TY  - GEN
TI  - Learning to Address Intra-segment Misclassification in Retinal Imaging
EP  - 492
Y1  - 2021/09/21/
AV  - public
ID  - discovery10137248
N2  - Accurate multi-class segmentation is a long-standing challenge in medical imaging, especially in scenarios where classes share strong similarity. Segmenting retinal blood vessels in retinal photographs is one such scenario, in which arteries and veins need to be identified and differentiated from each other and from the background. Intra-segment misclassification, i.e. veins classified as arteries or vice versa, frequently occurs when arteries and veins intersect, whereas in binary retinal vessel segmentation, error rates are much lower. We thus propose a new approach that decomposes multi-class segmentation into multiple binary, followed by a binary-to-multi-class fusion network. The network merges representations of artery, vein, and multi-class feature maps, each of which are supervised by expert vessel annotation in adversarial training. A skip-connection based merging process explicitly maintains class-specific gradients to avoid gradient vanishing in deep layers, to favor the discriminative features. The results show that, our model respectively improves F1-score by 4.4%, 5.1%, and 4.2% compared with three state-of-the-art deep learning based methods on DRIVE-AV, LES-AV, and HRF-AV data sets. Code: https://github.com/rmaphoh/Learning-AVSegmentation
PB  - Springer
CY  - Cham, Switzerland
KW  - Multi-class Segmentation
KW  -  Intra-segment Misclassification
KW  -   Retinal Vessel
KW  -  Binary-to-multi-class Fusion Network
T3  - Lecture Notes in Computer Science
SP  - 482
N1  - This version is the author accepted manuscript. For information on re-use, please refer to the publisher?s terms and conditions.
SN  - 1611-3349
UR  - https://doi.org/10.1007/978-3-030-87193-2_46
A1  - Zhou, Y
A1  - Xu, M
A1  - Hu, Y
A1  - Lin, H
A1  - Jacob, J
A1  - Keane, PA
A1  - Alexander, DC
ER  -