Zheen Zhao

Duke University Medical Center, Durham, NC, United States

Are you Zheen Zhao?

Claim your profile

Publications (16)20.69 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: To validate the accuracy of ROPtool software in measuring retinal vascular width and tortuosity in a large image set compared with expert diagnoses. Tortuosity and dilation indexes generated by ROPtool were compared with 3 expert consensus grades of normal, pre-plus, or plus disease for 368 quadrants in 92 RetCam (Clarity Medical Systems, Pleasanton, California) fundus images. Sensitivity and specificity of ROPtool software in diagnosing tortuosity and dilation sufficient for plus and pre-plus disease were calculated. These measures were compared with individual accuracies of 3 experienced pediatric ophthalmologists. The mean tortuosity indexes for expert-diagnosed categories of normal, pre-plus, and plus disease were 7.04, 18.73, and 34.62, respectively (P < .001), and the mean dilation indexes were 9.63, 12.05, and 13.61, respectively (P < .001). When optimal tortuosity and dilation index thresholds (from receiver operating characteristic curves) were applied, resultant sensitivity and specificity were 0.913 and 0.863, respectively, for plus tortuosity and 0.782 and 0.840, respectively, for plus dilation. These values were comparable to the performance of examiners judged against the same expert panel. ROPtool version 2.1.5 accurately measures tortuosity and dilation of posterior pole blood vessels in RetCam images, corresponding well with expert diagnostic categories of normal, pre-plus, and plus disease and performing comparably to experienced examiners.
    Archives of ophthalmology 07/2010; 128(7):847-52. · 3.86 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The primary indication for laser treatment in retinopathy of prematurity (ROP) is plus disease, or abnormal dilation and tortuosity of arterioles and venules. ROPtool is a computer program that traces retinal blood vessels and measures their width and tortuosity. Our purpose was to gain insight into the evolution of plus disease by applying ROPtool to RetCam images from eyes of infants who had serial photographs taken during their ROP screening period. Serial images were collected from eyes of 62 infants screened for ROP as part of another study. Fifty-nine images of one eye of 7 infants who developed plus disease were selected and analyzed by ROPtool. The average tortuosity of the most tortuous blood vessel and the average width of the most dilated vessel in each quadrant were calculated for each image. Tortuosity increased from an average of 7.72 units at the first examination to 24.44 units at the examination with maximum tortuosity, or an increase of 217% over a mean time period of 6.2 weeks. Two eyes had an increase in tortuosity of more than 500% from the first examination. Vessel width increased from an average of 8.60 units at the first examination to 11.03 units at the examination with maximum blood vessel width, or an increase of 28% over a mean time period of 5.1 weeks. ROPtool can measure changes in retinal vascular dilation and tortuosity in individual eyes over time. As plus disease develops, changes in tortuosity are sometimes very large, whereas changes in vessel width tend to be more subtle. Quantification of plus disease over time may help to improve our understanding of its mechanism and to monitor disease progression or response to treatment.
    Transactions of the American Ophthalmological Society 12/2009; 107:47-52.
  • [Show abstract] [Hide abstract]
    ABSTRACT: Plus disease is abnormal retinal vascular dilation and tortuosity, and it is now the primary indication for laser treatment in retinopathy of prematurity (ROP). ROPtool is a computer program that measures retinal arteriolar tortuosity. Our aim was to assess the accuracy of ROPtool's newly developed measurement of retinal vascular width (dilation). ROPtool was used to measure the width of 154 blood vessels in 20 high-quality RetCam images from 20 premature infants. ROPtool's accuracy was determined by comparing results with the mean grades of 2 authors who scored retinal vascular dilation using a 10-point scale. There was very good correlation (r = 0.80) between ROPtool's measurement of retinal vascular dilation and author judgment. Areas under receiver operating characteristics curves for identification of dilation sufficient for plus disease and for pre-plus disease were 0.93 and 0.90, respectively. At an optimal point on the receiver operating characteristics curve, ROPtool's sensitivity for diagnosing dilation sufficient for plus disease was 89% (24/27), and its specificity was 83% (106/127). In addition to measuring retinal vascular tortuosity, ROPtool now accurately measures retinal vascular width in high-quality RetCam images. Application of this technology has the potential to remove subjectivity from the assessment of plus disease.
