Segmenting Leukomalacia using Textural Information and Mathematical Morphology

In this article we present a technique for segmenting the affected tissue visible as white flaring in the ultrasound brain images of neonates with Leukomalacia (White Matter Damage). The technique combines both textural information of the investigated tissue as well as mathematical morphology in order to detect and delineate the boundaries of the affected parts of the brain. The reproducibility of the proposed technique is evaluated and the experimental results are validated by comparing them to the manual delineations of 12 expert medical doctors from different institutions. Although it seems hard to reach a consensus on the correct segmentation of the flaring, because of the lack of a golden standard in the ultrasound images, we show that our method outperforms the existing techniques based on active contours in speed and is more accurate. Keywords— Medical Ultrasound, Leukomalacia, segmentation, mathematical morphology