we presented two methods for tumour classification in mammograms. One system exploited the
use of neural networks using back-propagation and the second one was built employing association rule mining with
constraint form. The first method proved to be less sensitive to the database imbalance at a cost of high training
times. The second one, with a much more rapid training phase, obtained better results than reported in literature on a
well balance dataset. Both methods performed well which proves that association rules mining employed in classification process is worth further investigation.