    Retina (Philadelphia, Pa.) 06/2009; 29(8):1182-7. · 2.93 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Plus disease is the major criterion for laser treatment of retinopathy of prematurity. ROPtool is a computer program that traces retinal blood vessels and measures their tortuosity. Our objectives were to determine (1) whether examiners could accurately discriminate between arterioles and venules and (2) whether tortuosity sufficient for plus disease and pre-plus disease was assessed most accurately by considering arterioles, venules, or both. One hundred retinal vessels were identified in 25 images randomly selected from 184 total images. Three pediatric ophthalmologists independently designated vessels as arteriole or venule. Seventy-seven images that had at least 1 traceable arteriole and venule in each quadrant were analyzed by ROPtool, and the results were compared with the consensus of 3 expert examiners. Receiver operating characteristics (ROC) curves were generated and areas under the curves calculated to quantify the diagnostic utility of ROPtool's assessment of tortuosity of arterioles, venules, and both. Three pediatric ophthalmologists agreed on the designation of arteriole or venule for 83 of 100 blood vessels. With the use of expert consensus as the reference standard, areas under the ROC curves for identification of tortuosity sufficient for plus disease were 0.91, 0.70, and 0.93 for arterioles, venules, and both, respectively. Areas under the ROC curves for identification of tortuosity sufficient for pre-plus disease were 0.91, 0.63, and 0.90 for arterioles, venules, and both, respectively. When considering whether tortuosity is sufficient for plus or pre-plus disease, the assessment of either arterioles alone or of arterioles and venules together resulted in high diagnostic accuracy.
    Journal of AAPOS: the official publication of the American Association for Pediatric Ophthalmology and Strabismus / American Association for Pediatric Ophthalmology and Strabismus 05/2009; 13(2):181-5. · 1.07 Impact Factor
  • Journal of American Association for Pediatric Ophthalmology and Strabismus 02/2009; 13(1). · 1.14 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: "ROPtool" is a computer program that measures retinal blood vessel tortuosity. Our aim was to determine the feasibility and accuracy of analyzing images with ROPtool, which were obtained using video indirect ophthalmoscopy. Forty-five posterior pole still images captured from indirect ophthalmoscopy video clips were selected; 20 were selected for high quality and 25 were randomly selected. One of the authors (S.A.) used ROPtool to measure tortuosity for each quadrant of each image. Two of the authors (D.K.W. and S.F.F.) independently judged tortuosity on a 10-point scale, and their averaged grades were used as the reference standard. Among randomly selected images, ROPtool was able to trace at least two major vessels in 43 of 100 quadrants (43%). Lighter fundus pigment color was associated with ROPtool's ability to analyze images (P = 0.004). When considering analyzable images only, ROPtool's sensitivity in detecting tortuosity sufficient for plus disease was 83% (5/6) and specificity was 90% (18/20). ROPtool's sensitivity for pre-plus tortuosity was 100% (9/9) and specificity was 71% (12/17). ROPtool is useful for analyzing video indirect ophthalmoscopy images only when applied to those with high quality. When analyzing these images, ROPtool has very good accuracy compared to consensus of experienced examiners.
