Focusnet++: Attentive aggregated transformations for efficient and accurate medical image segmentation

Authors

Chaitanya Kaul, University of Glasgow
Nick Pears, University of York
Hang Dai, Mohamed Bin Zayed University of Artificial IntelligenceFollow
Roderick Murray-Smith, University of Glasgow
Suresh Manandhar, NAAMII

Document Type

Conference Proceeding

Publication Title

Proceedings - International Symposium on Biomedical Imaging

Abstract

We propose a new residual block for convolutional neural networks and demonstrate its state-of-the-art performance in medical image segmentation. We combine attention mechanisms with group convolutions to create our group attention mechanism, which forms the fundamental building block of our network, FocusNet++. We employ a hybrid loss based on balanced cross entropy, Tversky loss and the adaptive logarithmic loss to enhance the performance along with fast convergence. Our results show that FocusNet++ achieves state-of-the-art results across various benchmark metrics for the ISIC 2018 melanoma segmentation and the cell nuclei segmentation datasets with fewer parameters and FLOPs.

First Page

1042

Last Page

1046

DOI

10.1109/ISBI48211.2021.9433918

Publication Date

4-13-2021

Keywords

Group Attention, Medical Image Segmentation, Residual Learning

Comments

IR Deposit conditions: non-described

Recommended Citation

C. Kaul, N. Pears, H. Dai, R. Murray-Smith and S. Manandhar, "Focusnet++: Attentive Aggregated Transformations For Efficient And Accurate Medical Image Segmentation," in 2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI), 2021, pp. 1042-1046, doi: 10.1109/ISBI48211.2021.9433918.