Real-world diabetic retinopathy (DR) screening faces a paradox: the most diagnostically critical images are often the lowest in quality, because advanced disease itself produces vitreous hemorrhage, proliferative tissue, and media opacities that degrade fundus imagery. We characterize this quality-severity coupling quantitatively (Spearman ρ = 0.420, odds ratio 4.17 for referable DR in Reject vs. Good strata on DDR, p < 0.001) and show that conventional pipelines work against it: filtering low-quality images discards the most severe cases, while uniform processing leads to misclassification. Both behaviors stem from treating image quality assessment (IQA) as a binary preprocessing decision. We argue that quality should serve as a continuous guidance signal that conditions the diagnostic process, and propose QGDR, a quality-guided dynamic routing framework realizing this paradigm through three coordinated mechanisms: (i) a multi-level IQA module that extracts hierarchical quality features across backbone stages; (ii) a quality-conditioned context gating mechanism that modulates spatial attention according to the predicted quality state; and (iii) an adaptive gated fusion mechanism that routes inputs to scale-specialized experts, with high-quality images preferentially activating fine-scale experts for subtle lesions and degraded images relying on coarse-scale experts for robust global pattern recognition. On EyeQ and DDR, QGDR attains 78.32% accuracy with 0.6863 QWK and 80.85% accuracy with 0.8231 QWK respectively, outperforming representative CNN, transformer, and foundation-model baselines while remaining within the compute envelope of standard single-stream backbones. Counter-intuitively, performance is preserved or improved on the lowest-quality stratum (77.61% on EyeQ-Reject; 82.21% on DDR-Reject, exceeding the Good-quality accuracy on the same dataset), and a Shuffled-IQA counterfactual confirms that QGDR exploits the semantic content of quality information rather than a generic auxiliary signal. Test-only evaluation on the external IDRiD and DeepDRiD cohorts confirms cross-dataset generalization. By treating image quality as guidance rather than as a filter, QGDR preserves screening coverage without sacrificing diagnostic reliability.

Randomized controlled trials(RCTs)provide optimal evidence of the effectiveness and safety of a new drug,a new medical device,or a new therapeutic strategy with the necessary scientific design 1].

Diabetic macular edema (DME) stands as a leading cause for vision loss among the working-age population. Anti-vascular endothelial growth factor (VEGF) agents are currently recognized as the first-line treatment. However, a significant portion of patients remain insensitive to anti-VEGF, resulting in sustained visual impairment. Therefore, it's imperative to predict prognosis and formulate personalized therapeutic regimens. Generative adversarial networks (GANs) have demonstrated remarkably in forecasting prognosis of diseases, yet their performance is still constrained by the limited availability of real-world data and suboptimal image quality, which subsequently impacts the model's outputs. We endeavor to employ preoperative images along with postoperative OCT contours annotated and extracted via LabelMe and OpenCV to train the model in generating postoperative contours of critical OCT structures instead of previous whole retinal morphology, considerably alleviating the difficulty of output phase and diminishing the requisite quantity of training datasets. Our study reveals that the GAN could serve as an auxiliary instrument for ophthalmologists in determining the prognosis of individuals and screening patients with poor responses to anti-VEGF therapy.

Objective This study aims to develop ultrasound biomicroscopy (UBM)-based artificial intelligence (AI) models for preoperative differentiation of acute angle closure (AAC) with or without zonulopathy and to compare their comprehensive diagnostic performance against ophthalmologists as a cross-sectional study.Methods and analysis Three AI models were developed to differentiate AAC with or without zonular laxity or lens subluxation using UBM images and ocular parameters. Their diagnostic performances were analysed, with the best-performing model then compared with two diagnostic methods used by ophthalmologists (logistic regression and UBM image analysis). Additionally, a robustness validation dataset, including images from UBM and anterior segment optical coherence tomography (AS-OCT), was used to validate the robustness of the best-performing AI model.Results A total of 537 eyes were included in this study. The best-performing AI model was image-based and achieved a macro-area under the curve (AUC) of 0.9046 with a diagnostic processing time of 0.03 s per image in differentiating AAC with or without zonulopathy. The manually calculated multinomial logistic regression model achieved a macro-AUC of 0.9373, requiring 1200.00 s per analysis. UBM image analysis achieved a mean accuracy and processing time of 64.17% and 20.13 s, respectively, per image. Robustness validation of the image-based AI model showed an accuracy of 66.67% and 61.11% for UBM and AS-OCT images.Conclusions AI models and ophthalmologists effectively differentiated AAC with or without zonulopathy. However, when evaluated in terms of both accuracy and efficiency, the AI model showed superior comprehensive diagnostic performance, demonstrating high clinical applicability for preoperative diagnosis.

