Analysis of retinal fundus images for grading of diabetic retinopathy severity

Diabetic retinopathy (DR) is a sight threatening complication due to diabetes mellitus that affects the retina. In this article, a computerised DR grading system, which digitally analyses retinal fundus image, is used to measure foveal avascular zone. A v-fold cross-validation method is applied to the FINDeRS database to evaluate the performance of the DR system. It is shown that the system achieved sensitivity of >84%, specificity of >97% and accuracy of >95% for all DR stages. At high values of sensitivity (>95%), specificity (>97%) and accuracy (>98%) obtained for No DR and severe NPDR/PDR stages, the computerised DR grading system is suitable for early detection of DR and for effective treatment of severe cases.     

Automatic segmentation of age-related macular degeneration in retinal fundus images

Every year an increasing number of people are affected by age-related macular degeneration (ARMD). Consequently, vast amount of information is accumulated in medical databases and manual classification of this information is becoming more and more difficult. Therefore, there is an increasing interest in developing automated evaluation methods to follow up the diseases. In this paper, we have presented an automatic method for segmenting the ARMD in retinal fundus images. Previously used direct segmentation techniques, generating unsatisfactory results in some cases, are more complex and costly than our inverse method. This is because of the fact that the texture of unhealthy areas of macula is quite irregular and varies from eye to eye. Therefore, a simple inverse segmentation method is proposed to exploit the homogeneity of healthy areas of the macula rather than unhealthy areas. This method first extracts healthy areas of the macula by employing a simple region growing method. Then, blood vessels are also extracted and classified as healthy regions. In order to produce the final segmented image, the inverse image of the segmented image is generated as unhealthy region of the macula. The performance of the method is examined on various qualities of retinal fundus images. The segmentation method without any user involvement provides over 90% segmentation accuracy. Segmented images with reference invariants are also compared with consecutive images of the same patient to follow up the changes in the disease.     

一种基于相位相关法及数学形态学方法的眼底血管图像自动拼

Precise mosaic techniques are essential for quantitative evaluation of retinal images to make early detection of fundus anomalies feasible. Opening of a gray-scale image by a gray-scale structuring element(SE) can generate a background image. Image mosaic was achieved by subtraction this background image from the original image and then applying a phase correlation method to find translation difference. Because of the accuracy characteristics of the phase correlation method and the speed of the FFT hardware, this new algorithm can work very fast and accurately, compared to conventional techniques. The method was also characterized by an outstanding robustness against correlated noise and disturbances, such as those encountered with nonuniform illumination. The results confirm the robustness of the chosen approach.     

Automated determination of cup-to-disc ratio for classification of glaucomatous and normal eyes on stereo retinal fundus images

Early diagnosis of glaucoma, which is the second leading cause of blindness in the world, can halt or slow the progression of the disease. We propose an automated method for analyzing the optic disc and measuring the cup-to-disc ratio (CDR) on stereo retinal fundus images to improve ophthalmologists' diagnostic efficiency and potentially reduce the variation on the CDR measurement. The method was developed using 80 retinal fundus image pairs, including 25 glaucomatous, and 55 nonglaucomatous eyes, obtained at our institution. A disc region was segmented using the active contour method with the brightness and edge information. The segmentation of a cup region was performed using a depth map of the optic disc, which was reconstructed on the basis of the stereo disparity. The CDRs were measured and compared with those determined using the manual segmentation results by an expert ophthalmologist. The method was applied to a new database which consisted of 98 stereo image pairs including 60 and 30 pairs with and without signs of glaucoma, respectively. Using the CDRs, an area under the receiver operating characteristic curve of 0.90 was obtained for classification of the glaucomatous and nonglaucomatous eyes. The result indicates potential usefulness of the automated determination of CDRs for the diagnosis of glaucoma.     

Multi-generational analysis and visualization of the vascular tree in 3D micro-CT images

Micro-CT scanners can generate large high-resolution three-dimensional (3D) digital images of small-animal organs, such as rat hearts. Such images enable studies of basic physiologic questions on coronary branching geometry and fluid transport. Performing such an analysis requires three steps: (1) extract the arterial tree from the image; (2) compute quantitative geometric data from the extracted tree; and (3) perform a numerical analysis of the computed data. Because a typical coronary arterial tree consists of hundreds of branches and many generations, it is impractical to perform such an integrated study manually. An automatic method exists for performing step (1), extracting the tree, but little effort has been made on the other two steps. We propose an environment for performing a complete study. Quantitative measures for arterial-lumen cross-sectional area, inter-branch segment length, branch surface area and others at the generation, inter-branch, and intra-branch levels are computed. A human user can then work with the quantitative data in an interactive visualization system. The system provides various forms of viewing and permits interactive tree editing for "on the fly" correction of the quantitative data. We illustrate the methodology for 3D micro-CT rat heart images.     

