A combined fuzzy and level sets´ based approach for brain MRI image segmentation

The different tissues namely gray matter (GM) white matter (WM), and cerebrospinal fluid (CSF) are spread over the entire brain. It is difficult to demarcate them individually when a brain image is considered. The boundaries are not well defined. Modified fuzzy C means (MFCM) and level sets segmentation based methodology is proposed in this paper for automated brain MRI image segmentation into WM, GM and CSF. The initial segmentation is done by MFCM approach and the results thus obtained are input to the level set methodology. We have tested the methodology on 100 different brain MRI images. The results are compared by using individual MFCM and level set segmentation methods. We took the opinion of 10 expert radiologists to corroborate our results. The results are validated by radiologists as `Accurate´, `Satisfactory´, `Adequate´ and `Not acceptable´. The results obtained using only level set are `not acceptable´. Most of the results obtained using MFCM are `Adequate´. The results obtained using combined method are `Satisfactory´. Hence, the results obtained using combined MFCM and level sets based segmentation are considered better than using individual MFCM and level set segmentation methods. The manual intervention is avoided in the combined approach. The time required to segment using combined approach is also less compared to level set method. The segmentation using proposed methodology is helpful for radiologists in hospitals for brain MRI image analysis.