In doing so, we are starting to understand the cause and consequence of what Galvani was seeing in frog legs. #Bioelectricity in cancer cells!

In many ways our tools are essentially galvanoscopes! Albeit with far greater resolution, sensitivity, and applicability. A notable difference from a few centuries ago is my lab is developing #machinelearning and generativeAI to analyse data coming from microscopes and biosensors.

In parallel, we have a long-standing collaboration with Pantelis Georgiou to develop CMOS biochip sensors to measure bioelectric activity across tissues – setting the stage for diagnostics
https://buff.ly/4fQkohU

I can’t decide if in 2024 we have come a long way – or really not so far! Working with Amanda Foust and Mustafa Djamgoz at Imperial College London we have been imaging voltage using new microscopy in cancer.
https://buff.ly/4i9Rafp

Galvani also invented the #galvanoscope as a way to measure bioelectricity in frog legs.

#Sentinal4D and my lab at the Institute of Cancer Research have developed Al methods to analyze videos of #cancer cells, and measure how #cellshape changes over time.

We are using this technology to understand how drugs work, and find new cancer therapeutics.

link.springer.com/chapter/10.1...

You have to capture and analyze video footage.

The samee goes for cells. Studying fixed cells (whether an image, a transcriptome, or a proteome) will only get you so far. You won'ta really be able to understand how cancer develops, or how a drug works unless you look at cells over time.

But adding MYCN does make the output more interpretable (and likely trustable), by pathologists. That way decisions regarding treatment are not made based only what the Al 'sees'.

So... does adding more data always improve prediction accuracy? Not always. Integrating MYCN status doesn't necessarily enhance accuracy. In fact, image-based data alone yields comparable predictive power to multimodal data. And looking at the images is better than using MYCN status alone.