DeepFundus: A flow-cytometry-like image quality classifier for boosting the whole life cycle of medical artificial intelligence

Liu, Lixue; Wu, Xiaohang; Lin, Duoru; Zhao, Lanqin; Li, Mingyuan; Yun, Dongyuan; Lin, Zhenzhe; Pang, Jianyu; Li, Longhui; Wu, Yuxuan; Lai, Weiyi; Xiao, Wei; Shang, Yuanjun; Feng, Weibo; Tan, Xiao Jian; Li, Qiang; Liu, Shenzhen; Lin, Xinxin; Sun, Junqi; Zhao, Yiqi; Yang, Xiaohong; Ye, Qin-Ying; Zhong, Yuesi; Huang, Xi; He, Yuan; Fu, Ziwei; Xiang, Yong‐Bing; Zhang, Li; Zhao, Mingwei; Qu, Jingwei; Xu, Fan; Lü, Peng; Li, Jianqiao; Xu, Fabao; Wei, Wenbin; Dong, Li; Dai, Guangzheng; He, Xingru; Yan, Wentao; Zhu, Qiaolin; Liu, Lu; Zhang, Jiaying; Zhou, Wei; Meng, Xiangda; Li, Shiying; Shen, Mei; Jiang, Qin; Chen, Nan; Zhou, Xingtao; Li, Meiyan; Wang, Yan; Zou, Haohan; Zhong, Hua; Yang, Wenyan; Shou, Wulin; Zhong, Xingwu; Yang, Zhenduo; Lin, Ding; Hu, Yongcheng; Tan, Gang; He, Wanji; Zhao, Xin; Chen, Yuzhong; Liu, Yizhi; Lin, Haotian

doi:10.1016/j.xcrm.2022.100912

Cited by 5 publications

(7 citation statements)

References 25 publications

(32 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To assess the role of DeepFundus in the clinical deployment of medical AI, this image quality classifier, as a prescreening tool, was integrated into a certified AI diagnostic system to filter out poor-quality images prior to the detection of DR, AMD, and optic disc edema in real-world prospective cohorts. 1 The results showed that DeepFundus significantly enhanced the real-world performance of established AI diagnostic systems in detecting multiple fundus diseases, demonstrating that DeepFundus had a positive impact on the whole life cycle of medical AI practice.…”

Section: Main Textmentioning

confidence: 91%

“… 7 However, in real-world clinical settings, especially in community hospitals, approximately half of fundus images were of low quality due to multiple socio-environmental factors, indicating that the referral of cases with poor-quality images would pose a significant burden on the healthcare system since most of the cases were found to be disease-free. 1 …”

Section: Main Textmentioning

confidence: 99%

“…To address this issue, Liu et al. 1 developed DeepFundus, a deep-learning-based flow cytometry-like image quality classifier, based on 12,710 images from 4 different cohorts and evaluated it using 53,141 images from another 22 different cohorts nationwide and the Kaggle dataset from the United States. DeepFundus can not only detect poor-quality images but also identify reasons that cause the generation of these images in real time, which can guide a technician to take a precise solution to acquire good-quality images immediately rather than direct referral of cases with poor-quality images.…”

Section: Main Textmentioning

confidence: 99%

“…Based on this consideration, Liu et al. 1 established DeepFundus consisting of 13 AI models to evaluate images in terms of multidimensional quality properties including overall quality, clinical quality factors, affected retinal structures, and refractive media opacity. This design ensures that DeepFundus can be widely used in various clinical scenarios to enhance model performance for detecting multiple retinopathies.…”

Section: Main Textmentioning

confidence: 99%

“… Liu et al. 1 develop a deep-learning-based flow cytometry-like image quality classifier, DeepFundus, for the automated, high-throughput, and multidimensional classification of fundus image quality. DeepFundus significantly improves the real-world performance of established artificial intelligence diagnostics in detecting multiple retinopathies.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Solving data quality issues of fundus images in real-world settings by ophthalmic AI

Chen

2023

Cell Reports Medicine

View full text Add to dashboard Cite

Section: Main Textmentioning

confidence: 91%

Section: Main Textmentioning

confidence: 99%

Section: Main Textmentioning

confidence: 99%

Section: Main Textmentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

Solving data quality issues of fundus images in real-world settings by ophthalmic AI

Chen

2023

Cell Reports Medicine

View full text Add to dashboard Cite

Deep transfer learning from ordinary to capsule esophagogastroduodenoscopy for image quality controlling

Zhang,

Ding

et al. 2023

Engineering Reports

View full text Add to dashboard Cite

Quality controlling for capsule endoscopic images can be completed with the assistance of artificial intelligence, but the labeling process is time‐consuming. Domain adaption is a robust tool for cross‐domain learning to reach a consistent target. Current research aims to study the feasibility and effectiveness of domain adaption from ordinary endoscopic images to capsule endoscopic images in quality controlling. Dynamic adversarial adaptation network (DAAN) was trained to identify low‐quality images using ordinary endoscopic images with corresponding labels (source domain with supervision) and capsule endoscopic images without corresponding labels (target domain without supervision) so that image quality controlling can be transferred from ordinary to capsule endoscopic images. 62,850 images from capsule endoscopy and 17,434 images from ordinary endoscopy were included in developing deep learning models. In internal cross‐validation, DAAN achieved an average area under receiver operating characteristic curve (AUROC) of 0.8638 (95% confidence interval [CI] 0.6753–1.0000) in filtering low‐quality images for capsule endoscopic images, compared with CNN B/16 and L/32, which were also trained with ordinary endoscopic images with corresponding labels. 18,636 images from 355 patients who received capsule endoscopy were prospectively collected. The AUROC of DAAN reached 0.9471 (95% CI 0.9428–0.9511), which surpassed CNN (0.8570 and 95% CI [0.8529–0.8608]) and ViT (L/32: 0.8183 and 95% CI [0.8143–0.8220] and B/16: 0.7779 and 95% CI [0.7960–0.8036]). Domain adaption can complete image quality controlling task in capsule endoscopic images with the supervision of ordinary endoscopic images, whose quantity is smaller so that the annotation workload can be alleviated.

show abstract

Artificial intelligence in ophthalmology: The path to the real-world clinic

Wang

et al. 2023

Cell Reports Medicine

View full text Add to dashboard Cite

DeepFundus: A flow-cytometry-like image quality classifier for boosting the whole life cycle of medical artificial intelligence

Cited by 5 publications

References 25 publications

Solving data quality issues of fundus images in real-world settings by ophthalmic AI

Solving data quality issues of fundus images in real-world settings by ophthalmic AI

Deep transfer learning from ordinary to capsule esophagogastroduodenoscopy for image quality controlling

Artificial intelligence in ophthalmology: The path to the real-world clinic

Contact Info

Product

Resources

About