Deep learning based retinal OCT segmentation.

We look at the recent application of deep learning (DL) methods in automated fine-grained segmentation of spectral domain optical coherence tomography (OCT) images of the retina. We describe a new method combining fully convolutional networks (FCN) with Gaussian Processes for post processing. We report performance comparisons between the proposed approach, human clinicians, and other machine learning (ML) such as graph based approaches. The approach is demonstrated on an OCT dataset consisting of mild non-proliferative diabetic retinopathy from the University of Miami. The method is shown to have performance on par with humans, also compares favorably with the other ML methods, and appears to have as small or smaller mean unsigned error (equal to 1.06), versus errors ranging from 1.17 to 1.81 for other methods, and compared with human error of 1.10.

Mitral valve closure prediction with 3-D personalized anatomical models and anisotropic hyperelastic tissue assumptions.

This study is concerned with the development of patient-specific simulations of the mitral valve that use personalized anatomical models derived from 3-D transesophageal echocardiography (3-D TEE). The proposed method predicts the closed configuration of the mitral valve by solving for an equilibrium solution that balances various forces including blood pressure, tissue collision, valve tethering, and tissue elasticity. The model also incorporates realistic hyperelastic and anisotropic properties for the valve leaflets. This study compares hyperelastic tissue laws with a quasi-elastic law under various physiological parameters, and provides insights into error sensitivity to chordal placement, allowing for a preliminary comparison of the influence of the two factors (chords and models) on error. Predictive errors show the promise of the method, yielding aggregate median errors of the order of 1 mm, and computed strains and stresses show good correspondence with those reported in prior studies.

A personalized mitral valve closure simulator.

We describe a method for performing modeling and simulation to predict the closure of the Mitral Valve (MV) using patient specific anatomical information derived from 3D Transesophageal Echocardiography (3D TEE). The ability to predict the MV closure behavior is an important step along the way of developing personalized simulation tools that would allow a surgeon to perform preoperative planning and decide between various MV repair options. While the MV is an important use case because of its relevance and prevalence among reconstructive cardiac interventions, the study described here can provide a blueprint to perform pre-operative planning for other cardiac surgical interventions. The method reported here exploits the Saint Venant-Kirchhoff elasticity model that is tuned to match empirical observations of the MV strainstress behavior. Using intraoperative 3D TEE data, the proposed simulator was evaluated over 10 test cases and resulted in mean prediction absolute error values of 1.81 mm.

Valvular closure prediction using anisotropic and hyperelastic tissue models and individualized anatomy derived from RT3DE.

We describe a method for modeling the closure of the Mitral Valve (MV) and to compute realistic strain and stresses in MV tissues. This informs preoperative planning by allowing a surgeon to evaluate various MV repairs options. The modeling method exploits individualized (patient-specific) anatomical structure recovered from real-time 3D echocardiography (RT3DE). This study utilizes hyperelastic models of the MV tissues and employs patient specific leaflets, chordal length assessment and annulus shapes. We report experiments on ten intraoperative test cases, where we compute strain and stresses using several different tissue models from MV empirical studies by May-Newman and Holzapfel.

Automatic screening of age-related macular degeneration and retinal abnormalities.

We describe a novel approach for screening retinal imagery to detect evidence of abnormalities. In this paper, we focus our efforts on age-related macular degeneration (AMD), a pathology that may often go undetected in the early or intermediate stages, and can lead to a neovascular form often resulting in blindness, if untreated. Our strategy for retinal anomaly detection is to employ a single class classifier applied to fundus imagery. We use a multiresolution locally-adaptive scheme that identifies both normal and anomalous regions within the retina. We do this by using a hybrid parametric/non-parametric characterization of the support of the probability distribution of normal retinal tissue in color and intensity feature space. We apply this approach to screen for evidence of AMD on a dataset of 66 healthy and pathological cases and found a detection sensitivity and specificity of 95% and 96%.

Patient-specific modeling of stress/strain for surgical planning and guidance.

