Kivanc Kose

Entropy-Functional-Based Online Adaptive Decision Fusion Framework With Application to Wildfire Detection in Video

In this paper, an entropy-functional-based online adaptive decision fusion (EADF) framework is developed for image analysis and computer vision applications. In this framework, it is assumed that the compound algorithm consists of several subalgorithms, each of which yields its own decision as a real number centered around zero, representing the confidence level of that particular subalgorithm. Decision values are linearly combined with weights that are updated online according to an active fusion method based on performing entropic projections onto convex sets describing subalgorithms. It is assumed that there is an oracle, who is usually a human operator, providing feedback to the decision fusion method. A video-based wildfire detection system was developed to evaluate the performance of the decision fusion algorithm. In this case, image data arrive sequentially, and the oracle is the security guard of the forest lookout tower, verifying the decision of the combined algorithm. The simulation results are presented.

Filtered Variation method for denoising and sparse signal processing

We propose a new framework, called Filtered Variation (FV), for denoising and sparse signal processing applications. These problems are inherently ill-posed. Hence, we provide regularization to overcome this challenge by using discrete time filters that are widely used in signal processing. We mathematically define the FV problem, and solve it using alternating projections in space and transform domains. We provide a globally convergent algorithm based on the projections onto convex sets approach. We apply to our algorithm to real denoising problems and compare it with the total variation recovery.

Intraoperative imaging during Mohs surgery with reflectance confocal microscopy: initial clinical experience

Abstract. Mohs surgery for the removal of nonmelanoma skin cancers (NMSCs) is performed in stages, while being guided by the examination for residual tumor with frozen pathology. However, preparation of frozen pathology at each stage is time consuming and labor intensive. Real-time intraoperative reflectance confocal microscopy (RCM), combined with video mosaicking, may enable rapid detection of residual tumor directly in the surgical wounds on patients. We report our initial experience on 25 patients, using aluminum chloride for nuclear contrast. Imaging was performed in quadrants in the wound to simulate the Mohs surgeon’s examination of pathology. Images and videos of the epidermal and dermal margins were found to be of clinically acceptable quality. Bright nuclear morphology was identified at the epidermal margin and detectable in residual NMSC tumors. The presence of residual tumor and normal skin features could be detected in the peripheral and deep dermal margins. Intraoperative RCM imaging may enable detection of residual tumor directly on patients during Mohs surgery, and may serve as an adjunct for frozen pathology. Ultimately, for routine clinical utility, a stronger tumor-to-dermis contrast may be necessary, and also a smaller microscope with an automated approach for imaging in the entire wound in a rapid and controlled manner.

Video-mosaicing of reflectance confocal images for examination of extended areas of skin in vivo

With cellular-level resolution comparable to histology, reflectance confocal microscopy (RCM) imaging is a promising approach both for diagnosis of skin cancer in vivo with high sensitivity and specificity1,2, and for pre- and intra-operative detection of cancer margins to guide treatment.3–5 However, RCM images are limited to a field of view (FOV) of 1 mm -by- 1 mm, much smaller than the typical size of skin lesions. Many diagnostic features cannot be reliably identified in such small FOVs. Moreover, clinicians rely heavily on visual context of the surrounding tissue to perform diagnoses. Thus, larger areas must be imaged to evaluate cellular and morphologic features with high accuracy and repeatability. To address this concern, mosaicing approaches, which increase the FOV by acquiring a matrix of adjacent images and stitching them together to display a large area, have been developed for confocal microscopy6. In standard mosaicing, images are acquired while mechanically translating the microscope lens relative to the skin along pre-determined linear (straight-line) trajectories. This approach was implemented in the RCM scanner used in the cited studies1–5, and, in fact, is now routinely used on patients. However, the mechanics of translation limit speed and coverage to pre-selected small rectangular-shaped areas, currently up to 8 mm-by-8 mm, imaged in ~4.5 minutes. Coverage and speed could be increased, of course, with larger and faster mechanical translation systems, but would add significant size and cost to RCM scanners, and would certainly not be practical for routine use on patients. Miniaturized confocal endoscopes have been developed that allow the operator flexible control for imaging in vivo, without the constraints of mechanical translation7,8 Similar flexibility is now possible for imaging skin with the recent advent of smaller and miniaturized handheld confocal microscopes9,10,11. The operator manually moves the microscope along a desired curvilinear trajectory, with the lens gently pressed against the tissue, while acquiring a video sequence of images. Video microscopy enables the operator to choose the trajectory in real-time, allowing adaptive coverage of areas that can be selected in real-time during acquisition. Thus, an area with any shape and size may be rapidly imaged, without the previous constraints of straight-line trajectories and rectangular coverage. However, observing a video, by itself, merely as a time-sequence of small FOVs, does not readily provide the necessary visual context from the surrounding tissue. In this paper, we present results from an approach for computationally transforming such videos into mosaics that display the entire imaged area. Algorithms for video-mosaicing have been developed in the fields of computational photography and computer vision12, and their use has previously been reported for confocal endoscopic imaging7,8.We report here application of video-mosaicing to reflectance confocal images of human skin lesions and margins in vivo.

