Rapid Grading of Fundus Photographs for Diabetic Retinopathy Using Crowdsourcing

Background

Screening for diabetic retinopathy is both effective and cost-effective, but rates of screening compliance remain suboptimal. As screening improves, new methods to deal with screening data may help reduce the human resource needs. Crowdsourcing has been used in many contexts to harness distributed human intelligence for the completion of small tasks including image categorization.

Objective

Our goal was to develop and validate a novel method for fundus photograph grading.

Methods

An interface for fundus photo classification was developed for the Amazon Mechanical Turk crowdsourcing platform. We posted 19 expert-graded images for grading by Turkers, with 10 repetitions per photo for an initial proof-of-concept (Phase I). Turkers were paid US $0.10 per image. In Phase II, one prototypical image from each of the four grading categories received 500 unique Turker interpretations. Fifty draws of 1-50 Turkers were then used to estimate the variance in accuracy derived from randomly drawn samples of increasing crowd size to determine the minimum number of Turkers needed to produce valid results. In Phase III, the interface was modified to attempt to improve Turker grading.

Results

Across 230 grading instances in the normal versus abnormal arm of Phase I, 187 images (81.3%) were correctly classified by Turkers. Average time to grade each image was 25 seconds, including time to review training images. With the addition of grading categories, time to grade each image increased and percentage of images graded correctly decreased. In Phase II, area under the curve (AUC) of the receiver-operator characteristic (ROC) indicated that sensitivity and specificity were maximized after 7 graders for ratings of normal versus abnormal (AUC=0.98) but was significantly reduced (AUC=0.63) when Turkers were asked to specify the level of severity. With improvements to the interface in Phase III, correctly classified images by the mean Turker grade in four-category grading increased to a maximum of 52.6% (10/19 images) from 26.3% (5/19 images). Throughout all trials, 100% sensitivity for normal versus abnormal was maintained.

Conclusions

With minimal training, the Amazon Mechanical Turk workforce can rapidly and correctly categorize fundus photos of diabetic patients as normal or abnormal, though further refinement of the methodology is needed to improve Turker ratings of the degree of retinopathy. Images were interpreted for a total cost of US $1.10 per eye. Crowdsourcing may offer a novel and inexpensive means to reduce the skilled grader burden and increase screening for diabetic retinopathy.     

高血压患者外周血管壁腔比与脉搏传导速度的探讨

】     

Arteriovenous Anastomoses of Major Crural Vessels in Humans

Arteriovenous anastomoses of the major crural vessels were studied postmortem. The revealed anastomoses were examined by histological technique.     

Detection of Colorectal Adenocarcinoma and Grading Dysplasia on Histopathologic Slides Using Deep Learning

1. Download : Download high-res image (321KB)
2. Download : Download full-size image

     

单核细胞与淋巴细胞比值在糖尿病视网膜病变中的变化及意义

目的:分析单核细胞(M)与淋巴细胞(L)比值(MLR)在糖尿病视网膜病变(DR)中的变化及与DR分期的关系和临床诊断价值。方法:186例2型糖尿病(T2DM)患者,依其视网膜病变程度,分为单纯糖尿病组(T2DM组,n=78)、非增殖性糖尿病视网膜病变组(NPDR组,n=60)和增殖性糖尿病视网膜病变组(PDR组,n=48),检测各组白细胞计数(WBC)、中性粒细胞(N)、L、M、MLR、糖化血红蛋白(HbA1c)、血肌酐(Scr)、尿素氮(BUN)、总胆固醇(TC)、甘油三酯(TG)、高密度脂蛋白(HDL)及低密度脂蛋白(LDL)水平,并比较组间各指标的差异。Logistic回归分析DR的危险因素,Spearman相关分析危险因素与DR的相关性。ROC曲线分析MLR对DR的诊断价值。结果:各组病程、体重指数(BMI)、HbA1c、TC、LDL、Scr、BUN、N、L、M、MLR差异均有统计学意义(P0.05或P0.01)。与T2DM组比较,NPDR组病程延长,BMI、HbA1c、TC、LDL、Scr、BUN、N、M、MLR水平升高(P0.05或P0.01),L降低(P0.05),PDR组变化较NPDR组更显著(P0.05或P0.01)。Logistic回归分析显示病程、MLR为DR的独立危险因素,Spearman相关性分析显示病程、MLR与DR呈显著正相关(r分别为0.58、0.64,P0.05)。MLR诊断DR的最佳界值为0.22,ROC曲线下面积(AUC)为0.729,敏感度、特异度分别为80.16%、65.36%。结论:MLR是DR的独立危险因素,对DR有中度诊断价值。     

