A novel pipeline for adrenal tumour segmentation.

摘要: Abstract Background and objective Adrenal tumours occur on adrenal glands surrounded by organs osteoid. These can be categorized as either functional, non-functional, malign, or benign. Depending their appearance in the abdomen, arise from one gland (unilateral) both (bilateral) connect with other organs, including liver, spleen, pancreas, etc. This connection phenomenon constitutes most important handicap against tumour segmentation. Size change, variety of shape, diverse location, low contrast (similar grey values between various tissues) are disadvantages compounding segmentation difficulty. Few studies have considered segmentation, no significant improvement has been achieved for unilateral, bilateral, adherent, noncohesive There is also recognised pipeline method different size, location information. Methods study proposes an (ATUS) designed to eliminate above ATUS incorporates a number image methods, limited adaptive histogram equalization, split merge based quadtree decomposition, mean shift large level eliminator, region growing. Results Performance assessment was realised 32 arterial portal phase computed tomography images using six metrics: dice, jaccard, sensitivity, specificity, accuracy, structural similarity index. remarkable performance, not affected discussed handicaps, particularly adherence success rates 83.06%, 71.44%, 86.44%, 99.66%, 99.43%, 98.51% metrics, respectively, sufficient uptake. Conclusions The proposed system realises detailed avoids known preventing accurate