Updated: Oct 8, 2020
Gerard Sanroma uses Deep Learning to understand brains.
He’s new to working in an applied environment.
“It’s not a purely technical role of image processing, but from a biology perspective: knowing about the problem, not just the tools.”
“It’s important to formulate the relevant research questions in a way that makes sense, key for an applied researcher.”
You need to speak the same language for a question to be answerable. His job is “to understand the determinants of human brain ageing using image analysis and segmentation efforts.”
“Around forty, a lot starts to go downhill.”
It’s complex, he says, but mentions lifestyle: physical activity, diet and smoking. Not really news, is it? “I want to know exactly how these affect our biology. Getting a precise quantification.”
“There’s still significant variation outside the normative value. I’m curious about what determines this variation.”
“When you use complex models, like CNN, it’s more difficult to interpret. IMO, the true power of modelling is mapping the output from the input, not just getting the trend but leaving as little variance unexplained.”
Looking at brains all day: does it make you feel strange as a human? “Some people like them much more than I do!”