The complexity of signals and of algorithmic methods in medical signal processing tasks imposes considerable requirements on the computing hardware. The conflicting demands for computing power and portability can currently be satisfied only by massively parallel operation. In this contribution we present the CESAR architecture of a parallel processor array utilising the underlying Cellular Nonlinear Network principle. The array is embedded into a hardware-software system forming a highly configurable signal processing platform usable both in algorithm development and in real-world applications.