BACKGROUND/AIMS:The aim of this study was to develop and evaluate digital ray, based on preoperative and postoperative image pairs using style transfer generative adversarial networks (GANs), to enhance cataractous fundus images for improved retinopathy detection. METHODS:For eligible cataract patients, preoperative and postoperative colour fundus photographs (CFP) and ultra-wide field (UWF) images were captured. Then, both the original CycleGAN and a modified CycleGAN (C2ycleGAN) framework were adopted for image generation and quantitatively compared using Frechet Inception Distance (FID) and Kernel Inception Distance (KID). Additionally, CFP and UWF images from another cataract cohort were used to test model performances. Different panels of ophthalmologists evaluated the quality, authenticity and diagnostic efficacy of the generated images. RESULTS:A total of 959 CFP and 1009 UWF image pairs were included in model development. FID and KID indicated that images generated by C2ycleGAN presented significantly improved quality. Based on ophthalmologists' average ratings, the percentages of inadequate-quality images decreased from 32% to 18.8% for CFP, and from 18.7% to 14.7% for UWF. Only 24.8% and 13.8% of generated CFP and UWF images could be recognised as synthetic. The accuracy of retinopathy detection significantly increased from 78% to 91% for CFP and from 91% to 93% for UWF. For retinopathy subtype diagnosis, the accuracies also increased from 87%-94% to 91%-100% for CFP and from 87%-95% to 93%-97% for UWF. CONCLUSION:Digital ray could generate realistic postoperative CFP and UWF images with enhanced quality and accuracy for overall detection and subtype diagnosis of retinopathies, especially for CFP.\ TRIAL REGISTRATION NUMBER: This study was registered with ClinicalTrials.gov (NCT05491798).

Traditional Chinese Medicine (TCM) is one of the most promising programs for disease prevention and treatment. Meanwhile, the quality of TCM has garnered much attention. To ensure the quality of TCM, many works are based on the blockchain scheme to design the traceability scheme of TCM to trace its origin. Although these schemes can ensure the integrity, sharability, credibility, and immutability of TCM more effectively, many problems are exposed with the rapid growth of TCM data in blockchains, such as expensive overhead, performance bottlenecks, and the traditional blockchain architecture is unsuitable for TCM data with dynamic growth. Motivated by the aforementioned problems, we propose a novel and lightweight TCM traceability architecture based on the blockchain using sharding (LBS-TCM). Compared to the existing blockchain-based TCM traceability system, our architecture utilizes sharding to develop a novel traceability mechanism that supports more convenient traceability operations for TCM requirements such as uploading, querying, and downloading. Specifically, our architecture consists of a leader shard blockchain layer as its main component, which employs a sharding mechanism to conveniently TCM tracing. Empirical evaluations demonstrated that our architecture showed better performance in many aspects compared to traditional blockchain architectures, such as TCM transaction processing, TCM transaction querying, TCM uploading, etc. In our architecture, tracing TCM has become a very efficient operation, which ensures the quality of TCM and provides great convenience for subsequent TCM analysis and retrospective research.

随着大数据时代的到来,掌握数据科学的相关知识和方法将成为医工融合专业学生的一项核心竞争能力和必备能力.结合医学大数据与人工智能本科生创新实验室的建设,文章探索并实践如何把"本科生进实验室"与"毕业设计论文"有机结合的培养模式,充分利用大学四年时间培养医工融合专业学生的数据科学能力,以期满足"医工结合"背景下的数据科学素质能力要求.

Medical images play an important role in clinical diagnosis. However, the management of medical images has fallen into the dilemma of information islands. Blockchain technology is proposed to solve the problem of medical images’ information islands due to its decentralization, transparency, openness, autonomy, anonymity, and information tampering. However, each node needs to maintain the complete blockchain, and with the amount of medical image data rapidly increasing, the performance limitations of blockchain are emerging, such as poor performance in effectively querying and storing medical image data. Sharding technology is proposed to address the above limitations. In this paper, we propose an efficient management architecture based on a lightweight sharding blockchain for medical image management (EASBM). Our architecture consists of four layers: sharding blockchain layer, distributed storage layer, cache layer, and application layer. The sharding blockchain layer splits each block into head and body and divides each participant into corresponding shards, implementing a lightweight blockchain framework. The distributed storage layer saves the raw data and hash values of medical images, reducing the storage burden of blockchain and implementing off-chain storage. Our empirical evaluations suggest that the time spent querying a record is less than 4 ms in our proposed architecture, which contains a network of 1024 participating nodes.

