Multiple Sclerosis lesions detection by a hybrid Watershed-Clustering algorithm

Background Computer Aided Diagnosis (CAD) systems have been developing in the last years with aim of helping diagnosis and monitoring several diseases. We present a novel CAD system based on hybrid Watershed-Clustering algorithm for detection lesions Multiple Sclerosis. Methods Magnetic Resonance Imaging scans (FLAIR sequences without gadolinium) 20 patients affected by Sclerosis hyperintense were studied. The consisted following automated processing steps: images recording, segmentation Watershed algorithm, lesions, extraction both dynamic morphological features, classification Cluster Analysis. Results investigation was performed 316 suspect regions including 255 lesion 61 non-lesion cases. Receiver Operating Characteristic analysis revealed highly significant difference between non-lesions; diagnostic accuracy 87% (95% CI: 0.83–0.90), an appropriate cut-off 192.8; sensitivity 77% specificity 87%. Conclusions In conclusion, we developed using modified image which may discriminate MS from non-lesions. proposed method generates out-put that be support clinical evaluation.