A deep learning based fine-grained classification algorithm for grading of visual impairment in cataract patients

Abstract

Recent advancements in artificial intelligence (AI) have shown promising potential for the automated screening and grading of cataracts. However, the different types of visual impairment caused by cataracts exhibit similar phenotypes, posing significant challenges for accurately assessing the severity of visual impairment. To address this issue, we propose a dense convolution combined with attention mechanism and multi-level classifier (DAMC_Net) for visual impairment grading. First, the double-attention mechanism is utilized to enable the DAMC_Net to focus on lesions-related regions. Then, a hierarchical multi-level classifier is constructed to enhance the recognition ability in distinguishing the severities of visual impairment, while maintaining a better screening rate for normal samples. In addition, a cost-sensitive method is applied to address the problem of higher false-negative rate caused by the imbalanced dataset. Experimental results demonstrated that the DAMC_Net outperformed ResNet50 and dense convolutional network 121 (DenseNet121) models, with sensitivity improvements of 6.0% and 3.4% on the category of mild visual impairment caused by cataracts (MVICC), and 2.1% and 4.3% on the category of moderate to severe visual impairment caused by cataracts (MSVICC), respectively. The comparable performance on two external test datasets was achieved, further verifying the effectiveness and generalizability of the DAMC_Net.

References

[1] FlaxmanS R, BourneR R, ResnikoffS, et al.. Global causes of blindness and distance vision impairment 1990–2020: a systematic review and meta-analysis[J]. The lancet global health, 2017, 5(12):e1221-e1234

[2] LamD, RaoS K, RatraV, et al.. Cataract[J]. Nature reviews disease primers, 2015, 1(1):1-15

[3] DayA C, FindlO. Femtosecond laser-assisted vs conventional cataract surgery[J]. The lancet, 2020, 395(10219):170-171

[4] HeY, ZhangR, ZhangC, et al.. Clinical outcome of phacoemulsification combined with intraocular lens implantation for primary angle closure/glaucoma (PAC/PACG) with cataract[J]. American journal of translational research, 2021, 13(12):13498

[5] SchweitzerC, BrezinA, CochenerB, et al.. Femtosecond laser-assisted versus phacoemulsification cataract surgery (FEMCAT): a multicentre participant-masked randomised superiority and cost-effectiveness trial[J]. The lancet, 2020, 395(10219): 212-224

[6] World Health Organization. World report on vision[R], 2019, Geneva, WHO

[7] ResnikoffS, FelchW, GauthierT M, et al.. The number of ophthalmologists in practice and training worldwide: a growing gap despite more than 200 000 practitioners[J]. British journal of ophthalmology, 2012, 96(6):783-787

[8] ShenD, WuG, SukH I. Deep learning in medical image analysis[J]. Annual review of biomedical engineering, 2017, 19: 221

[9] KHOJASTE-SARAKHSI M, HAGHIGHI S S, GHOMI S F, et al. Deep learning for Alzheimer’s disease diagnosis: a survey[J]. Artificial intelligence in medicine, 2022: 102332.

[10] ChenS, QiuC, YangW, et al.. Combining edge guidance and feature pyramid for medical image segmentation[J]. Biomedical signal processing and control, 2022, 78: 103960

[11] LotterW, DiabA R, HaslamB, et al.. Robust breast cancer detection in mammography and digital breast tomosynthesis using an annotation-efficient deep learning approach[J]. Nature medicine, 2021, 27(2):244-249

[12] GrzybowskiA, BronaP, LimG, et al.. Artificial intelligence for diabetic retinopathy screening: a review[J]. Eye, 2020, 34(3):451-460

[13] YuK H, BeamA L, KohaneI S. Artificial intelligence in healthcare[J]. Nature biomedical engineering, 2018, 2(10):719-731

[14] SenguptaS, SinghA, LeopoldH A, et al.. Ophthalmic diagnosis using deep learning with fundus images-a critical review[J]. Artificial intelligence in medicine, 2020, 102: 101758

[15] LiZ, JiangJ, ChenK, et al.. Preventing corneal blindness caused by keratitis using artificial intelligence[J]. Nature communications, 2021, 12(1):1-12

[16] LiZ, QiangW, ChenH, et al.. Artificial intelligence to detect malignant eyelid tumors from photographic images[J]. NPJ digital medicine, 2022, 5(1):1-9

[17] ZhangH, NiuK, XiongY, et al.. Automatic cataract grading methods based on deep learning[J]. Computer methods and programs in biomedicine, 2019, 182: 104978

[18] JunayedM S, IslamM B, SadeghzadehA, et al.. CataractNet: an automated cataract detection system using deep learning for fundus images[J]. IEEE access, 2021, 9: 128799-128808

[19] XuX, ZhangL, LiJ, et al.. A hybrid global-local representation CNN model for automatic cataract grading[J]. IEEE journal of biomedical and health informatics, 2020, 24(2):556-567

[20] BloiceM D, RothP M, HolzingerA. Biomedical image augmentation using augmentor[J]. Bioinformatics, 2019, 35(21):4522-4524

[21] HuangG, LiuZ, Van Der MaatenL, et al.. Densely connected convolutional networks[C], 2017, New York, IEEE: 4700-4708

[22] WooS, ParkJ, LeeJ Y, et al.. CBAM: convolutional block attention module[C], 2018, Berlin, Springer: 3-1911211

[23] CholletF. Xception: deep learning with depthwise separable convolutions[C], 2017, New York, IEEE: 1800-1807

[24] JiangJ, WangL, FuH, et al.. Automatic classification of heterogeneous slit-illumination images using an ensemble of cost-sensitive convolutional neural networks[J]. Annals of translational medicine, 2021, 9(7): 550