Conventional spinal cord stimulation (SCS) delivers a fixed-input of energy into the dorsal column. A new SCS approach exists in controlling stimulation dose by measuring the recruitment of fibers in the dorsal column and by using the amplitude of the evoked compound action potentials (eCAPs) to maintain stimulation within an individualized therapeutic range. As a reminder, eCAP represents a synchronized response generated by a group of electrically activated nerve fibers to stimuli.
A key challenge in the measurement of evoked compound action potentials (eCAP) is the signal amplitude being very low. Spinal cord potential can be in the range of ten micro volts and a high precision (about one microvolt) measurement technique will be required to extract the eCAP. Further, if the initial stimulation potential is in volts, then stimulus artifacts would likely impact the evoked response unless the stimuli and measurement electrodes are spatially separated to minimize this artifact contamination.
Despite physical separation of stimuli from evoked signal measurement points, electrical circuit design techniques have critical impact on the measurement accuracy. Ideally a very high dynamic range circuit is desired to detect a very low signal from a high-amplitude stimulus but this is not practical.
Fortunately, there are several techniques in ultra-low noise amplifier design.
One may prefer to use an operational amplifier (Op Amp) or an instrumentation amplifier based design for simplicity, but a transistor-based discrete amplifier can have much lower noise, with a similar power consumption. Whether Op Amp or transistor based, noise of an amplifier can be reduced by paralleling amplifiers. This would lower noise by the square root of the number of parallel stages. A transistor-based design helps further, especially if JFET parts are used, since this would increase the input impedance.
It should be emphasized here that the resistor value selection is critical in low noise amplifier design, since a resistor contributes to circuit noise by about 4nV/rt Hz (nano volt per root hertz) for each 1 k ohms. In addition to its own noise contribution, resistor values also affect noise due to input noise current of amplifiers.
Each noise value Including the contribution of amplifier input noise voltage (multiplied by amplifier gain), amplifier output noise voltage, total circuit noise will add up as 'root of sums of squares of each contributor' and these can easily add up to hundreds of micro volts rms (root mean square) when calculated for the operating bandwidth of the circuit.
A low noise circuit designer should also be aware of the Gaussian distribution of random noise with the probability of +/- 3 sigma (6 sigma distribution). This implies 99.7% of total noise (in either plus and minus amplitude direction). This peak-to-peak noise will reach 6 times the rms value and can significantly shadow the low level evoked nerve signals being measured.
Please contact me if you are interested in additional details.