- Histological Stain Evaluation for Machine Learning Applications
Authors: Jimmy Azar, Christer Busch (1), Ingrid Carlbom
(1) Dept. Immunology, Genetics and Pathology
Conference: Workshop on: Histopathology Image Analysis (HIMA), 15th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI), Nice, France
Publisher: Journal of Pathology Informatics
Editors: Anant Madabhushi, Metin Gurcan, Nasir Rajpoot
Abstract: Machine learning and image analysis are increasingly important in pathology applications, such as for automatic analysis of histological tissue samples. Pathologists rely on multiple, contrasting stains to analyze tissue samples, but histological stains were developed for visual analysis and are not always ideal for automatic analysis. We present a methodology for evaluating histological stains in terms of their classification and clustering efficacy with the aim of improving segmentation and color decomposition. We evaluate the stains for both supervised and unsupervised classification of stain/tissue combinations. For supervised classification we measure the nonlinear support vector machines error rate and for unsupervised classification we use the Rand index and the F-measure to assess the clustering results of a Gaussian mixture model based on expectation-maximization. Finally we investigate class separability measures based on scatter criteria. We demonstrate that for a specific tissue type the same stains perform best according to all measures.
- Automated Classification of Immunostaining Patterns in Breast Tissue from the Human Protein Atlas
Authors: Swamidoss Issac Niwas, Andreas Kårsnäs, Virginie Uhlmann (2,3), P. Palanisamy (4), Caroline Kampf (5), Martin Simonsson (1), Carolina Wählby (1), Robin Strand
(1) SciLifeLab, UU
(2) Imaging Platform, Broad Institute of Harvard and MIT, Cambridge, Massachusetts MA, USA
(3) Biomedical Imaging Group, École Polytechnique Fédérale de Lausanne (EPFL), Switzerland
(4) Dept. of Electronics and Communication Engineering (ECE), National Institute of Technology (NIT), Tiruchirappalli, India
(5) Dept. of Immunology, Genetics and Pathology,UU
Conference: Workshop on: Histopathology Image Analysis (HIMA), 15th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI), Nice, France
Abstract: Background: The Human Protein Atlas (HPA) is an effort to map the location of all human proteins (http://www.proteinatlas.org/ ). It contains a large number of histological images of sections from human tissue. Tissue micro arrays are imaged by a slide scanning microscope, and each image represents a thin slice of a tissue core with a dark brown antibody specific stain and a blue counter stain. When generating antibodies for protein profiling of the human proteome, an important step in the quality control is to compare staining patterns of different antibodies directed towards the same protein. This comparison is an ultimate control that the antibody recognizes the right protein. In this paper, we propose and evaluate different approaches for classifying sub-cellular antibody staining patterns in breast tissue samples.
Methods and Material: The proposed methods include the computation of various features including gray level co-occurrence matrix (GLCM) features, complex wavelet co-occurrence matrix (CWCM) features and WND-CHARM-inspired features. The extracted features are used into two different multivariate classifiers (SVM and LDA classifier). Before extracting features, we use color deconvolution to separate different tissue components, such as the brownly stained positive regions and the blue cellular regions, in the immuno-stained TMA images of breast tissue.
Results: Good results have been obtained by using the combinations of GLCM and wavelets and texture features, edge features, histograms, transforms, etc. (WND-CHARM). The proposed complex wavelet features and the WND-CHARM features have accuracy similar to that of a human expert.
Conclusions: Both human experts and the proposed automated methods have difficulties discriminating between nuclear and cytoplasmic staining patterns. This is to a large extent due to mixed staining of nucleus and cytoplasm. Methods for quantification of staining patterns in histopathology have many applications, ranging from antibody quality control to tumour grading.
- Smart Paint - A New Interactive Segmentation Method Applied to MR Prostate Segmentation
Authors: Filip Malmberg, Robin Strand, Joel Kullberg (1), Richard Nordenskjöld (1), Ewert Bengtsson
(1) Dept. of Radiology, Oncology and Radiation Science, UU
Conference: Workshop on: Prostate MR Image Segmentation Grand Challenge (PROMISE'12), 15th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI), Nice, France
Abstract: This paper describes a general method for interactive segmentation, Smart Paint. The user interaction is inspired by the way an airbrush is used, objects are segmented by "sweeping" with the mouse cursor in the image. The user adds or removes details in 3D by the proposed segmentation tool and the user interface shows the segmentation result in 2D slices through the object. We use the novel method for prostate segmentation in transversal T2-weighted MR images from multiple centers and vendors and with differences in scanning protocol.