A simple algorithm for in vivo ocular fundus oximetry compensating for non-haemoglobin absorption and scattering

An algorithm is introduced for the compensation of the influence of non-haemoglobin absorption as well as tissue scattering on blood spectra used in optical oximetry at the ocular fundus. The in vivo measured spectra were corrected by a linear transformation in order to match the reference spectra of fully oxygenated and reduced blood, respectively, at three isosbestic points (522 nm, 569 nm and 586 nm). The oxygen saturation can then be determined at a wavelength showing a high contrast between oxygenated and reduced haemoglobin (e.g., 560 nm). Reflection measurements at blood flowing through cuvettes were used to validate the algorithm. The oxygen saturation values were compared to measurements of the same samples at a laboratory haemoximeter. The mean deviation was found to be 2.65%.     

Heterogeneous distribution of Weibel-Palade bodies and von Willebrand factor along the porcine vascular tree.

Vessels obtained from different levels of pig vascular tree were examined by transmission electron microscope, with the aim of determining whether or not their endothelial cells contain Weibel-Palade bodies (WPB). As these organelles are known to store the von Willebrand factor (vWF), a two-step immunogold labeling of this protein also was performed. Our results showed for the first time a heterogeneous distribution of WPB along the vascular tree of the normal pig: These structures were absent from the thoracic aorta, rare in the abdominal aorta, present in myocardial capillaries, and numerous in the inferior vena cava and pulmonary artery. Atypical WPB devoid of tubules were seen in all endothelial cells. The ultrastructural labeling of vWF demonstrated its presence only in the WPB, being absent in the subendothelium, and showed the same variation in its distribution along the vascular tree as for its storage organelle. Pigs homozygous for the von Willebrand disease were found to have only the atypical WPB, and do not express the vWF.     

Using a multi-agent system approach for microaneurysm detection in fundus images

ObjectiveMicroaneurysms represent the first sign of diabetic retinopathy, and their detection is fundamental for the prevention of vision impairment. Despite several research attempts to develop an automated system to detect microaneurysms in fundus images, none has shown the level of performance required for clinical practice. We propose a new approach, based on a multi-agent system model, for microaneurysm segmentation.Methods and materialsA multi-agent based approach, preceded by a preprocessing phase to allow construction of the environment in which agents are situated and interact, is presented. The proposed method is applied to two available online datasets and results are compared to other previously described approaches.ResultsMicroaneurysm segmentation emerges from agent interaction. The final score of the proposed approach was 0.240 in the Retinopathy Online Challenge.ConclusionsWe achieved competitive results, primarily in detecting microaneurysms close to vessels, compared to more conventional algorithms. Despite these results not being optimum, they are encouraging and reveal that some improvements may be made.     

On the automated definition of mobile target volumes from 4D-CT images for stereotactic body radiotherapy

Stereotactic body radiotherapy (SBRT) can be used to treat small lesions in the chest. A vacuum-based immobilization system is used in our clinic for SBRT, and a motion envelope is used in treatment planning. The purpose of this study is to automatically derive motion envelopes using deformable image registration of 4D-CT images, and to assess the effect of abdominal pressure on the motion envelopes. 4D-CT scans at ten phases were acquired prior to treatment for both free and restricted breathing using a vacuum-based immobilization system that includes an abdominal pressure pillow. To study the stability of the motion envelope over the course of treatment, a mid-treatment 4D-CT scan was obtained after delivery of the third fraction for two patients. The planning target volume excluding breathing motion (PTV(ex)) was defined on the image set at full exhalation phase and transformed into all other phases using displacement maps from deformable image registration. The motion envelope was obtained as the union of PTV(ex) masks of all phases. The ratios of the motion envelope to PTV(ex) volume ranged from 1.3 to 2.5. When pressure was applied, the ratios were reduced by as much as 29% compared to free breathing for some patients, but increased by up to 9% for others. The abdominal pressure pillow has more motion restriction effects on the anterior/inferior region of the lung. For one of the two patients for whom the 4D-CT scan was repeated at mid-treatment, the motion envelope was reproducible. However, for the other patient the tumor location and lung motion pattern significantly changed due to changes in the anatomy surrounding the tumor during the course of treatment, indicating that an image-guided approach to SBRT may increase the efficacy of this treatment.     

