Michelle Yan

Blood Vessel Classification into Arteries and Veins in Retinal Images

The prevalence of diabetes is expected to increase dramatically in coming years; already today it accounts for a major proportion of the health care budget in many countries. Diabetic Retinopathy (DR), a micro vascular complication very often seen in diabetes patients, is the most common cause of visual loss in working age population of developed countries today. Since the possibility of slowing or even stopping the progress of this disease depends on the early detection of DR, an automatic analysis of fundus images would be of great help to the ophthalmologist due to the small size of the symptoms and the large number of patients. An important symptom for DR are abnormally wide veins leading to an unusually low ratio of the average diameter of arteries to veins (AVR). There are also other diseases like high blood pressure or diseases of the pancreas with one symptom being an abnormal AVR value. To determine it, a classification of vessels as arteries or veins is indispensable. As to our knowledge despite the importance there have only been two approaches to vessel classification yet. Therefore we propose an improved method. We compare two feature extraction methods and two classification methods based on support vector machines and neural networks. Given a hand-segmentation of vessels our approach achieves 95.32% correctly classified vessel pixels. This value decreases by 10% on average, if the result of a segmentation algorithm is used as basis for the classification.

Using a Visual Discrimination Model for the Detection of Compression Artifacts in Virtual Pathology Images

A major issue in telepathology is the extremely large and growing size of digitized “virtual” slides, which can require several gigabytes of storage and cause significant delays in data transmission for remote image interpretation and interactive visualization by pathologists. Compression can reduce this massive amount of virtual slide data, but reversible (lossless) methods limit data reduction to less than 50%, while lossy compression can degrade image quality and diagnostic accuracy. “Visually lossless” compression offers the potential for using higher compression levels without noticeable artifacts, but requires a rate-control strategy that adapts to image content and loss visibility. We investigated the utility of a visual discrimination model (VDM) and other distortion metrics for predicting JPEG 2000 bit rates corresponding to visually lossless compression of virtual slides for breast biopsy specimens. Threshold bit rates were determined experimentally with human observers for a variety of tissue regions cropped from virtual slides. For test images compressed to their visually lossless thresholds, just-noticeable difference (JND) metrics computed by the VDM were nearly constant at the 95th percentile level or higher, and were significantly less variable than peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) metrics. Our results suggest that VDM metrics could be used to guide the compression of virtual slides to achieve visually lossless compression while providing 5-12 times the data reduction of reversible methods.

A local adaptive algorithm for microaneurysms detection in digital fundus images

Microaneurysms (MAs) detection is a critical step in diabetic retinopathy screening, since MAs are the earliest visible warning of potential future problems. A variety of thresholding based algorithms have been proposed for MAs detection in mass screening. Most of them process fundus images globally without a mechanism to take into account the local properties and changes. Their performance is often susceptible to nonuniform illumination and locations of MAs in different retinal regions. To keep sensitivity at a relatively high level, a low grey value threshold must be applied to the entire image globally, resulting in a much lower specificity in MAs detection. Therefore, post-processing steps, such as, feature extraction and classification, must be followed to improve the specificity at the cost of sensitivity. In order to address this problem, a local adaptive algorithm is proposed for automatic detection of MAs, where multiple subregions of each image are automatically analyzed to adapt to local intensity variation and properties, and furthermore prior structural features and pathology, such as, region and location information of vessel, optic disk and hard exudate are incorporated to further improve the detection accuracy. This algorithm effectively improves the specificity of MAs detection, without sacrificing the achieved sensitivity. It has potential to be used for automatic level-one grading of diabetic retinopathy screening.

A region based algorithm for vessel detection in retinal images

Accurate retinal blood vessel detection offers a great opportunity to predict and detect the stages of various ocular and systemic diseases, such as glaucoma, hypertension and congestive heart failure, since the change in width of blood vessels in retina has been reported as an independent and significant prospective risk factor for such diseases. In large-population studies of disease control and prevention, there exists an overwhelming need for an automatic tool that can reliably and accurately identify and measure retinal vessel diameters. To address requirements in this clinical setting, a vessel detection algorithm is proposed to quantitatively measure the salient properties of retinal vessel and combine the measurements by Bayesian decision to generate a confidence value for each detected vessel segment. The salient properties of vessels provide an alternative approach for retinal vessel detection at a level higher than detection at the pixel level. Experiments show superior detection performance than currently published results using a publicly available data set. More importantly, the proposed algorithm provides the confidence measurement that can be used as an objective criterion to select reliable vessel segments for diameter measurement.

Knowledge Based Image Enhancement Using Neural Networks

In this paper we combine the concept of adaptive filters with neural networks in order to be able to include high level knowledge about the contents of the image in the filtering process. Adaptive image enhancement algorithms often utilize low level knowledge like gradient information to guide filtering parameters. The advantage is that these filters do not need any specific knowledge and can thus be applied to a broad spectrum of images. However, for many problems this low level information is not sufficient to achieve good results. For example in medical imaging it is often very important that some features are preserved while others are suppressed. Usually these features cannot be distinguished by low level information. Therefore we propose a method to incorporate high level knowledge in the filtering process in order to adjust the parameters of any given filter thus creating a guided filter. We present a scheme for acquiring this high level knowledge which allows us to apply our method to all kinds of images using pattern recognition and special preprocessing techniques. The design of the guided filter itself is easy as for the high level knowledge only some sample pixels including their neighborhood and the desired parameters for these pixels are necessary