The method was evaluated on the training set obtained from http://promise12.grand-challenge.org. In the first round, all 50 volumes were segmented and the mean of Dice's coefficient was 0.82 with standard deviation 0.09. In a second round, the first 30 volumes were re-segmented by the same user and the result was slightly improved - Dice's coefficient 0.86 0.05 was obtained. For the training data, the mean time to segment a volume was 3 minutes and 30 seconds.
The proposed method is a generic tool for interactive image segmentation and this paper illustrates that it is well-suited for prostate segmentation.
- Cluster Detection in Cytology Images Using the Cellgraph Method
Authors: P. S. Chandran, N. B. Byju, R. U. Deepak, R. Kumar Rajesh, S. Sudhamony, Patrik Malm, Ewert Bengtsson
Conference: International Symposium on Information Technology in Medicine and Education (ITME), Hokodate, Hokkaido, Japan, number 2, pp 923-927
Abstract: Automated cervical cancer detection system is primarily based on delineating the cell nuclei and analyzing their textural and morphometric features for malignant characteristics. The presence of cell clusters in the slides have diagnostic value, since malignant cells have a greater tendency to stick together forming clusters than normal cells. However, cell clusters pose difficulty in delineating nucleus and extracting features reliably for malignancy detection in comparison to free lying cells. LBC slide preparation techniques remove biological artifacts and clustering to some extent but not completely. Hence cluster detection in automated cervical cancer screening becomes significant. In this work, a graph theoretical technique is adopted which can identify and compute quantitative metrics for this purpose. This method constructs a cell graph of the image in accordance with the Waxman model, using the positional coordinates of cells. The computed graph metrics from the cell graphs are used as the feature set for the classifier to deal with cell clusters. It is a preliminary exploration of using the topological analysis of the cellgraph to cytological images and the accuracy of classification using SVM showed that the results are well suited for cluster detection.
- Adaptive Structuring Elements Based on Salience Information
Authors: Vladimir Curic, Cris L. Luengo Hendriks
Conference: International Conference on Computer Vision and Graphics (ICCVG), Warsaw, Poland, volume 7594 of Lecture Notes in Computer Science, pp 321-328
Publisher: Springer Berlin/Heidelberg
Editors: L. Bolc, K. Wojciechowski, R. Tadeusiewicz, L.J. Chmielewski
Abstract: Adaptive structuring elements modify their shape and size according to the image content and may outperform fixed structuring elements. Without any restrictions, they suffer from a high computational complexity, which is often higher than linear with respect to the number of pixels in the image. This paper introduces adaptive structuring elements that have predefined shape, but where the size is adjusted to the local image structures. The size of adaptive structuring elements is determined by the salience map that corresponds to the salience of the edges in the image, which can be computed in linear time. We illustrate the difference between the new adaptive structuring elements and morphological amoebas. As an example of its usefulness, we show how the new adaptive morphological operations can isolate the text in historical documents.
- Towards User-Guided Quantitative Evaluation of Wrist Fractures in CT Images
Authors: Johan Nysjö, Albert Christersson (1), Filip Malmberg, Ida-Maria Sintorn, Ingela Nyström
(1) Dept. of Surgical Sciences, UU
Conference: International Conference on Computer Vision and Graphics (ICCVG), Warsaw, Poland, volume 7594 of Lecture Notes in Computer Science, pp 204-2011
Publisher: Springer Berlin/Heidelberg
Editors: L. Bolc, R. Tadeusiewicz, L.J. Chmielewski, K. Wojciechowski
Abstract: The wrist is the most common location for long-bone fractures in humans. To evaluate the healing process of such fractures, it is of interest to measure the fracture displacement, particularly the angle between the joint line and the long axis of the fractured long bone. We propose to measure this angle in 3D computed tomography (CT) images of fractured wrists. As a first step towards this goal, we here present a fast and precise semi-automatic method for determining the long axis of the radius bone in CT images. To facilitate user interaction in 3D, we utilize stereo graphics, head tracking, 3D input, and haptic feedback.
