I hope that wasn't a surprise!
I'm new to this... so yes! In a delightful / dreadful way it was a surprise to see radiation be stronger than the signal depending on the layout.
Oh dear! In reply#26 you seemed to be hinting at driving 100s of mA <<3ns edges to speed up the LEDs switching, I imagined multiple channels switching at exactly the same time only 1cm from the fast photo diode amps! But don't worry, I think everyone finds out the effects of fast edges the hard way.
In my simple LED switching experiments a 5mm UV seemed to have faster rise and fall times than most, but each edge ended in a horrible curve so I didn't try any further.
I don't know how you're turning the small photo diode currents back into a 400Mb/s logic level, but I know it's a lot harder than 'just use a 2ns photo diode'.

The new? ST60A2G0 looks like it could be a good solution.
I don't know much about high speed stuff at all.