We describe a method for performing modeling and simulation to predict the strain and stress experienced by tissues resulting from reconstructive cardiothoracic surgery. Stress computation is an important predictor of the quality and longevity of a repair and can therefore be used as guidance by a surgeon when deciding between various repair options. This paper uses the mitral valve repair as a use case because of its relevance and prevalence among reconstructive cardiac interventions. The modeling method presented here is informed by the patient specific anatomical structure recovered from real time 3D echocardiography. The method exploits hyperelastic models to infer realistic strain-stresses. We show through experiments using actual clinical data that results are in line with physiological expectations.

Computational hemodynamic modeling based on transesophageal echocardiographic imaging.

We address the problem of hemodynamic computational modeling in the left heart complex. The novelty of our approach lies in the exploitation of prior patient specific data resulting from image analysis of Transesophageal Echocardiographic Imagery (TEE). Kinematic and anatomical information in the form of left heart chambers and valve boundaries is recovered through a level-set-based user-in-the-loop segmentation on 2D TEE. The resulting boundaries in the TEE sequence are then interpolated to prescribe the motion displacements in a computational fluid dynamics (CFD) model implemented using Finite Element Modeling (FEM) applied on Arbitrary Lagrangian-Eulerian (ALE) meshes. Experimental results are presented.

Adaptive target detection in foliage-penetrating SAR images using alpha-stable models.

Detecting targets occluded by foliage in foliage-penetrating (FOPEN) ultra-wideband synthetic aperture radar (UWB SAR) images is an important and challenging problem. Given the different nature of target returns in foliage and nonfoliage regions and very low signal-to-clutter ratio in UWB imagery, conventional detection algorithms fail to yield robust target detection results. A new target detection algorithm is proposed that (1) incorporates symmetric alpha-stable (SalphaS) distributions for accurate clutter modeling, (2) constructs a two-dimensional (2-D) site model for deriving local context, and (3) exploits the site model for region-adaptive target detection. Theoretical and empirical evidence is given to support the use of the SalphaS model for image segmentation and constant false alarm rate (CFAR) detection. Results of our algorithm on real FOPEN images collected by the Army Research Laboratory are provided.

Image segmentation and labeling using the Polya urn model.

We propose a segmentation method based on Polya's (1931) urn model for contagious phenomena. A preliminary segmentation yields the initial composition of an urn representing the pixel. The resulting urns are then subjected to a modified urn sampling scheme mimicking the development of an infection to yield a segmentation of the image into homogeneous regions. This process is implemented using contagion urn processes and generalizes Polya's scheme by allowing spatial interactions. The composition of the urns is iteratively updated by assuming a spatial Markovian relationship between neighboring pixel labels. The asymptotic behavior of this process is examined and comparisons with simulated annealing and relaxation labeling are presented. Examples of the application of this scheme to the segmentation of synthetic texture images, ultra-wideband synthetic aperture radar (UWB SAR) images and magnetic resonance images (MRI) are provided.

An error resilient scheme for image transmission over noisy channels with memory.

This correspondence addresses the use of a joint source-channel coding strategy for enhancing the error resilience of images transmitted over a binary channel with additive Markov noise. In this scheme, inherent or residual (after source coding) image redundancy is exploited at the receiver via a maximum a posteriori (MAP) channel detector. This detector, which is optimal in terms of minimizing the probability of error, also exploits the larger capacity of the channel with memory as opposed to the interleaved (memoryless) channel. We first consider MAP channel decoding of uncompressed two-tone and bit-plane encoded grey-level images. Next, we propose a scheme relying on unequal error protection and MAP detection for transmitting grey-level images compressed using the discrete cosine transform (DCT), zonal coding, and quantization. Experimental results demonstrate that for various overall (source and channel) operational rates, significant performance improvements can be achieved over interleaved systems that do not incorporate image redundancy.

A spectral attentional mechanism tuned to object configurations.

This paper describes an attentional mechanism based on the interpretation of spectral signatures for detecting regular object configurations in areas of an image delineated using context information. The proposed global operator relies on the spectral analysis of edge structure and exploits spatial as well as frequency-domain constraints derived from known geometrical models of monitored objects. A decision theoretic method for learning acceptance detection regions is presented. Applications of this attentional mechanism are demonstrated for several aerial image interpretation tasks for attentional as well as recognition purposes. Specific examples are described for detecting vehicle formations (such as convoys), qualifying the geometry of detected formations, or monitoring the occupancy of regions of interest (such as parking areas, roads, or open areas). Experiments and sensitivity analysis results are reported.