-
Patch-based Convolutional Neural Network for Whole Slide Tissue Image Classification
Authors:
Le Hou,
Dimitris Samaras,
Tahsin M. Kurc,
Yi Gao,
James E. Davis,
Joel H. Saltz
Abstract:
Convolutional Neural Networks (CNN) are state-of-the-art models for many image classification tasks. However, to recognize cancer subtypes automatically, training a CNN on gigapixel resolution Whole Slide Tissue Images (WSI) is currently computationally impossible. The differentiation of cancer subtypes is based on cellular-level visual features observed on image patch scale. Therefore, we argue t…
▽ More
Convolutional Neural Networks (CNN) are state-of-the-art models for many image classification tasks. However, to recognize cancer subtypes automatically, training a CNN on gigapixel resolution Whole Slide Tissue Images (WSI) is currently computationally impossible. The differentiation of cancer subtypes is based on cellular-level visual features observed on image patch scale. Therefore, we argue that in this situation, training a patch-level classifier on image patches will perform better than or similar to an image-level classifier. The challenge becomes how to intelligently combine patch-level classification results and model the fact that not all patches will be discriminative. We propose to train a decision fusion model to aggregate patch-level predictions given by patch-level CNNs, which to the best of our knowledge has not been shown before. Furthermore, we formulate a novel Expectation-Maximization (EM) based method that automatically locates discriminative patches robustly by utilizing the spatial relationships of patches. We apply our method to the classification of glioma and non-small-cell lung carcinoma cases into subtypes. The classification accuracy of our method is similar to the inter-observer agreement between pathologists. Although it is impossible to train CNNs on WSIs, we experimentally demonstrate using a comparable non-cancer dataset of smaller images that a patch-based CNN can outperform an image-based CNN.
△ Less
Submitted 9 March, 2016; v1 submitted 29 April, 2015;
originally announced April 2015.
-
Statistical Characterization of the Chandra Source Catalog
Authors:
Francis A. Primini,
John C. Houck,
John E. Davis,
Michael A. Nowak,
Ian N. Evans,
Kenny J. Glotfelty,
Craig S. Anderson,
Nina R. Bonaventura,
Judy C. Chen,
Stephen M. Doe,
Janet D. Evans,
Giuseppina Fabbiano,
Elizabeth C. Galle,
Danny G. Gibbs II,
John D. Grier,
Roger M. Hain,
Diane M. Hall,
Peter N. Harbo,
Xiangqun,
He,
Margarita Karovska,
Vinay L. Kashyap,
Jennifer Lauer,
Michael L. McCollough,
Jonathan C. McDowell
, et al. (14 additional authors not shown)
Abstract:
The first release of the Chandra Source Catalog (CSC) contains ~95,000 X-ray sources in a total area of ~0.75% of the entire sky, using data from ~3,900 separate ACIS observations of a multitude of different types of X-ray sources. In order to maximize the scientific benefit of such a large, heterogeneous data-set, careful characterization of the statistical properties of the catalog, i.e., comple…
▽ More
The first release of the Chandra Source Catalog (CSC) contains ~95,000 X-ray sources in a total area of ~0.75% of the entire sky, using data from ~3,900 separate ACIS observations of a multitude of different types of X-ray sources. In order to maximize the scientific benefit of such a large, heterogeneous data-set, careful characterization of the statistical properties of the catalog, i.e., completeness, sensitivity, false source rate, and accuracy of source properties, is required. Characterization efforts of other, large Chandra catalogs, such as the ChaMP Point Source Catalog (Kim et al. 2007) or the 2 Mega-second Deep Field Surveys (Alexander et al. 2003), while informative, cannot serve this purpose, since the CSC analysis procedures are significantly different and the range of allowable data is much less restrictive. We describe here the characterization process for the CSC. This process includes both a comparison of real CSC results with those of other, deeper Chandra catalogs of the same targets and extensive simulations of blank-sky and point source populations.
△ Less
Submitted 5 May, 2011; v1 submitted 3 May, 2011;
originally announced May 2011.
