Fine-tuning Convolutional Neural Networks: a comprehensive guide and benchmark analysis for Glaucoma Screening

This work aimed at giving a comprehensive, in-detailed and benchmark guide on the route to fine-tuning Convolutional Neural Networks (CNNs) for glaucoma screening. Transfer learning consists in a promising alternative to train CNNs from scratch, to avoid the huge data and resources requirements. After a thorough study of five state-of-the-art CNNs architectures, a complete and well-explained strategy for fine-tuning these networks is proposed, using hyperparameter grid-searching and two-phase training approach. Excellent performance is reached on model evaluation, with a 0.9772 AUROC validation rate, giving arise to reliable glaucoma diagnosis-help systems. Also, a baseline benchmark analysis is conducted, studying the models according to performance indices such as model complexity and size, AUROC density and inference time. This in-depth analysis allows a rigorous comparison between model characteristics, and is useful for giving practioners important trademarks for prospective applications and deployments.

[1]  Sergey Ioffe,et al.  Rethinking the Inception Architecture for Computer Vision , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[2]  Tien Yin Wong,et al.  Glaucoma detection based on deep convolutional neural network , 2015, 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC).

[3]  Francisco Fumero,et al.  RIM-ONE: An open retinal image database for optic nerve evaluation , 2011, 2011 24th International Symposium on Computer-Based Medical Systems (CBMS).

[4]  Sangsoo Kim,et al.  A deep learning model for the detection of both advanced and early glaucoma using fundus photography , 2018, PloS one.

[5]  Nima Tajbakhsh,et al.  Convolutional Neural Networks for Medical Image Analysis: Full Training or Fine Tuning? , 2016, IEEE Transactions on Medical Imaging.

[6]  Bo Chen,et al.  MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications , 2017, ArXiv.

[7]  T. Wong,et al.  Global prevalence of glaucoma and projections of glaucoma burden through 2040: a systematic review and meta-analysis. , 2014, Ophthalmology.

[8]  Andreas K. Maier,et al.  Robust Vessel Segmentation in Fundus Images , 2013, Int. J. Biomed. Imaging.

[9]  S. Kingman.  Glaucoma is second leading cause of blindness globally. , 2004, Bulletin of the World Health Organization.

[10]  François Chollet,et al.  Keras: The Python Deep Learning library , 2018 .

[11]  M. He,et al.  Efficacy of a Deep Learning System for Detecting Glaucomatous Optic Neuropathy Based on Color Fundus Photographs. , 2018, Ophthalmology.

[12]  Guigang Zhang,et al.  Deep Learning , 2016, Int. J. Semantic Comput..

[13]  Jost B. Jonas,et al.  Optic disc morphometry in chronic primary open-angle glaucoma , 1988, Graefe's Archive for Clinical and Experimental Ophthalmology.

[14]  Yoshua Bengio,et al.  Gradient-based learning applied to document recognition , 1998, Proc. IEEE.

[15]  Xiaochun Cao,et al.  Disc-Aware Ensemble Network for Glaucoma Screening From Fundus Image , 2018, IEEE Transactions on Medical Imaging.

[16]  M. Sonka,et al.  Retinal Imaging and Image Analysis. , 2010, IEEE transactions on medical imaging.

[17]  Jayanthi Sivaswamy,et al.  A Comprehensive Retinal Image Dataset for the Assessment of Glaucoma from the Optic Nerve Head Analysis , 2015 .

[18]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[19]  M F Armaly,et al.  The cup-disc ratio. The findings of tonometry and tonography in the normal eye. , 1969, Archives of ophthalmology.

[20]  Weidong (Tom) Cai,et al.  A Deep Learning-Based Algorithm Identifies Glaucomatous Discs Using Monoscopic Fundus Photographs. , 2018, Ophthalmology. Glaucoma.

[21]  Mohamed Akil,et al.  Fully automated method for glaucoma screening using robust optic nerve head detection and unsupervised segmentation based cup-to-disc ratio computation in retinal fundus images , 2019, Comput. Medical Imaging Graph..

[22]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[23]  Tien Yin Wong,et al.  Superpixel Classification Based Optic Disc and Optic Cup Segmentation for Glaucoma Screening , 2013, IEEE Transactions on Medical Imaging.

[24]  S. Drance,et al.  Neuroretinal rim area in early glaucoma. , 1985, American journal of ophthalmology.

[25]  Valery Naranjo,et al.  CNNs for automatic glaucoma assessment using fundus images: an extensive validation , 2019, BioMedical Engineering OnLine.

[26]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[27]  Matthew B. Blaschko,et al.  Convolutional neural network transfer for automated glaucoma identification , 2017, Symposium on Medical Information Processing and Analysis.

[28]  Faraz Oloumi,et al.  Deep learning in ophthalmology: a review. , 2018, Canadian journal of ophthalmology. Journal canadien d'ophtalmologie.

[29]  Kilian Q. Weinberger,et al.  Densely Connected Convolutional Networks , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).