Purpose To develop a deep learning system to differentiate demyelinating optic neuritis (ON) and non-arteritic anterior ischemic optic neuropathy (NAION) with overlapping clinical profiles at the acute phase. Methods We developed a deep learning system (ONION) to distinguish ON from NAION at the acute phase. Color fundus photographs (CFPs) from 871 eyes of 547 patients were included, including 396 ON from 232 patients and 475 NAION from 315 patients. Efficientnet-B0 was used to train the model, and the performance was measured by calculating the sensitivity, specificity, and area under the receiver operating characteristic curve (AUC). Also, Cohen’s kappa coefficients were obtained to compare the system’s performance to that of different ophthalmologists. Results In the validation data set, the ONION system distinguished between acute ON and NAION achieved the following mean performance: time-consuming (23 s), AUC 0.903 (95% CI 0.827–0.947), sensitivity 0.796 (95% CI 0.704–0.864), and specificity 0.865 (95% CI 0.783–0.920). Testing data set: time-consuming (17 s), AUC 0.902 (95% CI 0.832–0.944), sensitivity 0.814 (95% CI 0.732–0.875), and specificity 0.841 (95% CI 0.762–0.897). The performance (κ = 0.805) was comparable to that of a retinal expert (κ = 0.749) and was better than the other four ophthalmologists (κ = 0.309–0.609). Conclusion The ONION system performed satisfactorily distinguishing ON from NAION at the acute phase. It might greatly benefit the challenging differentiation between ON and NAION.

Medical image synthesis is a key technology in computer aided diagnosis due to its wide applications. Recently, the end-to-end models based on generative adversarial networks（GAN） learn the true data distribution to guide the image generation via the competing generator and discriminator. However, the high resolution of the medical images and the lack of labeled samples make these models hard to train. Furthermore, they are not robust against data perturbations and vulnerable to malicious attacks. In this paper, we propose a novel image synthesis model using robust conditional GAN,namely MiSrc-GAN. The MiSrc-GAN contains a progressive resolution generator, a multi-scale discriminator and a paired adversarial examples generating module. Through effective integrating GAN framework and adversarial examples, the MiSrc-GAN is able to simultaneously improve the robustness of the multi-scale discriminator and the quality of synthetic images under the joint probability distribution of the original medical images and their translating versions. The extensive experiments show that the proposed method achieves state-of-the-art image synthesis results on both CSC and REFUGE datasets.

Purpose: This study sought to assess the predictive performance of optical coherence tomography (OCT) images for the response of diabetic macular edema (DME) patients to anti-vascular endothelial growth factor (VEGF) therapy generated from baseline images using generative adversarial networks (GANs). Methods: Patient information, including clinical and imaging data, was obtained from inpatients at the Ophthalmology Department of Qilu Hospital. 715 and 103 pairs of pre-and post-treatment OCT images of DME patients were included in the training and validation sets, respectively. The post-treatment OCT images were used to assess the validity of the generated images. Six different GAN models (CycleGAN, PairGAN, Pix2pixHD, RegGAN, SPADE, UNIT) were applied to predict the efficacy of anti-VEGF treatment by generating OCT images. Independent screening and evaluation experiments were conducted to validate the quality and comparability of images generated by different GAN models. Results: OCT images generated f GAN models exhibited high comparability to the real images, especially for edema absorption. RegGAN exhibited the highest prediction accuracy over the CycleGAN, PairGAN, Pix2pixHD, SPADE, and UNIT models. Further analyses were conducted based on the RegGAN. Most post-therapeutic OCT images (95/103) were difficult to differentiate from the real OCT images by retinal specialists. A mean absolute error of 26.74 +/- 21.28 mu m was observed for central macular thickness (CMT) between the synthetic and real OCT images. Conclusion: Different generative adversarial networks have different prognostic efficacy for DME, and RegGAN yielded the best performance in our study. Different GAN models yielded good accuracy in predicting the OCTbased response to anti-VEGF treatment at one month. Overall, the application of GAN models can assist clinicians in prognosis prediction of patients with DME to design better treatment strategies and follow-up schedules.