Detection of Neovascularization in Diabetic Retinopathy

Diabetic retinopathy has become an increasingly important cause of blindness. Nevertheless, vision loss can be prevented from early detection of diabetic retinopathy and monitor with regular examination. Common automatic detection of retinal abnormalities is for microaneurysms, hemorrhages, hard exudates, and cotton wool spot. However, there is a worse case of retinal abnormality, but not much research was done to detect it. It is neovascularization where new blood vessels grow due to extensive lack of oxygen in the retinal capillaries. This paper shows that various combination of techniques such as image normalization, compactness classifier, morphology-based operator, Gaussian filtering, and thresholding techniques were used in developing of neovascularization detection. A function matrix box was added in order to classify the neovascularization from natural blood vessel. A region-based neovascularization classification was attempted as a diagnostic accuracy. The developed method was tested on images from different database sources with varying quality and image resolution. It shows that specificity and sensitivity results were 89.4% and 63.9%, respectively. The proposed approach yield encouraging results for future development.     

Assistive Framework for Automatic Detection of All the Zones in Retinopathy of Prematurity Using Deep Learning

Journal of Digital Imaging - Retinopathy of prematurity (ROP) is a potentially blinding disorder seen in low birth weight preterm infants. In India, the burden of ROP is high, with nearly 200,000...     

2型糖尿病及其视网膜病变患者中性粒细胞/淋巴细胞比值变化和意义

目的:分析糖尿病视网膜病变(DR)患者中性粒细胞(N)与淋巴细胞(L)比值(NLR)的变化和临床意义。方法:纳入研究对象183人,包括正常对照组(NC组,n=55),单纯2型糖尿病组(DM组,n=64),非增生性DR组(NPDR组,n=28)和增生性DR组(PDR组,n=36),各组均禁食10h后采集空腹静脉血,采用核酸荧光染色和激光流式技术测定全血N及L计数,计算NLR,采用葡萄糖氧化酶法测定空腹血糖(FPG)、免疫比浊法测定糖化血红蛋白(HbA1c)、电化学发光免疫分析法测定空腹胰岛素(FIN),计算稳态胰岛素抵抗指数(HOMA-IR)。比较各指标组间差异并分析NLR与HOMA-IR的相关性。结果:与NC组比较,DM组、NPDR组和PDR组患者L值、NLR值均显著增高(P0.05),PDR组较NPDR组升高更明显(P0.05)。DM组、NPDR组和PDR组HbA1c、FPG、FINS、HOMA-IR水平均较NC组升高(P0.05),尤以PDR组升高更显著(P0.05)。相关性分析显示DR患者NLR与HOMA-IR水平呈明显正相关(r=0.354,P0.05)。结论:NLR值可协同其它生物指标反映DR患者病情严重程度。     

Increased Matrix Metalloproteinase-1 Activation Enhances Disruption and Regression of k-RasV12–Expressing Arteriovenous Malformation-Like Vessels

1. Download : Download high-res image (80KB)
2. Download : Download full-size image

     

