What is the principle of the oscilloscope's isolation probe?

Recently, Ti sent an email introducing the design of its oscilloscope front-end probe, which seems to explain the principle of oscilloscope probes quite clearly.

As can be seen from the figure, the input signal passes through the pi-shaped attenuation network (in this design, the pi-shaped attenuation network can be converted into three gears of 1:1, 2:1, and 5:1). LMH5559 provides a reference potential to make the input signal within the input range of the differential amplifier (FDA). LMH5401 is an FDA with an input impedance of 50 [formula], which completes the conversion of single-ended input to dual-ended output. Then, through the digitally controlled adjustable amplitude amplifier (DVGA) LMH6401, the output signal is further adjusted in amplitude to meet the ADC input range and make full use of the ADC. Before entering the ADC, the high-frequency noise is first removed through a low-pass filter. TRF3765 is a fractional-N frequency-locked loop that provides a sampling clock for the ADC. After the ADC completes sampling, it is sent to the FPGA through the FMC interface (FPGA Mezzanine Card) for data processing.

The input of the isolation probe mentioned by the questioner is a pi-shaped network that can be connected to the double-ended input. After passing through the fully differential amplifier, the instrument amplifier is used to subtract the inputs at both ends and finally output a single-ended signal.