-
Corona, Jet, and Relativistic Line Models for Suzaku/RXTE/Chandra-HETG Observations of the Cygnus X-1 Hard State
Authors:
Michael A. Nowak,
Manfred Hanke,
Sarah N. Trowbridge,
Sera B. Markoff,
Joern Wilms,
Katja Pottschmidt,
Paolo Coppi,
Dipankar Maitra,
John E. Davis,
Frank Tramper
Abstract:
Using Suzaku and the Rossi X-ray Timing Explorer, we have conducted a series of four simultaneous observations of the galactic black hole candidate Cyg X-1 in what were historically faint and spectrally hard low states. Additionally, all of these observations occurred near superior conjunction with our line of sight to the X-ray source passing through the dense phases of the focused wind from the…
▽ More
Using Suzaku and the Rossi X-ray Timing Explorer, we have conducted a series of four simultaneous observations of the galactic black hole candidate Cyg X-1 in what were historically faint and spectrally hard low states. Additionally, all of these observations occurred near superior conjunction with our line of sight to the X-ray source passing through the dense phases of the focused wind from the mass donating secondary. One of our observations was also simultaneous with observations by the Chandra-High Energy Transmission Grating. These latter spectra are crucial for revealing the ionized absorption due to the secondary's focused wind. Such absorption is present and must be accounted for in all four spectra. These simultaneous data give an unprecedented view of the 0.8-300 keV spectrum of Cyg X-1, and hence bear upon both corona and X-ray emitting jet models of black hole hard states. Three models fit the spectra well: coronae with thermal or mixed thermal/non-thermal electron populations, and jets. All three models require a soft component that we fit with a low temperature disk spectrum with an inner radius of only a few tens of GM/c^2. All three models also agree that the known spectral break at 10\,keV is not solely due to the presence of reflection, but each gives a different underlying explanation for the augmentation of this break. Thus whereas all three models require that there is a relativistically broadened Fe line, the strength and inner radius of such a line is dependent upon the specific model, {thus making premature line-based estimates of the black hole spin in the Cyg X-1 system. We look at the relativistic line in detail, accounting for the narrow Fe emission and ionized absorption detected by HETG. Although the specific relativistic parameters of the line are continuum-dependent, none of the broad line fits allow for an inner disk radius that is >40 GM/c^2.
△ Less
Submitted 21 December, 2010;
originally announced December 2010.
-
The Three-Dimensional Structure of Cassiopeia A
Authors:
Tracey DeLaney,
Lawrence Rudnick,
M. D. Stage,
J. D. Smith,
Karl Isensee,
Jeonghee Rho,
Glenn E. Allen,
Haley Gomez,
Takashi Kozasa,
William T. Reach,
J. E. Davis,
J. C. Houck
Abstract:
We used the Spitzer Space Telescope's Infrared Spectrograph to map nearly the entire extent of Cassiopeia A between 5-40 micron. Using infrared and Chandra X-ray Doppler velocity measurements, along with the locations of optical ejecta beyond the forward shock, we constructed a 3-D model of the remnant. The structure of Cas A can be characterized into a spherical component, a tilted thick disk, an…
▽ More
We used the Spitzer Space Telescope's Infrared Spectrograph to map nearly the entire extent of Cassiopeia A between 5-40 micron. Using infrared and Chandra X-ray Doppler velocity measurements, along with the locations of optical ejecta beyond the forward shock, we constructed a 3-D model of the remnant. The structure of Cas A can be characterized into a spherical component, a tilted thick disk, and multiple ejecta jets/pistons and optical fast-moving knots all populating the thick disk plane. The Bright Ring in Cas A identifies the intersection between the thick plane/pistons and a roughly spherical reverse shock. The ejecta pistons indicate a radial velocity gradient in the explosion. Some ejecta pistons are bipolar with oppositely-directed flows about the expansion center while some ejecta pistons show no such symmetry. Some ejecta pistons appear to maintain the integrity of the nuclear burning layers while others appear to have punched through the outer layers. The ejecta pistons indicate a radial velocity gradient in the explosion. In 3-D, the Fe jet in the southeast occupies a "hole" in the Si-group emission and does not represent "overturning", as previously thought. Although interaction with the circumstellar medium affects the detailed appearance of the remnant and may affect the visibility of the southeast Fe jet, the bulk of the symmetries and asymmetries in Cas A are intrinsic to the explosion.
△ Less
Submitted 16 November, 2010;
originally announced November 2010.