Projections onto convex sets (POCS) based optimization by lifting

Summary form only given. A new optimization technique based on the projections onto convex space (POCS) framework for solving convex and some non-convex optimization problems are presented. The dimension of the minimization problem is lifted by one and sets corresponding to the cost function are defined. If the cost function is a convex function in RN the corresponding set which is the epigraph of the cost function is also a convex set in RN+1. The iterative optimization approach starts with an arbitrary initial estimate in RN+1 and an orthogonal projection is performed onto one of the sets in a sequential manner at each step of the optimization problem. The method provides globally optimal solutions in total-variation, filtered variation, l1, and entropic cost functions. It is also experimentally observed that cost functions based on lp; p <; 1 may be handled by using the supporting hyperplane concept. The new POCS based method can be used in image deblurring, restoration and compressive sensing problems.

Automated Delineation of Dermal–Epidermal Junction in Reflectance Confocal Microscopy Image Stacks of Human Skin

Reflectance confocal microscopy (RCM) images skin noninvasively, with optical sectioning and nuclear-level resolution comparable with that of pathology. On the basis of the assessment of the dermal-epidermal junction (DEJ) and morphologic features in its vicinity, skin cancer can be diagnosed in vivo with high sensitivity and specificity. However, the current visual, qualitative approach for reading images leads to subjective variability in diagnosis. We hypothesize that machine learning-based algorithms may enable a more quantitative, objective approach. Testing and validation were performed with two algorithms that can automatically delineate the DEJ in RCM stacks of normal human skin. The test set was composed of 15 fair- and 15 dark-skin stacks (30 subjects) with expert labelings. In dark skin, in which the contrast is high owing to melanin, the algorithm produced an average error of 7.9±6.4 μm. In fair skin, the algorithm delineated the DEJ as a transition zone, with average error of 8.3±5.8 μm for the epidermis-to-transition zone boundary and 7.6±5.6 μm for the transition zone-to-dermis. Our results suggest that automated algorithms may quantitatively guide the delineation of the DEJ, to assist in objective reading of RCM images. Further development of such algorithms may guide assessment of abnormal morphological features at the DEJ.

Image Classification of Human Carcinoma Cells Using Complex Wavelet-Based Covariance Descriptors

Cancer cell lines are widely used for research purposes in laboratories all over the world. Computer-assisted classification of cancer cells can alleviate the burden of manual labeling and help cancer research. In this paper, we present a novel computerized method for cancer cell line image classification. The aim is to automatically classify 14 different classes of cell lines including 7 classes of breast and 7 classes of liver cancer cells. Microscopic images containing irregular carcinoma cell patterns are represented by subwindows which correspond to foreground pixels. For each subwindow, a covariance descriptor utilizing the dual-tree complex wavelet transform (DT-WT) coefficients and several morphological attributes are computed. Directionally selective DT-WT feature parameters are preferred primarily because of their ability to characterize edges at multiple orientations which is the characteristic feature of carcinoma cell line images. A Support Vector Machine (SVM) classifier with radial basis function (RBF) kernel is employed for final classification. Over a dataset of 840 images, we achieve an accuracy above 98%, which outperforms the classical covariance-based methods. The proposed system can be used as a reliable decision maker for laboratory studies. Our tool provides an automated, time- and cost-efficient analysis of cancer cell morphology to classify different cancer cell lines using image-processing techniques, which can be used as an alternative to the costly short tandem repeat (STR) analysis. The data set used in this manuscript is available as supplementary material through http://signal.ee.bilkent.edu.tr/cancerCellLineClassificationSampleImages.html.

