In the health field, several thousand images are generated every day in medical imaging establishments. The volume of information involved is still far from being fully controlled. On the other hand, the development of machine learning tools today opens the way to a new generation of image analysis in this context of "BigData". Moreover, our approach is part of this research dynamic. In order to test the robustness of our algorithm and its degree of adaptation to BigData, we tested, in a first phase of analysis, our algorithm on an image-database containing 320 mammograms. The precision obtained is estimated at 75% for a recall of 33%. In a second analysis phase, we performed the test on an image database containing 1000 medical images. The precision obtained is estimated at nearly 70% for a recall of 33%. Although the precision obtained in this first step is far from perfect, our processing algorithm remains promising and shows a good adaptation to the management of "Digdata