    Retina (Philadelphia, Pa.) 08/2008; 28(10):1458-62. · 2.93 Impact Factor
  • Source
    ZheEn Zhao, Eam Khwang Teoh
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper describes a novel scheme to build a 3D Point Distribution Model (PDM) from a set of segmented training volumetric images. This approach is based on a deformable model algorithm to find correspondences across a set of surfaces/samples. It selects one sample as the template, and then deforms the template to approximate all other samples. These approximations carry the correspondences from the template to all other samples. The challenge is that a single template cannot guarantee accurate approximations to all other samples. The proposed solution is first to select the template sample, which brings the most accurate approximations to others among all samples. For each sample, which is not approximated accurately by the template, a “bridge” sample is chosen so that the bridge approximates accurately the current sample and the template approximates accurately to the bridge. The correspondences are then carried over from the template to the current sample via the “bridge”. A PDM is then constructed from the set of template’s approximations to all samples. This method is applied to construct four PDMs from 3D human brain Magnetic Resonance Images (MRIs). The four 3D PDMs constructed show considerable improvement on the approximation accuracy as compared to that constructed by adapting arbitrary templates. This improvement is important, as the approximation accuracy is the major concern of the deformable model-based approaches for the construction of PDMs.
    Image and Vision Computing 02/2008; · 1.58 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper we present improvements to a software application, named ROPtool, that aids in the timely and accurate detection and diagnosis of retinopathy of prematurity (ROP). ROP occurs in 68% of infants less than 1251 grams at birth, and it is a leading cause of blindness for prematurely born infants. The standard of care for its diagnosis is the subjective assessment of retinal vessel dilation and tortuosity. There is significant inter-observer variation in those assessments. ROPtool analyzes retinal images, extracts user-selected blood vessels from those images, and quantifies the tortuosity of those vessels. The presence of ROP is then gauged by comparing the tortuosity of an infant's retinal vessels with measures made from a clinical-standard image of severely tortuous retinal vessels. The presence of such tortuous retinal vessels is referred to as 'plus disease'. In this paper, a novel metric of tortuosity is proposed. From the ophthalmologist's point of view, the new metric is an improvement from our previously published algorithm, since it uses smooth curves instead of straight lines to simulate 'normal vessels'. Another advantage of the new ROPtool is that minimal user interactions are required. ROPtool utilizes a ridge traversal algorithm to extract retinal vessels. The algorithm reconstructs connectivity along a vessel automatically. This paper supports its claims by reporting ROC curves from a pilot study involving 20 retinal images. The areas under two ROC curves, from two experts in ROP, using the new metric to diagnose 'tortuosity sufficient for plus disease', varied from 0.86 to 0.91.
    Proceedings of SPIE - The International Society for Optical Engineering 01/2008; · 0.20 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: To prospectively determine if tortuosity assessment by a computer program (ROPtool) that traces retinal blood vessels and measures their tortuosity was more accurate than that of individual pediatric ophthalmologists. One hundred eighty-five high-quality RetCam images from premature infants were circulated to 3 retinopathy of prematurity (ROP) experts and 3 other pediatric ophthalmologists ("examiners") who graded the tortuosity in each quadrant as normal, pre-plus, or plus. These same images were analyzed using ROPtool. Using expert consensus as the standard, ROPtool's overall accuracy of 95% (175 of 185) for identifying tortuosity sufficient for plus disease was similar to that of examiner 1 (93%; 172 of 185; P = .50), examiner 2 (93%; 172 of 185; P = .50), and examiner 3 (91%; 168 of 185; P = .10). ROPtool's sensitivity of 97% (36 of 37) compared favorably with that of examiner 1 (65%; 24 of 37; P < .001), examiner 2 (70%; 26 of 37; P < .001), and examiner 3 (81%; 30 of 37; P = .06). Computer-assisted analysis of retinal images can potentially reduce subjectivity in the diagnosis of plus disease and optimize timing of follow-up and treatment for ROP.