- An Efficient Preconditioner and a Modified RANSAC for Fast and Robust Feature Matching
Authors: Anders Hast, Andrea Marchetti (1)
(1) CNR, Institute of Informatics and Telematics, Pisa, Italy
Conference: International Conferences in Central Europe on Computer Graphics, Visualization and Computer Vision (WSCG'12), Communcations Paper, pp 11-18
Abstract: Standard RANSAC does not perform very well for contaminated sets, when there is a majority of outliers. We present a method that overcomes this problem by transforming the problem into a 2D position vector space, where an ordinary cluster algorithm can be used to find a set of putative inliers. This set can then easily be handled by a modified version of RANSAC that draws samples from this set only and scores using the entire set. This approach works well for moderate differences in scale and rotation. For contaminated sets the increase in performance is in several orders of magnitude. We present results from testing the algorithm using the Direct Linear Transformation on aerial images and photographs used for panographs.
- Putative Match Analysis : A Repeatable Alternative to RANSAC for Matching of Aerial Images
Authors: Anders Hast, Andrea Marchetti (1)
(1) Institute of Informatics and Telematics, Pisa, Italy
Conference: International Conference on Computer Vision Theory and Applications (VISAPP), Rome, Italy, Volume 2, pp 341-344
Publisher: SciTePress
Editors: Gabriela Csurka, José Braz
Abstract: One disadvantage with RANSAC is that it is based on randomness and will therefore often yield a different set of inliers in each run, especially if the dataset contains a large number of outliers. A repeatable algorithm for finding both matches and the homography is proposed, which in our case is used for image stitching and the obtained points are also used for georeferencing. This algorithm will yield the same set of matches every time and is therefore a useful tool when trying to evaluate other algorithms involved and their parameters. Moreover a refining step is proposed that finds the best matches depending on what geometric transformation is used, which also can be utilized as a refining step for RANSAC.
- Rigid Template Registration in MET Images Using CUDA
Authors: Lennart Svensson, Johan Nysjö, Anders Brun, Ingela Nyström, Ida-Maria Sintorn
Conference: International Conference on Computer Vision Theory and Applications (VISAPP), Rome, Italy, Volume 2, pp 418-422
Publisher: SciTePress
Editors: Gabriela Csurka, José Braz
Abstract: Rigid registration is a basic tool in many applications, especially in Molecular Electron Tomography (MET), and also in, e.g., registration of rigid implants in medical images and as initialization for deformable registration. As MET volumes have a low signal to noise ratio, a complete search of the six-dimensional (6D) parameter space is often employed. In this paper, we describe how rigid registration with normalized cross-correlation can be implemented on the GPU using NVIDIA's parallel computing architecture CUDA. We compare the performance to the Colores software and two Matlab implementations, one of which is using the GPU accelerated JACKET library. With well-aligned padding and using CUDA, the performance increases by an order of a magnitude, making it feasible to work with three-dimensional fitness landscapes, here denoted scoring volumes, that are generated on the fly. This will eventually enable the biologists to interactively register macromolecule chains in MET volumes piece by piece.
- A Novel Algorithm for Computing Riemannian Geodesic Distance in Rectangular 2D Grids
Authors: Ola Nilsson (1), Martin Reimers, Ken Museth, Anders Brun
(1) Dept. of Science and Technology, Linköping University
(2) Dept. of Informatics, University of Oslo, Norway
Conference: Advances in Visual Computing : 8th International Symposium (ISVC), Rethymnon, Crete, Greece, Revised Selected Papers, Part II, volume 7432 of Lecture Notes in Computer Science, pp 265-274
Publisher: Springer
Abstract: We present a novel way to efficiently compute Riemannian geodesic distance over a two-dimensional domain. It is based on a previ- ously presented method for computation of geodesic distances on surface meshes. Our method is adapted for rectangular grids, equipped with a variable anisotropic metric tensor. Processing and visualization of such tensor fields is common in certain applications, for instance structure ten- sor fields in image analysis and diffusion tensor fields in medical imaging. The included benchmark study shows that our method provides signif- icantly better results in anisotropic regions and is faster than current stat-of-the-art solvers. Additionally, our method is straightforward to code; the test implementation is less than 150 lines of C++ code.