An automated segmentation method for three-dimensional carotid ultrasound images

We have developed an automated segmentation method for three-dimensional vascular ultrasound images. The method consists of two steps: an automated initial contour identification, followed by application of a geometrically deformable model (GDM). The formation of the initial contours requires the input of a single seed point by the user, and was shown to be insensitive to the placement of the seed within a structure. The GDM minimizes contour energy, providing a smoothed final result. It requires only three simple parameters, all with easily selectable values. The algorithm is fast, performing segmentation on a 336 x 352 x 200 volume in 25 s when running on a 100 MHz 9500 Power Macintosh prototype. The segmentation algorithm was tested on stenosed vessel phantoms with known geometry, and the segmentation of the cross-sectional areas was found to be within 3% of the true area. The algorithm was also applied to two sets of patient carotid images, one acquired with a mechanical scanner and the other with a freehand scanning system, with good results on both.     

System for the automated photothermal treatment of cutaneous vascular lesions

It is well known that the use of tightly focused continuous wave lasers can be an effective treatment of common telangiactasia. In general, the technique requires the skills of a highly dexterous surgeon using the aid of optical magnification. Due to the nature of this approach, it has proven to be largely impractical. To overcome this, we have developed an automated system that alleviates the strain on the user associated with the manual tracing method. The device makes use of high contrast illumination, simple monochromatic imaging, and machine vision to determine the location of blood vessels in the area of interest. The vessel coordinates are then used as input to a two-dimensional laser scanner via a near real-time feedback loop to target, track, and treat. Such mechanization should result in increased overall treatment success, and decreased patient morbidity. Additionally, this approach enables the use of laser systems that are considerably smaller than those currently used, and consequently the potential for significant cost savings. Here we present an overview of a proof-of-principle system, and results using examples involving in vivo imaging of human skin.     

The preprocessing of retinal images for the detection of fluorescein leakage

Images of the human retina are routinely used in clinical practice for the diagnosis and management of eye disease. Increased permeability of retinal blood vessels, which is a clinically significant feature, can be visualized with a process known as fluorescein angiography as leakage of fluorescence dye into the surrounding tissues. Analyses of such images can be quantified but significant degradation of images due to uneven illumination or occluded optical pathways is often incurred during image capture. We describe a procedure to restore fluorescein angiographic retinal images so that quantitative computation can be reliably performed. Analysis of the image acquisition system reveals that captured images are composed of two functions, one describing the true underlying image and the other the degradation incurred. These two functions are independent of one another and it is possible to estimate the degradation from an isolated captured image and restore it appropriately. Any leakage of fluorescein dye is then detected by analysing the restored angiographic sequence over time and finding areas of the image that do not have the usual decrease in fluorescence intensity.     

Evidence of the participation of peribiliary mast cells in regulation of the peribiliary vascular plexus along the intrahepatic biliary tree

Our pilot study disclosed that tryptase-positive mast cells (MC) were densely distributed around the intrahepatic bile ducts (peribiliary MC). In this study, the pathophysiologic roles of these MC were examined with respect to the microcirculation around the bile duct in 71 cases of histologically normal liver, 24 cases of chronic hepatitis, and 45 cases of liver cirrhosis. The tryptase-positive MC were very close to the microvessels of the peribiliary vascular plexus (PVP), which supply the intrahepatic biliary tree. The tryptase-positive MC were frequently found adjacent to vascular smooth muscle cells, including pericytes. The location of the tryptase-positive MC was confirmed by ultrastructural analysis. In cirrhosis, the numbers of both microvessels of PVP and peribiliary MC increased in parallel. Peribiliary MC were immunoreactive for endothelin 1 (ET-1), and were variably immunoreactive for histamine, chymase, inducible nitric oxide synthase (iNOS), and endothelin A and B (ET(A) and ET(B)) receptors, particularly in cirrhotic livers. On vascular endothelial cells of PVP, endothelial nitric oxide synthase (eNOS) and ET-1 were consistently detectable, and ET(A) receptors, ET(B) receptors, and iNOS were variably detectable. Pericytes of PVP expressed ET(A) and ET(B) receptors in addition to ET-1 and iNOS. Biliary epithelial cells also focally expressed iNOS, ET-1, and ET(A) and ET(B) receptors. These vasoactive substances were strongly expressed on the cellular components in cirrhotic liver. By in situ hybridization, iNOS mRNA signals were observed on iNOS-immunoreactive cell components, including peribiliary MC. These morphologic and immunohistochemical findings suggest that the cellular components displaying vasoactive substances in the milieu of the intrahepatic biliary tree are very dynamic in the vasoregulation of PVP in normal livers, even more so in cirrhosis, and that peribiliary MC exert local effects on the microcirculation of PVP, directly and indirectly.     