Phase and TV Based Convex Sets for Blind Deconvolution of Microscopic Images

In this paper, two closed and convex sets for blind deconvolution problem are proposed. Most blurring functions in microscopy are symmetric with respect to the origin. Therefore, they do not modify the phase of the Fourier transform (FT) of the original image. As a result blurred image and the original image have the same FT phase. Therefore, the set of images with a prescribed FT phase can be used as a constraint set in blind deconvolution problems. Another convex set that can be used during the image reconstruction process is the Epigraph Set of Total Variation (ESTV) function. This set does not need a prescribed upper bound on the Total Variation (TV) of the image. The upper bound is automatically adjusted according to the current image of the restoration process. Both the TV of the image and the blurring filter are regularized using the ESTV set. Both the phase information set and the ESTV are closed and convex sets. Therefore they can be used as a part of any blind deconvolution algorithm. Simulation examples are presented.

Growth-Curve Modeling of Nevi With a Peripheral Globular Pattern.

Importance Although nevi with a peripheral rim of globules (peripheral globular nevi [PGN]) observed with dermoscopy are associated with enlarging melanocytic nevi, their actual growth dynamics remain unknown. Because change is a sensitive but nonspecific marker for melanoma, beginning to understand the growth patterns of nevi may improve the ability of physicians to differentiate normal from abnormal growth and reduce unnecessary biopsies. Objective To study the growth dynamics and morphologic evolution of PGN on dermoscopy. Design, Setting, and Participants A total of 84 participants with 121 PGN from September 1, 1999, through May 1, 2013, were identified retrospectively. Cohorts were recruited from the Memorial Sloan Kettering Cancer Center; Melanoma Unit of the Hospital Clinic, University of Barcelona; and Study of Nevi in Children. All 3 cohorts underwent longitudinal monitoring with serial dermoscopic imaging of their PGN. Data analysis was performed from May 1, 2014, through April 1, 2015. Main Outcomes and Measures Establishment of the natural growth curve of PGN. The secondary aim was to establish the median time to growth cessation in those PGN for which the size eventually stabilized and/or had begun to decrease during the study period. Results The median duration of follow-up was 25.1 (range, 2.0-114.4) months. Most of the nevi (116 [95.9%]) enlarged at some point during sequential monitoring. The rate of increase in the surface area of PGN varied among cohorts and ranged from -0.47 to 2.26 mm2/mo (mean rate, 0.25 [95% CI, 0.14-0.36] mm2/mo). The median time to growth cessation in the 26 PGN that stabilized or decreased in size (21.5%) was 58.6 months. All lesions changed in a symmetric manner and 91 (75.2%) displayed a decrease in the density of peripheral globules over time. Conclusions and Relevance Nevi displaying a peripheral globular pattern enlarged symmetrically with apparent growth cessation occurring during a span of 4 to 5 years. Our results reiterate the important concept that not all growth is associated with malignancy.

Denoising using projections onto the epigraph set of convex cost functions

A new denoising algorithm based on orthogonal projections onto the epigraph set of a convex cost function is presented. In this algorithm, the dimension of the minimization problem is lifted by one and feasibility sets corresponding to the cost function using the epigraph concept are defined. As the utilized cost function is a convex function in RN, the corresponding epigraph set is also a convex set in RN+1. The denoising algorithm starts with an arbitrary initial estimate in RN+1. At each step of the iterative denoising, an orthogonal projection is performed onto one of the constraint sets associated with the cost function in a sequential manner. The method provides globally optimal solutions for total-variation, ℓ1, ℓ2, and entropic cost functions.