The Cosmic Microwave Background is one of the most important sources of information about the Early Universe, since it is composed by photons emitted when the Universe was only 377 000 years old. Observing it, however, entails several difficulties. One of them is the presence of unresolved point sources in the foreground. In this work, we implement a recent algorithm that takes advantage of needlet filtering and multiple testing Statistics. We create a flexible and complete programme and apply it to simulations and Planck anisotropy maps. In the simulations, the algorithm is able to recover essentially all point sources with an intensity above 3 to 4 with very low percentage of false detections. On some of the Planck maps, the algorithm detects a population of maxima incompatible with the assumed gaussianity of the Cosmic Microwave Background.