Roger is this you ? Old fellow from Darmstadt and Birmingham ?

Roger, if i understand your circuit right then you do not have a kind of opto isolation from the high side to the mcu side.

you have inserted the bc547 to get a strong pulldown output that has a pullup on C againt vcc. C controls the input of the high side switch. BTS441 (we use BSP450).
your additional 100nF is connected from B to gnd if i understand that right.
so this cap (i tis only cap for signal noise lower than 5MHz, above it is pure inductive) does nothing more than just a simple lowpass filter (c of 100nF and the internal pullup resistor of your mcu port what is about 10k). if i am right so far then you do not have anything done against emi. these emi pulses are so strong that they can even "jump" over an opto isolator, the leads of your cap connected to the transistor act as a inductor in your case and the emi does not see the cap, as the cap itself is far away from the emiflow.

emi-noise ......*...... mcuport
!
CAP
!
GND

in this star configuration the emi noise goes directly to the mcuport, as it does not see the cap because the lead from the point where you have connected it to you basis of your transistor and the cap is inductive for emi.

what you need in such cases is

emi noise .... CAP ... mcu port
!
GND

also try some ferrit over the signals.
your ignotioncoil creates hughe spikes up to several 100MHz.
it could (and i guess it will) be that the noise also comes across the gnd.

if you give me more information about your design, i will try to help you. i have solved this a lot of this.

Michael