A Fast and Robust Approach to Liver Nodule Detection in MR Images

In this paper, a novel and fast method to detect liver nodules automatically based on the original image, which avoids problems raised by liver segmentation was proposed. Algorithms were applied to determine the candidate regions which may contain nodules and the gray value of a nodule is typically considered to be a local extremum, but not necessarily a maximum, as it differs in signal from the surrounding parenchyma. Fuzzy C-means(FCM), which is an optimization algorithm to classify the image into C clusters based on the distance of the pixels to the cluster center in the feature domain, was used to segment the image. This is followed by a morphological smoothing, identification and analysis of candidate regions by computing properties which include area, centroid location, shape in a two-dimensional slice and connectivity in three dimensional space. Candidate regions whose properties are similar to that of nodules are extracted as the final result.