On the mining of association rules in medical image data sets

This paper presents an effective association rule mining algorithm suitable for obtaining rules that describe patterns in large databases with long patterns. In particular, this algorithm is applicable to medical images, which are typical examples of such large and long data sets. Our algorithm improves upon the FP-Growth algorithm, known for its fast implementation , by applying heuristics for parallel processing, thereby obtaining significant speedup during the mining process. A comparison of existing algorithms and the partitioned FP-growth algorithm is also presented. Finally, we present a summary of the report and make suggestions for future research.