TY - CHAP
T1 - 2D-Based Coarse-to-Fine Approaches for Small Target Segmentation in Abdominal CT Scans
AU - Zhou, Yuyin
AU - Yu, Qihang
AU - Wang, Yan
AU - Xie, Lingxi
AU - Shen, Wei
AU - Fishman, Elliot K.
AU - Yuille, Alan L.
PY - 2019
Y1 - 2019
N2 - Deep neural networks have been widely adopted for automatic organ segmentation from abdominal CT scans. However, the segmentation accuracy of small organs (e.g., pancreas) or neoplasms (e.g., pancreatic cyst) is sometimes below satisfaction, arguably because deep networks are easily disrupted by the complex and variable background regions which occupy a large fraction of the input volume. In this chapter, we propose two coarse-to-fine mechanisms which use prediction from the first (coarse) stage to shrink the input region for the second (fine) stage. More specifically, the two stages in the first method are trained individually in a step-wise manner, so that the entire input region and the region cropped according to the bounding box are treated separately. While the second method inserts a saliency transformation module between the two stages so that the segmentation probability map from the previous iteration can be repeatedly converted as spatial weights to the current iteration. In training, it allows joint optimization over the deep networks. In testing, it propagates multi-stage visual information throughout iterations to improve segmentation accuracy. Experiments are performed on several CT datasets, including NIH pancreas, JHMI multi-organ, and JHMI pancreatic cyst dataset. Our proposed approach gives strong results in terms of DSC.
AB - Deep neural networks have been widely adopted for automatic organ segmentation from abdominal CT scans. However, the segmentation accuracy of small organs (e.g., pancreas) or neoplasms (e.g., pancreatic cyst) is sometimes below satisfaction, arguably because deep networks are easily disrupted by the complex and variable background regions which occupy a large fraction of the input volume. In this chapter, we propose two coarse-to-fine mechanisms which use prediction from the first (coarse) stage to shrink the input region for the second (fine) stage. More specifically, the two stages in the first method are trained individually in a step-wise manner, so that the entire input region and the region cropped according to the bounding box are treated separately. While the second method inserts a saliency transformation module between the two stages so that the segmentation probability map from the previous iteration can be repeatedly converted as spatial weights to the current iteration. In training, it allows joint optimization over the deep networks. In testing, it propagates multi-stage visual information throughout iterations to improve segmentation accuracy. Experiments are performed on several CT datasets, including NIH pancreas, JHMI multi-organ, and JHMI pancreatic cyst dataset. Our proposed approach gives strong results in terms of DSC.
UR - http://www.scopus.com/inward/record.url?scp=85073205304&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85073205304&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-13969-8_3
DO - 10.1007/978-3-030-13969-8_3
M3 - Chapter
AN - SCOPUS:85073205304
T3 - Advances in Computer Vision and Pattern Recognition
SP - 43
EP - 67
BT - Advances in Computer Vision and Pattern Recognition
PB - Springer London
ER -