En route to an intraoperative imaging system for brain surgery, a thermographic camera associated with time-resolved data analysis is used for the identification of functional areas and the detection of brain tumours. In this contribution we propose a Cellular Nonlinear Network based image processing system supporting massively-parallel and real-time processing of thermographic video data. In the current setup the system is implemented on a Xilinx Zynq FPGA and is used for the preprocessing of the raw data. The development and implementation of a digital low-pass filter for the compensation of local motion artefacts will be demonstrated as a typical application example of the freely programmable system.