The elimination of low-frequency noise of breath and motion artifacts is one of the most difficult challenges of preprocessing rheographic signal. The data filtering is the conventional way to separate useful signal from noise and interferences. Conventionally, linear filtering is used to easy design and implementation. However, in some cases such techniques are difficult, if possible, to apply, since the data frequency range is overlapped with one of interferences. Specifically, it happens in aortic rheography, where some breathing process and pulmonary blood flow contributions are unavoidable. We suggest an alternative approach for breathing interference reduction, based on adaptive reconstruction of baseline deviation. Specifically, the computational scheme based on multiple calculation of Akima splines is suggested, implemented using C# language and validated using surrogate data. The applications of proposed technique to the real data processing deliver the better quality of aortic valve opening detection.