- Rendering Stiffness with a Prototype Haptic Glove Actuated by an Integrated Piezoelectric Motor
Authors: Pontus Olsson, Stefan Johansson (1), Fredrik Nysjö, Ingrid Carlbom
(1) Dept. of Engineering Sciences UU
Conference: Haptics: Perception, Devices, Mobility, and Communication: Part I, EuroHaptics, Tampere, Finland, volume 7282 of Lecture Notes in Computer Science, pp 361-372
Publisher: Springer Berlin/Heidelberg
Abstract: Bi-directional haptic devices incorporate both sensors and actuators. While small and compact sensors are readily available, actuators in haptic interfaces require a significant volume to produce needed forces. With many actuated degrees of freedom, the mass and size of the actuators become a problem in devices such as haptic gloves. Piezo-technology offers the possibility of compact actuators which can be controlled with high accuracy. We describe a prototype admittance-type haptic device for the hand with a compact integrated piezoelectric motor. The current implementation provides one degree of freedom, but it could be extended with more motors for additional degrees of freedom. We demonstrate both the accuracy with which the device can reproduce force-displacement responses of non-linear elastic material stiffness and the device's fast and stable response to an applied load.
- Physically Co-Located Haptic Interaction with 3D Displays
Authors: Pontus Olsson, Fredrik Nysjö, Stefan Seipel, Ingrid Carlbom
Conference: Haptics Symposium (HAPTICS), pp 267-272
Publisher: IEEE Computer Society
Abstract: Studies indicate that haptic interaction with a computer generated virtual scene may become more intuitive by aligning (co-locating) the visual and haptic workspaces so that the visual and haptic feedback coincide as they do in the real world. Co-located haptics may gain importance when more advanced haptic interfaces, such as high-fidelity whole hand devices, become available. We describe a user study that investigates the pros and cons with physically co-located versus non-collocated haptics on two different display types: a commercial half-transparent mirror 3D display with shutter glasses and a prototype autostereoscopic display based on a Holographic Optical Element (HOE). We use two accuracy tasks with spatial accuracy as the dependent variable and one manipulation task with time as the dependent variable. The study shows that on both displays co-location significantly improves completion time in the manipulation task. However, the study shows that co-location does not improve the accuracy in the spatial accuracy tasks.
- Solving Combined Geospatial Tasks Using 2D and 3D Bar Charts
Authors: Stefan Seipel, L. Carvalho
Conference: 16th International Conference on Information Visualisation (IV), Montpellier, France, pp 157-163
Abstract: This paper presents a user study that investigates 2D and 3D visualizations of bar charts in geographic maps. The task to be solved by the participants in this study required estimation of the ratio of two different spatial distance measures and relative ranking among potential candidates. The results of this experiment show that subjects were equally fast and accurate when using both the 2D and 3D visualizations. Visual discomfort was reported by almost half of the test population, but performance was not affected. Our study also showed that frequent game players did not benefit more from a 3D visualization than inexperienced game-players.
- Seeded Segmentation Based on Object Homogeneity
Authors: Filip Malmberg, Robin Strand, Richard Nordenskjöld (1), Joel Kullberg (1)
(1) Dept. of Radiology, Oncology and Radiation Science, UU
Conference: 21st International Conference on Pattern Recognition (ICPR), Tsukuba, Japan
Abstract: Seeded segmentation methods attempt to solve the segmentation problem in the presence of prior knowledge in the form of a partial segmentation, where a small subset of the image elements (seed-points) have been assigned correct segmentation labels. Common for most of the leading methods in this area is that they seek to find a segmentation where the boundaries of the segmented regions coincide with sharp edges in the image. Here, we instead propose a method for seeded segmentation that seeks to divide the image into areas of homogeneous pixel values. The method is based on the computation of minimal cost paths in a discrete representation of the image, using a novel path-cost function. The utility of the proposed method is demonstrated in a case study on segmentation of white matter hyper intensities in MR images of the human brain.
- Comparison of Restoration Quality on Square and Hexagonal Grids using Normalized Convolution
Authors: Elisabeth Linnér, Robin Strand
Conference: 21st International Conference on Pattern Recognition (ICPR), Tsukuba, Japan
Abstract: Normalized convolution can be used to restore information that has been lost from an image, such as dead pixels, using the remaining information, and ignoring the incorrect pixels. It is known that the representation quality of an image consisting of a given number of pixels depends on how these pixels are distributed. In this paper, we investigate whether the ability to restore information using normalized convolution is affected by the sampling grid of the image. We compare square and hexagonal grids, and find that, in general, more pixels can be restored in hexagonal grids.
