The C-pathologist - courtesy of the Economist

http://www.economist.com/blogs/babbage/2011/11/diagnosing-cancer

They report in this week's Science Translational Medicine that they have written a program which can distinguish between grades of breast-cancer cell—and can do so in a way that provides a more accurate prognosis than a human pathologist can manage.

Click to expand...

I thought this was quite interesting. The last point is also quite good.

Pladio said:

http://www.economist.com/blogs/babbage/2011/11/diagnosing-cancer



I thought this was quite interesting. The last point is also quite good.

Click to expand...

Quite Interesting, indeed. But there are a lot of questions:

1. how well it performs under less controlled conditions
2. how good it is at evaluating different types of breast cancer.
3. performance at different stages in tumour development
4. the number of false positives (it seems that around 30% of early cancers detected during screening either disappear or are not cancers after all, resulting in many unnecessary operations and radiotherapy)

All this require lots of studies, eventually involving tens of thousands of patients followed over several years. But very interesting, and it's one of the areas where computer aided diagnosis (CAD) is most likely to succeed (IMO). Hopefully it will also succeed at detecting other types of cancers, which may require some serious modding - as cancer is a group of many diseases with very different characteristics.

Btw: I notice that one of thos commenting claimed that cancer is the number 1 killer in the world. It's not.

pibbur who thinks Stephen Fry is not a very good doctor.