    Archives of Ophthalmology 11/2007; 125(11):1523-30. · 4.49 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The accurate diagnosis of plus disease is critical to optimize the timing of laser treatment. Unfortunately, it is highly subjective and error-prone. "ROPtool" is a computer program that automatically traces retinal blood vessels and measures their tortuosity and dilation. Our aims were to pilot ROPtool, determine its reliability and validity, and establish appropriate numerical thresholds for plus and pre-plus disease. Twenty high-quality images of the posterior poles of premature infants were collected. Two of the authors (DKW and SFF) independently judged tortuosity and dilation separately as plus, pre-plus, or normal for each quadrant of each image. Disagreements were adjudicated, and the results were considered to be the standard for comparison to ROPtool. These two authors then separately used ROPtool to analyze the same 20 images. For determination of tortuosity sufficient for plus disease, ROPtool interuser agreement was 95% (19/20), compared with 90% (18/20) agreement by investigator judgment. Eye-level (2 MDs x 20 eyes) sensitivity of ROPtool in detecting tortuosity sufficient for plus disease averaged 95% (21/22) and specificity averaged 78% (14/18). Quadrant-level (2 MDs x 20 eyes x 4 quadrants) sensitivity averaged 85% (66/78) and specificity averaged 77% (63/82). A numeric threshold for pre-plus disease equal to 70% of the average tortuosity of the standard photograph of plus disease resulted in mean sensitivity of 89% (103/116) and mean specificity of 82% (36/44) in distinguishing quadrant-level tortuosity sufficient for pre-plus disease or worse from normal. ROPtool can reduce subjectivity and thereby enhance the evaluation of plus and pre-plus disease.
    Journal of American Association for Pediatric Ophthalmology and Strabismus 09/2007; 11(4):381-7. · 1.14 Impact Factor
  • Journal of American Association for Pediatric Ophthalmology and Strabismus 02/2007; 11(1):104-104. · 1.14 Impact Factor
  • Zheen Zhao, Eam Khwang Teoh
    [Show abstract] [Hide abstract]
    ABSTRACT: A 3D partitioned active shape model (PASM) is proposed in this paper to address the problems of 3D active shape models (ASM) brought by small training sets, which is usually the case in 3D applications. When numbers of training samples are limited, 3D ASMs tend to be restrictive, because the plausible area/allowable region spanned by relatively few eigenvectors cannot capture the full range of shape variability. 3D PASMs overcome this limitation by using a partitioned representation of ASM. Given a point distribution model, the mean mesh is partitioned into a group of small tiles. The statistical priors of tiles are estimated by applying principal component analysis to each tile, and the priors serve as constraints for corresponding tiles during deformations. To avoid the shape inconsistency introduced by the independent estimations between tiles, samples and deformed model points are projected as curves in one hyperspace, instead of point clouds in several hyperspaces. The deformed model points are then fitted into the allowable region of the model using a curve alignment scheme. The experiments on 3D human brain MRIs show that the 3D PASMs segment objects more accurately and are more robust to noise and low contrast in images than two other current active shape models. Furthermore, a study for the PASM's sensitivity to different initializations shows that PASMs perform stable when initialization positions change
    Ninth International Conference on Control, Automation, Robotics and Vision, ICARCV 2006, Singapore, 5-8 December 2006, Proceedings; 01/2006
  • Zheen Zhao, Eam Khwang Teoh
    [Show abstract] [Hide abstract]
    ABSTRACT: A 3D Partitioned Active Shape Model (PASM) is proposed in this paper to address the problems of 3D Active Shape Models (ASM) caused by the limited numbers of training samples, which is usually the case in 3D segmentation. When training sets are small, 3D ASMs tend to be restrictive, because the plausible area/allowable region spanned by relatively few eigenvectors cannot capture the full range of shape variability. 3D PASMs overcome this limitation by using a partitioned representation of the ASM. Given a Point Distribution Model (PDM), the mean mesh is partitioned into a group of small tiles. The statistical priors of tiles are estimated by applying Principal Component Analysis to each tile to constrain corresponding tiles during deformation. To avoid the inconsistency of shapes between tiles, samples are projected as curves in one hyperspace, instead of point clouds in several hyperspaces. The deformed model points are then fitted into the allowable region of the model by using a curve alignment scheme. The experiments on 3D human brain MRIs show that when the numbers of the training samples are limited, the 3D PASMs significantly improve the segmentation results as compared to 3D ASMs and 3D Hierarchical ASMs, which are the extension of the 2D Hierarchical ASM to the 3D case.