- The Vectorial Minimum Barrier Distance
Authors: Andreas Kårsnäs, Robin Strand, Punam K. Saha (1)
(1) Dept. of Electrical and Computer Engineering and the Dept. of Radiology, The University of Iowa, Iowa City, IA, USA
Conference: 21st International Conference on Pattern Recognition (ICPR), Tsukuba, Japan
Abstract: We introduce the vectorial Minimum Barrier Distance (MBD), a method for computing a gray-weighted distance transform while also incorporating information from vectorial data. Compared to other similar tools that use vectorial data, the proposed method requires no training and does not assume having only one background class. We describe a region growing algorithm for computing the vectorial MBD efficiently.
The method is evaluated on two types of multi-channel images: color images and textural features. Different path-cost functions for calculating the multi-dimensional path-cost distance are also compared.
The results show that by combining multi-channel images into vectorial information the performance ofthe vectorial MBD segmentation is improved compared to when one channel is used. This implies that the method can be a good way of incorporating multi-channel information in interactive segmentation.
- Regional Zernike Moments for Texture Recognition
Authors: Ida-Maria Sintorn, Gustaf Kylberg
Conference: 21st International Conference on Pattern Recognition (ICPR), Tsukuba, Japan, pp 1635-1638
Abstract: Zernike moments are commonly used in pattern recognition but are not suited for texture analysis. In this paper we introduce regional Zernike moments (RZM) where we combine the Zernike moments for the pixels in a region to create a measure suitable for texture analysis. We compare our proposed measures to texture measures based on Gabor filters, Haralick co-occurrence matrices and local binary patterns on two different texture image sets, and show that they are noise insensitive and very well suited for texture recognition.
- Graph Based Line Segmentation on Cluttered Handwritten Manuscripts
Authors: Fredrik Wahlberg, Anders Brun
Conference: 21st International Conference on Pattern Recognition (ICPR), Tsukuba, Japan, pp 1635-1638
Abstract: We propose a two phase line segmentation method for handwritten pre-modern densely writ- ten manuscripts. The proposed method combines the robustness of projection based methods with the flexibility of graph based methods. The result are cut-outs of the image containing each text line. Overlapping characters, help lines and degradation can create foreground elements spanning several lines that are hard to separate. We treat the problem of finding a cut through the text line separation as a graph optimization problem, which allows for flexible separation of entangled components.
The proposed method has been tested on two medieval sources with satisfying results. A comparison to similar methods, using standard metrics, is presented.
- Identifying all Individuals in a Honeybee Hive : Progress Towards Mapping all Social Interactions
Authors: Cris L. Luengo Hendriks, Zi Quan Yu, Antoine Lecocq (1,2), Teatske Bakker(1), Barbara Locke (1), Olle Terenius (1)
(1) Dept. of Ecology, SLU, Uppsala
(2) Currently at: Department of Agriculture & Ecology, University of Copenhagen
Conference: Workshop on: Visual observation and analysis of animal and insect behavior, 21st International Conference on Pattern Recognition (ICPR), Tsukuba, Japan, pp 5-8
Editors: R. Fisher, J. Hallam, B. Boom
Abstract: Here we present work in progress towards a fully automatic system that monitors a honeybee hive over many days, deriving information on the position and velocity of each bee, and detecting and identifying each instance of a social interaction. Each bee is tagged with a unique identifier, enabling the system to know exactly which individuals interacted in each case. The final result should be a map of all interactions, from which it is possible to derive, for example, a sociogram.
- Analyzing Tubular Tissue in Histopathological Thin Sections
Authors: Azadeh Fakhrzadeh, Ellinor Spörndly-Nees (1), Lena Holm (1), Cris L. Luengo Hendriks
(1) Department of Anatomy, Physiology and Biochemistry, SLU, Uppsala
Conference: International Conference on Digital Image Computing Techniques and Applications (DICTA), Perth, Australia, pp 1-6
Publisher: IEEE Publications
Abstract: We propose a method for automatic segmentation of tubules in the stained thin sections of various tissue types. Tubules consist of one or more layers of cells surrounding a cavity. The segmented tubules can be used to study the morphology of the tissue. Some research has been done to automatically estimate the density of tubules. To the best of our knowledge, no one has been able to, fully automatically, segment the whole tubule. Usually the border between tubules is subtle and appears broken in a straight-forward segmentation. Here we suggest delineating these borders using the geodesic distance transform. We apply this method on images of Periodic Acid Shiffs (PAS) stained thin sections of testicular tissue, delineating 89% of the tubules correctly.
