Slic-seg: Slice-by-slice segmentation propagation of the placenta in fetal MRI using one-plane scribbles and online learning

TY - CHAP

T1 - Slic-seg

T2 - 18th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2015

AU - Wang, Guotai

AU - Zuluaga, Maria A.

AU - Pratt, Rosalind

AU - Aertsen, Michael

AU - David, Anna L.

AU - Deprest, Jan

AU - Vercauteren, Tom

AU - Ourselin, Sebastien

PY - 2015/11/18

Y1 - 2015/11/18

N2 - Segmentation of the placenta from fetal MRI is critical for planning of fetal surgical procedures. Unfortunately, it is made difficult by poor image quality due to sparse acquisition, inter-slice motion, and the widely varying position and orientation of the placenta between pregnant women. We propose a minimally interactive online learning-based method named Slic-Seg to obtain accurate placenta segmentations from MRI. An online random forest is first trained on data coming from scribbles provided by the user in one single selected start slice. This then forms the basis for a slice-by-slice framework that segments subsequent slices before incorporating them into the training set on the fly. The proposed method was compared with its offline counterpart that is with no retraining, and with two other widely used interactive methods. Experiments show that our method 1) has a high performance in the start slice even in cases where sparse scribbles provided by the user lead to poor results with the competitive approaches, 2) has a robust segmentation in subsequent slices, and 3) results in less variability between users.

AB - Segmentation of the placenta from fetal MRI is critical for planning of fetal surgical procedures. Unfortunately, it is made difficult by poor image quality due to sparse acquisition, inter-slice motion, and the widely varying position and orientation of the placenta between pregnant women. We propose a minimally interactive online learning-based method named Slic-Seg to obtain accurate placenta segmentations from MRI. An online random forest is first trained on data coming from scribbles provided by the user in one single selected start slice. This then forms the basis for a slice-by-slice framework that segments subsequent slices before incorporating them into the training set on the fly. The proposed method was compared with its offline counterpart that is with no retraining, and with two other widely used interactive methods. Experiments show that our method 1) has a high performance in the start slice even in cases where sparse scribbles provided by the user lead to poor results with the competitive approaches, 2) has a robust segmentation in subsequent slices, and 3) results in less variability between users.

UR - https://www.scopus.com/pages/publications/84951834714

U2 - 10.1007/978-3-319-24574-4_4

DO - 10.1007/978-3-319-24574-4_4

M3 - Conference paper

AN - SCOPUS:84951834714

SN - 9783319245737

VL - 9351

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 29

EP - 37

BT - Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015 - 18th International Conference, Proceedings

PB - Springer Verlag

Y2 - 5 October 2015 through 9 October 2015

ER -