Generalised Dice Overlap as a Deep Learning Loss Function for Highly Unbalanced Segmentations
Deep-learning has proved in recent years to be a powerful tool for image analysis and is now widely used to segment both 2D and 3D medical images. Deep-learning segmentation frameworks rely not only on the choice of network architecture but also on the choice of loss function. When the segmentation process targets rare observations, a severe class imbalance is likely to occur between candidate labels, thus resulting in sub-optimal performance. In order to mitigate this issue, strategies such as the weighted cross-entropy function, the sensitivity function or the Dice loss function, have been proposed. In this work, we investigate the behavior of these loss functions and their sensitivity to learning rate tuning in the presence of different rates of label imbalance across 2D and 3D segmentation tasks. We also propose to use the class re-balancing properties of the Generalized Dice overlap, a known metric for segmentation assessment, as a robust and accurate deep-learning loss function for unbalanced tasks.
This work made use of Emerald, a GPU accelerated HPC, made available by the Science & Engineering South Consortium operated in partnership with the STFC Rutherford-Appleton Laboratory. This work was funded by the EPSRC (EP/H046410/1, EP/J020990/1, EP/K005278, EP/H046410/1), the MRC (MR/J01107X/1), the EU-FP7 project VPH-DARE@ IT (FP7-ICT-2011-9-601055), the Wellcome Trust (WT101957), the NIHR Biomedical Research Unit (Dementia) at UCL and the NIHR University College London Hospitals BRC (NIHR BRC UCLH/UCL High Impact Initiative- BW.mn.BRC10269).
- 1.Brosch, T., Yoo, Y., Tang, L.Y.W., Li, D.K.B., Traboulsee, A., Tam, R.: Deep convolutional encoder networks for multiple sclerosis lesion segmentation. In: Navab, N., Hornegger, J., Wells, W.M., Frangi, A.F. (eds.) MICCAI 2015. LNCS, vol. 9351, pp. 3–11. Springer, Cham (2015). doi: 10.1007/978-3-319-24574-4_1 CrossRefGoogle Scholar
- 2.Crum, W., Camara, O., Hill, D.: Generalized overlap measures for evaluation and validation in medical image analysis. IEEE TMI 25(11), 1451–1461 (2006)Google Scholar
- 3.Havaei, M., Davy, A., Warde-Farley, D., Biard, A., Courville, A., Bengio, Y., Pal, C., Jodoin, P.M., Larochelle, H.: Brain tumor segmentation with deep neural networks. MIA 35, 18–31 (2017)Google Scholar
- 4.Kamnitsas, K., Ledig, C., Newcombe, V.F., Simpson, J.P., Kane, A.D., Menon, D.K., Rueckert, D., Glocker, B.: Efficient multi-scale 3D CNN with fully connected CRF for accurate brain lesion segmentation. MIA 36, 61–78 (2017)Google Scholar
- 5.Lai, M.: Deep learning for medical image segmentation arXiv:1505.02000 (2015)
- 6.Li, W., Wang, G., Fidon, L., Ourselin, S., Cardoso, M.J., Vercauteren, T.: On the compactness, efficiency, and representation of 3D convolutional networks: brain parcellation as a pretext task. In: Niethammer, M., Styner, M., Aylward, S., Zhu, H., Oguz, I., Yap, P.-T., Shen, D. (eds.) IPMI 2017. LNCS, vol. 10265, pp. 348–360. Springer, Cham (2017). doi: 10.1007/978-3-319-59050-9_28 CrossRefGoogle Scholar
- 7.Menze, B.H., et al.: The multimodal brain tumor image segmentation benchmark (BRATS). IEEE TMI 34(10), 1993–2024 (2015)Google Scholar
- 8.Milletari, F., Navab, N., Ahmadi, S.A.: V-Net: fully convolutional neural networks for volumetric medical image segmentation. In: 2016 Fourth International Conference on 3D Vision (3DV), pp. 565–571. IEEE, October 2016Google Scholar
- 10.Zheng, Y., Liu, D., Georgescu, B., Nguyen, H., Comaniciu, D.: 3D deep learning for efficient and robust landmark detection in volumetric data. In: Navab, N., Hornegger, J., Wells, W.M., Frangi, A.F. (eds.) MICCAI 2015. LNCS, vol. 9349, pp. 565–572. Springer, Cham (2015). doi: 10.1007/978-3-319-24553-9_69 CrossRefGoogle Scholar