Performance of three algorithms for breathing sounds sources estimation

This work evaluates the discrete, linear solutions for the breathing sounds (BS) inverse problem under different simulated scenarios. Three different inverse algorithms are analysed and compared: overdetermined least-square, underdetermined minimum norm and weighted minimum norm (FOCUSS). The tests evaluate the performance of the algorithms in the estimation of single and multiple sources and the influence on the solution of the source depth; simulated scenarios include point and non-point BS sources. Of the three solutions tested, the FOCUSS algorithm seems to be the more appropriate to localize the BS generators in 3D space.