Abstract Anemia is one of the most widespread clinical symptoms all over the world, which could bring adverse effects on people's daily life and work. Considering the universality of anemia detection and the inconvenience of traditional blood testing methods, many deep learning detection methods based on image recognition have been developed in recent years, including the methods of anemia detection with individuals’ images of conjunctiva. However, existing methods using one single conjunctiva image could not reach comparable accuracy in anemia detection in many real-world application scenarios. To enhance intelligent anemia detection using conjunctiva images, we proposed a new algorithmic framework which could make full use of the data information contained in the image. To be concrete, we proposed to fully explore the global and local information in the image, and adopted a two-branch neural network architecture to unify the information of these two aspects. Compared with the existing methods, our method can fully explore the information contained in a single conjunctiva image and achieve more reliable anemia detection effect. Compared with other existing methods, the experimental results verified the effectiveness of the new algorithm.

PURPOSE:To generate and evaluate individualized post-therapeutic optical coherence tomography (OCT) images that could predict the short-term response of anti-vascular endothelial growth factor (VEGF) therapy for diabetic macular edema (DME) based on pre-therapeutic images using generative adversarial network (GAN).METHODS:Real-world imaging data were collected at the Department of Ophthalmology, Qilu Hospital. A total of 561 pairs of pre-therapeutic and post-therapeutic OCT images of patients with DME were retrospectively included in the training set, 71 pre-therapeutic OCT images were included in the validation set, and their corresponding post-therapeutic OCT images were used to evaluate the synthetic images. A pix2pixHD method was adopted to predict post-therapeutic OCT images in DME patients that received anti-VEGF therapy. The quality and similarity of synthetic OCT images were evaluated independently by a screening experiment and an evaluation experiment.RESULTS:The post-therapeutic OCT images generated by the GAN model based on big data were comparable to the actual images, and the response of edema resorption was also close to the ground truth. Most synthetic images (65/71) were difficult to differentiate from the actual OCT images by retinal specialists. The mean absolute error (MAE) of the central macular thickness (CMT) between the synthetic OCT images and the actual images was 24.51 ± 18.56 μm.CONCLUSIONS:The application of GAN can objectively demonstrate the individual short-term response of anti-VEGF therapy one month in advance based on OCT images with high accuracy, which could potentially help to improve treatment compliance of DME patients, identify patients who are not responding well to treatment and optimize the treatment program.

Late fetal growth restriction (FGR) is a common complication of pregnancy characterized by chronic hypoxia. However, late FGR is in a dilemma of the high incidence but low detection rate. Depending on the non-invasiveness and convenient operation, the routine cardiotocography (CTG) allows continuous monitoring fetal heart rate (FHR) to assess fetal intrauterine stockpiling ability. In this paper, we aimed to explore the FHR pattern of late FGR in routine CTG. For analysis, the FHR features were acquired using routine CTG in a population of 160 healthy and 102 late FGR fetuses published in IEEE Dataport. First, we explored the relationships among FHR features and their importance on late FGR assessment by utilizing hypothesis testing, principal component analysis (PCA) and Spearman correlation analysis. Second, we presented a regression coefficient-based backward-stepwise-selection of association rules analysis (ARA) called backward-stepwise Max-R 2 Apriori ARA, to find the optimum itemset that helps diagnose late FGRs from healthy fetuses. The hypothesis testing, PCA and Spearman correlation analysis found eight FHR features were highly relevant to the late FGR. Moreover, the backward-stepwise Max-R2 Apriori ARA validated the correlation and interpretation about FHR features of late FGR. In conclusion, the analysis results are consistent with clinical knowledge on late FGR and help screen late FGR in antepartum fetal monitoring.

Aiming at the imbalance and cost-sensitive problem of sample categories in actual fetal monitoring, as well as actual needs, we proposed a category imbalance fetal contraction monitoring model based on GBDT (Gradient Boosting Decision Tree) combined learning. Subsets with balanced category were generated by random under-sampling and applied to train several GBDT base classifiers using the method of feature selection. We integrated the base classifiers by the simple average method and calculated the final prediction probability. In this study, AUC and cost-sensitive error rate were used as evaluation indicators to compare with the commonly used single learning models such as Decision Tree, Logistic Regression and combined learning models like Random Forest to verify the effectiveness of the model.