自发性高血压鼠脑底动脉血管活性肠多肽能神经纤维的起源

应用免疫组织化学 ABC法和神经节切除术 ,观察了 2 0只自发性高血压鼠 (分手术 、 、 组和对照组 )脑底动脉血管活性肠多肽能神经纤维的起源。结果显示 ,对照组脑底血管的大脑前动脉、大脑中动脉、大脑后动脉和基底动脉壁上均可见棕褐争的免疫反应阳性纤维 ,纤维似细曲线状 ,呈网状走行。手术 组作一侧蝶腭神经节切除术 ,双侧脑底动脉主要分支的阳性纤维明显减少 ;手术 组作一侧耳神经节切除术 ,双侧大脑后动脉和基底动脉壁上的阳性纤维密度减少 ;手术 组作一侧颈上神经节切除术 ,基底动脉的阳性纤维减少。结果表明 :自发性高血压鼠一侧脑底动脉主要分支的血管活性肠多肽能神经纤维起源于双侧蝶腭神经节、耳神经节和颈上神经节。提示血管活性肠多肽能神经可能在高血压鼠的脑血管运动和高血压的发生和发展方面起着重要的作用     

Visualization of Blood Vessels Using Optical Coherence Tomography

New methods of optical coherence tomography (OCT) have been developed. OCT is a modern technique for noninvasive in vivo examination of upper layers of human skin and subcutaneous blood vessels using nonionizing radiation. The instruments described in this work use a low-power (0.2–0.4 mW) superluminescent diode. Several wavelengths can be applied simultaneously. Raster scanning and raster averaging in the interferometer sample arm increase the signal-to-noise ratio by 4–10 dB.     

Automated Detection and Grading of Diabetic Maculopathy in Digital Retinal Images

Diabetic maculopathy is one of the retinal abnormalities in which a diabetic patient suffers from severe vision loss due to the affected macula. It affects the central vision of the person and causes blindness in severe cases. In this article, we propose an automated medical system for the grading of diabetic maculopathy that will assist the ophthalmologists in early detection of the disease. The proposed system extracts the macula from digital retinal image using the vascular structure and optic disc location. It creates a binary map for possible exudate regions using filter banks and formulates a detailed feature vector for all regions. The system uses a Gaussian Mixture Model-based classifier to the retinal image in different stages of maculopathy by using the macula coordinates and exudate feature set. The evaluation of proposed system is performed by using publicly available standard retinal image databases. The results of our system have been compared with other methods in the literature in terms of sensitivity, specificity, positive predictive value and accuracy. Our system gives higher values as compared to others on the same databases which makes it suitable for an automated medical system for grading of diabetic maculopathy.     

Vasoactivity of Blood Vessels Using a Novel Isovolumic Myograph

The vasoactivity of blood vessel is impaired by cardiovascular disease. The ex-vivo measurements of vasoactivity (wire and pressure myographs) have some inherent difficulties in the characterization of pharmacodynamics and biomechanics. Here, we introduce a novel isovolumic myograph which allows simultaneous measurement of both dynamic pressure and dimensional changes during vasoconstriction or vasodilation. The principle of the isovolumic myograph is based on the utility of a hydraulically closed system with low compliance such that contraction against an incompressible fluid increases the pressure with the closed system while dilation decreases it. We demonstrate the sensitivity and utility of the new technique by the response of elastic (carotid), muscular (femoral) arteries, and femoral veins. The results show characteristic dynamic pressure response for each vessel type. Furthermore, our results show a Hill-type equation for the tension–velocity relation for smooth muscles as well as characteristic curves for each contraction. These novel findings for blood vessels will advance our knowledge of endothelium and vascular smooth muscle mechanics and pharmacodynamics.     

