.
2 Comments
How interesting that while thinking about the issue of improving the working conditions of a radiologist, I came to the same conclusion that switching between contexts, between a browser and a word processor, contributes greatly to fatigue at the end of the working day. In Ukraine, where I work as a radiologist, we do not have standardized, employer-provided dictation systems, so based on the available systems, namely dictation in Google Docs, LLM Gemini, the Mouse without Borders program, and an advanced text notebook written in Python (of course, with the help of the same Gemini), I was able to radically improve my working conditions. Having 2 jobs and physically 3 workplaces and the need to sometimes open an examination on the go, I made a modular architecture, where my laptop acts as a universal work module, on whose shoulders is placed the entire routine of dictation, speech recognition, processing the received data using Gem-Bot, which is configured according to the rules and features, checking grammar and typing errors (part of the work still consists of typing, sometimes it’s easier that way). All the setup and writing of the program took half a year, but now I have received a boost of about 70-150 percent in productivity and can either do more or do less work faster. I have become much less tired. Not to mention some minor improvements on the work PC using AHK scripts and small programs. Thank you for your posts, they are unique, I have spied several ideas for improvement in them.
Thank you for sharing your experience here!