文摘
Computerized bone age assessment was widely studied for the physicians and radiologists to evaluate children with endocrinological disorders, growth retardation and treatment monitoring. Unfortunately, the morphological assessment of phalanges and epiphyseal/metaphyseal region (EMROIs) in hand radiogram with non-uniform intensity or full of the noise is intricacies and hardly to be segmented, and which was also time-consuming. So, a new segmentation technique for handy skeleton was proposed. The first step, the properties of position and brightness on phalanges and EMORIs are analyzed by a series of image preprocessing procedures. Next, 14 EMORIs of five phalanges were examined by the sum-variance (SV) scheme for the 100 boys and 100 girls. Last, four statistical indices are used to evaluate the segmentation. The experiment performance is shown the better result than the gradient vector flow (GVF) snake, and the approximation of adaptive two-means clustering. The study showed the proposed hierarchical algorithm for the extraction and segmentation of phalangeal bone regions of interests (PROIs) and EMROIs are effective, especially for the EMROIs segmentation by using the proposed algorithm.