Open AccessDeep learning-based medical image segmentation with limited labels
Author(s) -
Weicheng Chi,
Lin Ma,
Junmei Wu,
Mingli Chen,
Weiguo Lu,
Xuejun Gu
Publication year - 2020
Publication title -
physics in medicine and biology/physics in medicine and biology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.312
H-Index - 191
eISSN - 1361-6560
pISSN - 0031-9155
DOI - 10.1088/1361-6560/abc363
Subject(s) - segmentation , artificial intelligence , computer science , voxel , training set , pattern recognition (psychology) , labeled data , deep learning , sørensen–dice coefficient , image segmentation , atlas (anatomy) , robustness (evolution) , computer vision , medical imaging , anatomy , medicine , biochemistry , gene , chemistry
Deep learning (DL)-based auto-segmentation has the potential for accurate organ delineation in radiotherapy applications but requires large amounts of clean labeled data to train a robust model. However, annotating medical images is extremely time-consuming and requires clinical expertise, especially for segmentation that demands voxel-wise labels. On the other hand, medical images without annotations are abundant and highly accessible. To alleviate the influence of the limited number of clean labels, we propose a weakly supervised DL training approach using deformable image registration (DIR)-based annotations, leveraging the abundance of unlabeled data. We generate pseudo-contours by utilizing DIR to propagate atlas contours onto abundant unlabeled images and train a robust DL-based segmentation model. With 10 labeled TCIA dataset and 50 unlabeled CT scans from our institution, our model achieved Dice similarity coefficient of 87.9%, 73.4%, 73.4%, 63.2% and 61.0% on mandible, left & right parotid glands and left & right submandibular glands of TCIA test set and competitive performance on our institutional clinical dataset and a third party (PDDCA) dataset. Experimental results demonstrated the proposed method outperformed traditional multi-atlas DIR methods and fully supervised limited data training and is promising for DL-based medical image segmentation application with limited annotated data.