TY - GEN
T1 - Learning to Detect Brain Lesions from Noisy Annotations
AU - Karimi, Davood
AU - Peters, Jurriaan M.
AU - Ouaalam, Abdelhakim
AU - Prabhu, Sanjay P.
AU - Sahin, Mustafa
AU - Krueger, Darcy A.
AU - Kolevzon, Alexander
AU - Eng, Charis
AU - Warfield, Simon K.
AU - Gholipour, Ali
N1 - Funding Information:
This study was supported in part by the NIH grants R01EB019483, R01EB018988, and R01NS106030. The data was acquired as part of the Developmental Synaptopathies Consortium (U54NS092090).
Publisher Copyright:
© 2020 IEEE.
PY - 2020/4
Y1 - 2020/4
N2 - Supervised training of deep neural networks in medical imaging applications relies heavily on expert-provided annotations. These annotations, however, are often imperfect, as voxel-by-voxel labeling of structures on 3D images is difficult and laborious. In this paper, we focus on one common type of label imperfection, namely, false negatives. Focusing on brain lesion detection, we propose a method to train a convolutional neural network (CNN) to segment lesions while simultaneously improving the quality of the training labels by identifying false negatives and adding them to the training labels. To identify lesions missed by annotators in the training data, our method makes use of the 1) CNN predictions, 2) prediction uncertainty estimated during training, and 3) prior knowledge about lesion size and features. On a dataset of 165 scans of children with tuberous sclerosis complex from five centers, our method achieved better lesion detection and segmentation accuracy than the baseline CNN trained on the noisy labels, and than several alternative techniques.
AB - Supervised training of deep neural networks in medical imaging applications relies heavily on expert-provided annotations. These annotations, however, are often imperfect, as voxel-by-voxel labeling of structures on 3D images is difficult and laborious. In this paper, we focus on one common type of label imperfection, namely, false negatives. Focusing on brain lesion detection, we propose a method to train a convolutional neural network (CNN) to segment lesions while simultaneously improving the quality of the training labels by identifying false negatives and adding them to the training labels. To identify lesions missed by annotators in the training data, our method makes use of the 1) CNN predictions, 2) prediction uncertainty estimated during training, and 3) prior knowledge about lesion size and features. On a dataset of 165 scans of children with tuberous sclerosis complex from five centers, our method achieved better lesion detection and segmentation accuracy than the baseline CNN trained on the noisy labels, and than several alternative techniques.
KW - brain lesion detection
KW - deep learning
KW - imperfect labels
KW - noisy labels
KW - tuberous sclerosis complex
UR - http://www.scopus.com/inward/record.url?scp=85085860519&partnerID=8YFLogxK
U2 - 10.1109/ISBI45749.2020.9098599
DO - 10.1109/ISBI45749.2020.9098599
M3 - Conference contribution
AN - SCOPUS:85085860519
T3 - Proceedings - International Symposium on Biomedical Imaging
SP - 1910
EP - 1914
BT - ISBI 2020 - 2020 IEEE International Symposium on Biomedical Imaging
PB - IEEE Computer Society
T2 - 17th IEEE International Symposium on Biomedical Imaging, ISBI 2020
Y2 - 3 April 2020 through 7 April 2020
ER -