Long-term adaptation to prism-induced inversion of the retinal images

For 1 week, healthy human participants ( n=7) were devoid of normal vision by exposure to prism lenses that optically rotated their perceived world around the line of sight by 180 degrees. Adaptation to such prisms involved sustained and vigorous practice of the ability to redirect the unadapted efferent motor command; because prior to all visually guided movements, the to-be-executed efferent command was based on incorrect (prismatically reversed) spatial information. The time course of this sort of adaptation was systematically explored in Cooper-Shepard mental rotation (MR) tests and in naturalistic motor-tasks for the purpose of investigating whether mental rotations of the direction of the intended movement share common aspects with the process of MR. A control group ( n=7) intermittently exposed to the distorted spatial organization of the central visual field was studied in parallel. The main results were as follows: (a) the MR reaction times (RTs) day 1 with prisms appeared to be very similar to the normal RTs (day 1, no-prisms) with the one exception that subjects now responded within a prism (rotated) frame of spatial reference rather than within the environmentally upright. The visuomotor performance became grossly irregular and dysmetric. (b) The majority of the visuomotor adaptation functions began to level off on the 3rd day. (c) The increases in natural motor proficiency were accompanied by a systematic and noticeable decrease in magnitude of the MR Y-intercept obtained from the linear regression line calculated between each subject's RT and the various stimulus angles. MR slopes were stable through days 1-7 for both the experimental and control group. An increased correlation between rotational stimulus angle and RT suggested that the MR function also became progressively more tightly coupled to the stimulus angles. (d) Postadaptation measures of performance indicated the occurrence of selective and minimal adaptation in the natural motor tasks only. It is suggested that these results reflect an improved attentional (strategic) ability to replace incorrect (error producing) control signals with correct (error reducing) control signals. As a result, perceptual-motor start-up processes directly related to spatial coding and to the planning, initiation and correction of the intended direction of motor-or-mental movement improved while the subprocess ("stage") concerned with transformations of such movements remained unchanged. Visuomotor adaptation to inverting prisms engages, and thereby stimulates, a cortical system also invoked in the preparatory process of MR.     

Identification and red blood cell automated counting from blood smear images using computer-aided system

Red blood cell count plays a vital role in identifying the overall health of the patient. Hospitals use the hemocytometer to count the blood cells. Conventional method of placing the smear under microscope and counting the cells manually lead to erroneous results, and medical laboratory technicians are put under stress. A computer-aided system will help to attain precise results in less amount of time. This research work proposes an image-processing technique for counting the number of red blood cells. It aims to examine and process the blood smear image, in order to support the counting of red blood cells and identify the number of normal and abnormal cells in the image automatically. K-medoids algorithm which is robust to external noise is used to extract the WBCs from the image. Granulometric analysis is used to separate the red blood cells from the white blood cells. The red blood cells obtained are counted using the labeling algorithm and circular Hough transform. The radius range for the circle-drawing algorithm is estimated by computing the distance of the pixels from the boundary which automates the entire algorithm. A comparison is done between the counts obtained using the labeling algorithm and circular Hough transform. Results of the work showed that circular Hough transform was more accurate in counting the red blood cells than the labeling algorithm as it was successful in identifying even the overlapping cells. The work also intends to compare the results of cell count done using the proposed methodology and manual approach. The work is designed to address all the drawbacks of the previous research work. The research work can be extended to extract various texture and shape features of abnormal cells identified so that diseases like anemia of inflammation and chronic disease can be detected at the earliest.     

A modular supervised algorithm for vessel segmentation in red-free retinal images

In this paper, a supervised algorithm for vessel segmentation in red-free images of the human retina is proposed. The algorithm is modular and made up of two fundamental blocks. The optimal values of two algorithm parameters are found out by maximizing proper measures of performances (MOPs) able to evaluate from a quantitative point of view the results provided by the proposed algorithm. The choice of the MOP allows one to tailor the solution to the specific image features to be emphasized. The performances of the algorithm are compared with those of other methods described in the literature. The simulation results show a good trade-off between quality and processing speed times. For instance, in terms of the maximum average accuracy (MAA), K value, and specificity (SP), the best performance outcomes are 0.9587, 0.8069 and 0.9477, respectively.     