    Advances in Visual Computing, Second International Symposium, ISVC 2006, Lake Tahoe, NV, USA, November 6-8, 2006 Proceedings, Part I; 01/2006
  • [Show abstract] [Hide abstract]
    ABSTRACT: A 3D Partitioned Active Shape Model (PASM) is proposed in this paper to address the problems of the 3D Active Shape Models (ASM). When training sets are small. It is usually the case in 3D segmentation, 3D ASMs tend to be restrictive. This is because the allowable region spanned by relatively few eigenvectors cannot capture the full range of shape variability. The 3D PASM overcomes this limitation by using a partitioned representation of the ASM. Given a Point Distribution Model (PDM), the mean mesh is partitioned into a group of small tiles. In order to constrain deformation of tiles, the statistical priors of tiles are estimated by applying Principal Component Analysis to each tile. To avoid the inconsistency of shapes between tiles, training samples are projected as curves in one hyperspace instead of point clouds in several hyperspaces. The deformed points are then fitted into the allowable region of the model by using a curve alignment scheme. The experiments on 3D human brain MRIs show that when the numbers of the training samples are limited, the 3D PASMs significantly improve the segmentation results as compared to 3D ASMs and 3D Hierarchical ASMs.
    Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention. 02/2005; 8(Pt 1):221-8.
  • Source
    Zheen Zhao, Eam Khwang Teoh
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper presents a non-linear automated point distribution model building method (NAMB). In this algorithm, each training sample is registered to all other training samples by a non-linear corresponder. When landmarks are chosen by regarding all shapes of samples, a point distribution model (PDM) is constructed. The contribution of this paper is an accurate non-linear points correspondence detection algorithm and a framework that preserves shape details of training samples. In the experiment, the NAMB is compared with the manual model construction and other automated method to construct PDMs. The comparison illustrated that the NAMB is faster than manual method. At the same time, NAMB is more capable than the other automated method in preserving shape details.
    Control, Automation, Robotics and Vision Conference, 2004. ICARCV 2004 8th; 01/2005
  • Zheen Zhao, Eam Khwang Teoh
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper describes a novel framework to build 3D Point Distribution Model (PDM) from a set of segmented volumetric images. This method is based on a deformable model algorithm. Each training sample deforms to approximate all other training shapes. The training sample with best approximation results is then chosen as the template. Finally, the poor approximation results from this template are improved by the "bridge over" scheme, which deforms the template to approximate intermediate training shapes and then deforms the approximations to outliers. The method is applied to construct a 3D PDM of 20 human brain ventricles. The results show that the algorithm leads to more accurate representation than traditional framework. Also, the performance of the PDM of soft tissue is comparable with the PDM of bone structures by a previous method. The traditional framework of deformable model based approach selects the template arbitrarily and deforms the template to approximate training shapes directly. The limitation of the traditional framework is that the representation accuracy of the PDM entirely depends on the direct approximation. Moreover, the arbitrary template selection deteriorates the accuracy of the approximation. Our framework that features template selection and indirect approximation solves the shortcomings and improves the PDM representation accuracy. Furthermore, the "bridge over" framework could be used with any deformable model algorithm. In this sense, the method is a generic framework open to future investigation.
    Proceedings of SPIE - The International Society for Optical Engineering 01/2005; 5747:303-314. · 0.20 Impact Factor

Publication Stats

173 Citations
20.69 Total Impact Points

Institutions

  • 2008–2010
    • Duke University Medical Center
      • Department of Ophthalmology
      Durham, NC, United States
  • 2006–2009
    • Duke University
      • Department of Biomedical Engineering (BME)
      Durham, North Carolina, United States
  • 2005–2008
    • Nanyang Technological University
      • School of Electrical and Electronic Engineering
      Tumasik, Singapore
    • University of North Carolina at Chapel Hill
      North Carolina, United States