Automatic detection of pelvic lymph nodes using multiple MR sequences

A system for automatic detection of pelvic lymph nodes is developed by incorporating complementary information extracted from multiple MR sequences. A single MR sequence lacks sufficient diagnostic information for lymph node localization and staging. Correct diagnosis often requires input from multiple complementary sequences which makes manual detection of lymph nodes very labor intensive. Small lymph nodes are often missed even by highly-trained radiologists. The proposed system is aimed at assisting radiologists in finding lymph nodes faster and more accurately. To the best of our knowledge, this is the first such system reported in the literature. A 3-dimensional (3D) MR angiography (MRA) image is employed for extracting blood vessels that serve as a guide in searching for pelvic lymph nodes. Segmentation, shape and location analysis of potential lymph nodes are then performed using a high resolution 3D T1-weighted VIBE (T1-vibe) MR sequence acquired by Siemens 3T scanner. An optional contrast-agent enhanced MR image, such as post ferumoxtran-10 T2*-weighted MEDIC sequence, can also be incorporated to further improve detection accuracy of malignant nodes. The system outputs a list of potential lymph node locations that are overlaid onto the corresponding MR sequences and presents them to users with associated confidence levels as well as their sizes and lengths in each axis. Preliminary studies demonstrates the feasibility of automatic lymph node detection and scenarios in which this system may be used to assist radiologists in diagnosis and reporting.

Spatio-temporal analysis for automatic contrast injection detection on angiography during trans-catheter aortic valve implantation

Registration of 3-D aortic model onto X-ray images provides anatomical details for optimal valve deployment in Trans-catheter aortic valve implantation (TAVI) procedures. Fast and automatic contrast detection in the aortic root in fluoroscopic sequences facilitates a seamless workflow by triggering 2-D/3-D registration automatically when necessary. In this paper, we propose an integrated spatial and temporal analysis on fluoroscopic sequences to improve the performance of the contrast feature curve based method in our previous work [4]. For sequences with a high likelihood of containing contrasted aorta, a cascaded classifier is used to separate the contrasted aorta from the contrasted balloons. A local training procedure is then performed in sequences with contrasted aorta to identify the range of well-contrasted frames. Next, a multi-layer sparse Gabor feature based classifier is adopted to recognize faint contrast. Experiments on 69 sequences acquired during TAVI procedures achieved the correct classification rate of 12/12 for non-contrasted sequences and 56/57 for contrasted sequences.

Automatic detection of contrast injection on fluoroscopy and angiography for image guided trans-catheter aortic valve implantations (TAVI)

Presentation of detailed anatomical structures via 3-D models helps navigation and deployment of the prosthetic valve in TAVI procedures. Fast and automatic contrast detection in the aortic root on X-ray images facilitates a seamless workflow to utilize the 3-D models by triggering 2-D/3-D registration automatically when motion compensation is needed. In this paper, we propose a novel method for automatic detection of contrast injection in the aortic root on fluoroscopic and angiographic sequences. The proposed method is based on histogram analysis and likelihood ratio test, and is robust to variations in the background, the density and volume of the injected contrast, and the size of the aorta. The performance of the proposed algorithm was evaluated on 26 sequences from 5 patients and 3 clinical sites, with 16 out of 17 contrast injections correctly detected and zero false detections. The proposed method is of general form and can be extended for detection of contrast injection in other organs and/or applications.

Visually lossless compression of breast biopsy virtual slides for telepathology

A major issue in telepathology is the extreme size of digitized slides, which require several gigabytes of storage and cause significant delays in image delivery to pathologists. We investigated the utility of a visual discrimination model (VDM) to predict bit rates for visually lossless JPEG2000 compression of breast biopsy virtual slides. Visually lossless bit rates were determined experimentally with human observers. VDM metrics computed for those bit rates were nearly constant, suggesting that VDMs could be used to achieve visually lossless image quality while providing about four times the data reduction of reversible compression.

Integrated spatiotemporal analysis for automatic contrast agent inflow detection on angiography and fluoroscopy during transcatheter aortic valve implantation.

PURPOSE In this paper, the authors propose an integrated spatial and temporal analysis for automatic detection of contrast agent inflow at the aortic root on fluoroscopic and angiographic sequences during transcatheter aortic valve implantation procedures as a means to automatically trigger registration of 3D aortic models. METHODS The proposed contrast agent inflow detection method is based on a contrast feature curve, calculated using histogram analysis and a likelihood ratio test. Several image preprocessing steps are performed to enhance the properties of the contrast feature curve. For sequences with a dominant peak on its contrast feature curve, a cascaded classifier is then applied to differentiate the contrast-enhanced aorta from contrast-enhanced balloons. Finally, a multilayer classifier based on sparse Gabor features is used to recognize sequences containing a faint contrast-enhanced aorta. RESULTS The algorithm was evaluated using 105 sequences consisting of more than 12,000 frames, and achieved a detection accuracy of 99.1% (100% sensitivity and 98.5% specificity). The computation time for a typical sequence of 150 frames was ≈ 1 s on a single-core Dell PC with a 1 GHz Intel Xeon processor and 2 GB of RAM. CONCLUSIONS The authors developed a novel, automatic method for contrast agent inflow detection on x-ray sequences. With the achieved efficiency and accuracy, the proposed method is potentially feasible for clinical use as it facilitates a seamless workflow in utilizing patient-specific 3D models to provide anatomical details during transcatheter aortic valve implantation procedures.