电子病历包含病人的隐私信息,如何在保密情况下进行电子病历内容搜索是医院电子病历数据有效利用的难点.针对电子病历密文搜索,提出一种基于布隆过滤器(BF)和B+树的快速搜索方法.该技术对电子病历建立BF,按照倒排索引方式组织起来并建立B+树,能够支持在不泄露用户搜索关键词的前提下,对电子病历密文直接进行查询,安全级别达到IND-CKA级别.实验结果表明:由于使用BF与B+树结构,空间过滤特性高,无须解密即可达99％的过滤效率,查询效率高,且支持布尔查询,有效促进加密电子病历数据的应用.

To predict visual acuity (VA) and post-therapeutic optical coherence tomography (OCT) images 1, 3, and 6 months after laser treatment in patients with central serous chorioretinopathy (CSC) by artificial intelligence (AI). Real-world clinical and imaging data were collected at Zhongshan Ophthalmic Center (ZOC) and Xiamen Eye Center (XEC). The data obtained from ZOC (416 eyes of 401 patients) were used as the training set; the data obtained from XEC (64 eyes of 60 patients) were used as the test set. Six different machine learning algorithms and a blending algorithm were used to predict VA, and a pix2pixHD method was adopted to predict post-therapeutic OCT images in patients after laser treatment. The data for VA predictions included clinical features obtained from electronic medical records (20 features) and measured features obtained from fundus fluorescein angiography, indocyanine green angiography, and OCT (145 features). The data for OCT predictions included 480 pairs of pre- and post-therapeutic OCT images. The VA and OCT images predicted by AI were compared with the ground truth. In the VA predictions of XEC dataset, the mean absolute errors (MAEs) were 0.074–0.098 logMAR (within four to five letters), and the root mean square errors were 0.096–0.127 logMAR (within five to seven letters) for the 1-, 3-, and 6-month predictions, respectively; in the post-therapeutic OCT predictions, only about 5.15% (5 of 97) of synthetic OCT images could be accurately identified as synthetic images. The MAEs of central macular thickness of synthetic OCT images were 30.15 ± 13.28 μm and 22.46 ± 9.71 μm for the 1- and 3-month predictions, respectively. This is the first study to apply AI to predict VA and post-therapeutic OCT of patients with CSC. This work establishes a reliable method of predicting prognosis 6 months in advance; the application of AI has the potential to help reduce patient anxiety and serve as a reference for ophthalmologists when choosing optimal laser treatments.

Cardiotocography (CTG) is frequently used as a method of diagnosing fetal distress during pregnancy and delivery, including listening to Fetal Heart Rate (FHR) and monitoring Uterine Contractions (UC). Many scholars have contributed to classify CTG data through machine learning methods, intending to reduce the consuming-time and mistakes during obstetricians’ identification. In this study, we used repeated holdout cross-validation for data pre-processing and classified the CTG dataset as a normal class and pathological class. Sensitivity, specificity, accuracy, and AUC were involved to measure the performance of the models. According to the results of ASSEMBLE AdaBoost and AdaBoost, we can conclude that ASSEMBLE AdaBoost is robust in the classification of CTG data. Training with few labeled data and amounts of unlabeled data, ASSEMBLE AdaBoost achieved an accuracy of over 90%, which was superior to AdaBoost.

Intelligent classification of antepartum cardiotocography (CTG) can assist obstetricians to make clinical decisions, which is helpful to improve the accuracy of fetal abnormality detection. However, most of the existing machine learning based classification models fail to meet the clinical requirements. In this work, a method is proposed to improve the classification accuracy by using Deep Forest (DF) algorithm. Principal component analysis and visualization were adopted to mine the distribution characteristics of the CTG data. After data preprocessing, deep forest multi-granularity scanning phase was used to explore the connection between the case characteristic. Then the cascade forest phase, which was designed to integrate Random Forest (RF), Weighted Random Forest (WRF), Completely Random Forest (CRF) and Gradient Boosting Decision Tree (GBDT) as the basic classifiers, performed deep iterations and finally got the best performance model. Compared with the traditional machine learning models, deep neural network and the state-of-art CTG classification models, the results show that the accuracy value, average F1 value and Area Under the Curve (AUC) value were 92.64 %, 92.01 % and 0.990 respectively in the external public data set, and were 91.64 %, 88.92 % and 0.9493 respectively in the internal private data set, which were the most excellent among all comparison models. In conclusion, the proposed DF model is effective and feasible, and has a good application prospect in the intelligent assessment of antenatal fetal health status.

采用深度森林框架构建基于不平衡电子胎心宫缩监护数据的多分类判别模型,验证模型有效性,结果表明该模型预测性能较好,极大降低误判率,在产前胎儿健康状况智能评估中有良好应用前景.