视网膜血管高血压交叉压迫征的量化研究

研究视网膜血管高血压交叉压迫征的量化方法,采用局部血管图像增强的匹配滤波进行预处理、数学形态学的闭合运算进行图像平滑、线化无级扩大法确定血管边界,实施管径可逆量化自动寻优算法,为获取相关疾病诊断的数据提供一种辅助手段。     

高血压鼠脑底动脉神经肽Y能神经与颈上神经节、星状神经节的关系

目的　探讨高血压鼠脑底动脉神经肽Y能神经与颈上神经节、星状神经节的关系。方法　应用神经节切除术和免疫组织化学ABC法 ,对 16只自发性高血压鼠脑底动脉神经肽Y能神经纤维的分布进行了观察。结果　对照组自发性高血压鼠大脑前动脉、大脑中动脉、大脑后动脉和基底动脉壁上均可见神经肽Y能阳性纤维 ,纤维似曲线状 ,多呈网状走行 ,密度较高。手术Ⅰ组作双侧颈上神经节切除术 ,脑底主要动脉的阳性纤维明显减少 ;手术Ⅱ组作双侧星状神经节切除 ,脑底主要动脉壁上的阳性纤维明显减少 ;手术Ⅲ组作双侧颈上神经节和星状神经节切除术 ,脑底主要动脉的阳性纤维完全消失。结论　自发性高血压鼠脑底主要动脉的神经肽Y能神经纤维起源于双侧颈上神经节和双侧星状神经节 ,神经肽Y能神经可能在高血压发病中起作用     

改进的快速FCM及SVM实现糖网白色病灶的自动检测

为构建基于眼底图像的糖尿病视网膜病变(糖网)自动筛查系统,提出一种基于改进的快速FCM(IFFCM)及SVM的糖网白色病灶自动检测算法.首先,利用改进的快速FCM算法,对彩色眼底图像进行粗分割获取糖网白色病灶候选区域,由于该算法将中值滤波添加到FCM算法的准则函数中,同时利用K-means算法的聚类结果对FCM进行聚类中心初始化,使得该算法克服了传统FCM算法计算复杂度高以及对噪声敏感的缺点;其次,采用两层级联分类结构的SVM对候选区域进行分类,即先利用SVM根据候选区域的特征集将白色病灶提取出来,再利用SVM根据另外的特征集将白色病灶中的硬性渗出与棉绒斑区分开,从而实现眼底图像中糖网白色病灶的自动检测.利用该方法对65幅眼底图像进行糖网白色病灶的自动检测,得到图像水平灵敏度100％,特异性95.0％,准确率98.46％;病灶区域水平(硬性渗出/棉绒斑)灵敏度96.42％/97.15％,阳性预测值90.03％/91.18％;平均一幅图像处理时间35.56 s.结果表明:将改进的快速FCM算法所提供的良好粗分割结果与识别率较高的分类器SVM相结合,使得对糖网白色病灶的自动检测结果较优,即该算法能够高效地自动检测出眼底图像中的糖网白色病灶.     

High-Throughput Variation Detection and Genotyping Using Microarrays

The genetic dissection of complex traits may ultimately require a large number of SNPs to be genotyped in multiple individuals who exhibit phenotypic variation in a trait of interest. Microarray technology can enable rapid genotyping of variation specific to study samples. To facilitate their use, we have developed an automated statistical method (ABACUS) to analyze microarray hybridization data and applied this method to Affymetrix Variation Detection Arrays (VDAs). ABACUS provides a quality score to individual genotypes, allowing investigators to focus their attention on sites that give accurate information. We have applied ABACUS to an experiment encompassing 32 autosomal and eight X-linked genomic regions, each consisting of approximately 50 kb of unique sequence spanning a 100-kb region, in 40 humans. At sufficiently high-quality scores, we are able to read approximately 80% of all sites. To assess the accuracy of SNP detection, 108 of 108 SNPs have been experimentally confirmed; an additional 371 SNPs have been confirmed electronically. To access the accuracy of diploid genotypes at segregating autosomal sites, we confirmed 1515 of 1515 homozygous calls, and 420 of 423 (99.29%) heterozygotes. In replicate experiments, consisting of independent amplification of identical samples followed by hybridization to distinct microarrays of the same design, genotyping is highly repeatable. In an autosomal replicate experiment, 813,295 of 813,295 genotypes are called identically (including 351 heterozygotes); at an X-linked locus in males (haploid), 841,236 of 841,236 sites are called identically.