Morphological spot counting from stacked images for automated analysis of gene copy numbers by fluorescence in situ hybridization

Fluorescence in situ hybridization (FISH) is a molecular diagnostic technique in which a fluorescent labeled probe hybridizes to a target nucleotide sequence of deoxyribose nucleic acid. Upon excitation, each chromosome containing the target sequence produces a fluorescent signal (spot). Because fluorescent spot counting is tedious and often subjective, automated digital algorithms to count spots are desirable. New technology provides a stack of images on multiple focal planes throughout a tissue sample. Multiple-focal-plane imaging helps overcome the biases and imprecision inherent in single-focal-plane methods. This paper proposes an algorithm for global spot counting in stacked three-dimensional slice FISH images without the necessity of nuclei segmentation. It is designed to work in complex backgrounds, when there are agglomerated nuclei, and in the presence of illumination gradients. It is based on the morphological top-hat transform, which locates intensity spikes on irregular backgrounds. After finding signals in the slice images, the algorithm groups these together to form three-dimensional spots. Filters are employed to separate legitimate spots from fluorescent noise. The algorithm is set in a comprehensive toolbox that provides visualization and analytic facilities. It includes simulation software that allows examination of algorithm performance for various image and algorithm parameter settings, including signal size, signal density, and the number of slices.     

啮齿类动物视网膜血管系统的发育

目的 探讨啮齿类视网膜血管系统的发育。方法 选取生后不同年龄阶段的啮齿类动物各12只,用免疫组织化学染色、明胶墨汁灌流、
透射电镜显示其视网膜血管和血-视网膜屏障结构。结果 啮齿类视网膜血管系统是生后由视盘处开始,并以视乳头为中心呈放射状蔓延生长,
逐步覆盖整个视网膜,先形成视网膜内表面浅层血管网,继而伸入视网膜内核层的内、外侧边缘形成两个平行于浅层血管网的深层血管网;血
-视网膜屏障由幼稚内皮细胞、厚薄不均一的基膜层和多层较厚的胶质细胞终足成分构成,逐步发育为由内壁光滑连接紧密的内皮细胞、均一的
基膜、较薄的终足构成的成熟血-视网膜屏障。结论 啮齿类动物视网膜血管系统的发育成熟与视网膜细胞的分化成熟在时空上存在一致性;成
熟的血-视网膜屏障使视网膜具有一定的抗感染能力。     

Xenopus sonic hedgehog guides retinal axons along the optic tract

The role of classic morphogens such as Sonic hedgehog (Shh) as axon guidance cues has been reported in a variety of vertebrate organisms (Charron and Tessier‐Lavigne [ 2005 ] Development 132:2251–2262). In this work, we provide the first evidence that Xenopus sonic hedgehog (Xshh) signaling is involved in guiding retinal ganglion cell (RGC) axons along the optic tract. Xshh is expressed in the brain during retinal axon extension, adjacent to these axons in the ventral diencephalon. Retinal axons themselves express Patched 1 and Smoothened co‐receptors during RGC axon growth. Blocking Shh signaling causes abnormal ventral pathfinding, and targeting errors at the optic tectum. Misexpression of exogenous N‐Shh peptide in vivo also causes pathfinding errors. Retinal axons grown in culture respond to N‐Shh in a dose‐dependent manner, either by decreasing extension at lower concentrations, or retracting axons in the presence of higher doses. These data suggest that Shh signaling is required for normal RGC axon pathfinding and tectal targeting in the developing visual system of Xenopus. We propose that Shh serves as a ventral optic tract repellent that helps to define the caudal boundary for retinal axons in the diencephalon, and that this signaling is also required for initial target recognition at the optic tectum. Developmental Dynamics 239:2921–2932, 2010. © 2010 Wiley‐Liss, Inc.     

Real-time emulation of neural images in the outer retinal circuit

We describe a novel real-time system that emulates the architecture and functionality of the vertebrate retina. This system reconstructs the neural images formed by the retinal neurons in real time by using a combination of analog and digital systems consisting of a neuromorphic silicon retina chip, a field-programmable gate array, and a digital computer. While the silicon retina carries out the spatial filtering of input images instantaneously, using the embedded resistive networks that emulate the receptive field structure of the outer retinal neurons, the digital computer carries out the temporal filtering of the spatially filtered images to emulate the dynamical properties of the outer retinal circuits. The emulations of the neural image, including 128 x 128 bipolar cells, are carried out at a frame rate of 62.5 Hz. The emulation of the response to the Hermann grid and a spot of light and an annulus of lights has demonstrated that the system responds as expected by previous physiological and psychophysical observations. Furthermore, the emulated dynamics of neural images in response to natural scenes revealed the complex nature of retinal neuron activity. We have concluded that the system reflects the spatiotemporal responses of bipolar cells in the vertebrate retina. The proposed emulation system is expected to aid in understanding the visual computation in the retina and the brain.