TY - GEN
T1 - Weakly supervised localisation for fetal ultrasound images
AU - Toussaint, Nicolas
AU - Khanal, Bishesh
AU - Sinclair, Matthew
AU - Gomez, Alberto
AU - Skelton, Emily
AU - Matthew, Jacqueline
AU - Schnabel, Julia A.
N1 - Publisher Copyright:
© Springer Nature Switzerland AG 2018.
PY - 2018
Y1 - 2018
N2 - This paper addresses the task of detecting and localising fetal anatomical regions in 2D ultrasound images, where only image-level labels are present at training, i.e. without any localisation or segmentation information. We examine the use of convolutional neural network architectures coupled with soft proposal layers. The resulting network simultaneously performs anatomical region detection (classification) and localisation tasks. We generate a proposal map describing the attention of the network for a particular class. The network is trained on 85,500 2D fetal Ultrasound images and their associated labels. Labels correspond to six anatomical regions: head, spine, thorax, abdomen, limbs, and placenta. Detection achieves an average accuracy of 90% on individual regions, and show that the proposal maps correlate well with relevant anatomical structures. This work presents itself as a powerful and essential step towards subsequent tasks such as fetal position and pose estimation, organ-specific segmentation, or image-guided navigation.
AB - This paper addresses the task of detecting and localising fetal anatomical regions in 2D ultrasound images, where only image-level labels are present at training, i.e. without any localisation or segmentation information. We examine the use of convolutional neural network architectures coupled with soft proposal layers. The resulting network simultaneously performs anatomical region detection (classification) and localisation tasks. We generate a proposal map describing the attention of the network for a particular class. The network is trained on 85,500 2D fetal Ultrasound images and their associated labels. Labels correspond to six anatomical regions: head, spine, thorax, abdomen, limbs, and placenta. Detection achieves an average accuracy of 90% on individual regions, and show that the proposal maps correlate well with relevant anatomical structures. This work presents itself as a powerful and essential step towards subsequent tasks such as fetal position and pose estimation, organ-specific segmentation, or image-guided navigation.
KW - Deep learning
KW - Fetal imaging
KW - Image-guided navigation
KW - Object localisation
KW - Ultrasound
KW - Weakly supervised learning
UR - https://www.scopus.com/pages/publications/85057229178
U2 - 10.1007/978-3-030-00889-5_22
DO - 10.1007/978-3-030-00889-5_22
M3 - Conference contribution
AN - SCOPUS:85057229178
SN - 9783030008888
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 192
EP - 200
BT - Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support - 4th International Workshop, DLMIA 2018 and 8th International Workshop, ML-CDS 2018 Held in Conjunction with MICCAI 2018
A2 - Maier-Hein, Lena
A2 - Syeda-Mahmood, Tanveer
A2 - Taylor, Zeike
A2 - Lu, Zhi
A2 - Stoyanov, Danail
A2 - Madabhushi, Anant
A2 - Tavares, João Manuel R.S.
A2 - Nascimento, Jacinto C.
A2 - Moradi, Mehdi
A2 - Martel, Anne
A2 - Papa, Joao Paulo
A2 - Conjeti, Sailesh
A2 - Belagiannis, Vasileios
A2 - Greenspan, Hayit
A2 - Carneiro, Gustavo
A2 - Bradley, Andrew
PB - Springer Verlag
T2 - 4th International Workshop on Deep Learning in Medical Image Analysis, DLMIA 2018 and 8th International Workshop on Multimodal Learning for Clinical Decision Support, ML-CDS 2018 Held in Conjunction with MICCAI 2018
Y2 - 20 September 2018 through 20 September 2018
ER -