-
The Chandra Source Catalog
Authors:
Ian N. Evans,
Francis A. Primini,
Kenny J. Glotfelty,
Craig S. Anderson,
Nina R. Bonaventura,
Judy C. Chen,
John E. Davis,
Stephen M. Doe,
Janet D. Evans,
Giuseppina Fabbiano,
Elizabeth C. Galle,
Danny G. Gibbs II,
John D. Grier,
Roger M. Hain,
Diane M. Hall,
Peter N. Harbo,
Xiangqun,
He,
John C. Houck,
Margarita Karovska,
Vinay L. Kashyap,
Jennifer Lauer,
Michael L. McCollough,
Jonathan C. McDowell,
Joseph B. Miller
, et al. (14 additional authors not shown)
Abstract:
The Chandra Source Catalog (CSC) is a general purpose virtual X-ray astrophysics facility that provides access to a carefully selected set of generally useful quantities for individual X-ray sources, and is designed to satisfy the needs of a broad-based group of scientists, including those who may be less familiar with astronomical data analysis in the X-ray regime. The first release of the CSC in…
▽ More
The Chandra Source Catalog (CSC) is a general purpose virtual X-ray astrophysics facility that provides access to a carefully selected set of generally useful quantities for individual X-ray sources, and is designed to satisfy the needs of a broad-based group of scientists, including those who may be less familiar with astronomical data analysis in the X-ray regime. The first release of the CSC includes information about 94,676 distinct X-ray sources detected in a subset of public ACIS imaging observations from roughly the first eight years of the Chandra mission. This release of the catalog includes point and compact sources with observed spatial extents <~ 30''. The catalog (1) provides access to the best estimates of the X-ray source properties for detected sources, with good scientific fidelity, and directly supports scientific analysis using the individual source data; (2) facilitates analysis of a wide range of statistical properties for classes of X-ray sources; and (3) provides efficient access to calibrated observational data and ancillary data products for individual X-ray sources, so that users can perform detailed further analysis using existing tools. The catalog includes real X-ray sources detected with flux estimates that are at least 3 times their estimated 1 sigma uncertainties in at least one energy band, while maintaining the number of spurious sources at a level of <~ 1 false source per field for a 100 ks observation. For each detected source, the CSC provides commonly tabulated quantities, including source position, extent, multi-band fluxes, hardness ratios, and variability statistics, derived from the observations in which the source is detected. In addition to these traditional catalog elements, for each X-ray source the CSC includes an extensive set of file-based data products that can be manipulated interactively.
△ Less
Submitted 25 May, 2010;
originally announced May 2010.
-
Cosmic ray diffusion near the Bohm limit in the Cassiopeia A supernova remnant
Authors:
Michael D. Stage,
Glenn E. Allen,
John C. Houck,
John E. Davis
Abstract:
Supernova remnants (SNRs) are believed to be the primary location of the acceleration of Galactic cosmic rays, via diffusive shock (Fermi) acceleration. Despite considerable theoretical work the precise details are still unknown, in part because of the difficulty in directly observing nucleons that are accelerated to TeV energies in, and affect the structure of, the SNR shocks. However, for the…
▽ More
Supernova remnants (SNRs) are believed to be the primary location of the acceleration of Galactic cosmic rays, via diffusive shock (Fermi) acceleration. Despite considerable theoretical work the precise details are still unknown, in part because of the difficulty in directly observing nucleons that are accelerated to TeV energies in, and affect the structure of, the SNR shocks. However, for the last ten years, X-ray observatories ASCA, and more recently Chandra, XMM-Newton, and Suzaku have made it possible to image the synchrotron emission at keV energies produced by cosmic-ray electrons accelerated in the SNR shocks. In this article, we describe a spatially-resolved spectroscopic analysis of Chandra observations of the Galactic SNR Cassiopeia A to map the cutoff frequencies of electrons accelerated in the forward shock. We set upper limits on the electron diffusion coefficient and find locations where particles appear to be accelerated nearly as fast as theoretically possible (the Bohm limit).
△ Less
Submitted 18 August, 2006;
originally announced August 2006.
-
Using the Parallel Virtual Machine for Everyday Analysis
Authors:
M. S. Noble,
J. C. Houck,
J. E. Davis,
A. Young,
M. Nowak
Abstract:
A review of the literature reveals that while parallel computing is sometimes employed by astronomers for custom, large-scale calculations, no package fosters the routine application of parallel methods to standard problems in astronomical data analysis. This paper describes our attempt to close that gap by wrap** the Parallel Virtual Machine (PVM) as a scriptable S-Lang module. Using PVM with…
▽ More
A review of the literature reveals that while parallel computing is sometimes employed by astronomers for custom, large-scale calculations, no package fosters the routine application of parallel methods to standard problems in astronomical data analysis. This paper describes our attempt to close that gap by wrap** the Parallel Virtual Machine (PVM) as a scriptable S-Lang module. Using PVM within ISIS, the Interactive Spectral Interpretation System, we've distributed a number of representive calculations over a network of 25+ CPUs to achieve dramatic reductions in execution times. We discuss how the approach applies to a wide class of modeling problems, outline our efforts to make it more transparent for common use, and note its growing importance in the context of the large, multi-wavelength datasets used in modern analysis.
△ Less
Submitted 24 October, 2005;
originally announced October 2005.
-
The Chandra High Energy Transmission Grating: Design, Fabrication, Ground Calibration and Five Years in Flight
Authors:
Claude R. Canizares,
John E. Davis,
Daniel Dewey,
Kathryn A. Flanagan,
Eugene B. Galton,
David P. Huenemoerder,
Kazunori Ishibashi,
Thomas H. Markert,
Herman L. Marshall,
Michael McGuirk,
Mark L. Schattenburg,
Norbert S. Schulz,
Henry I. Smith,
Michael Wise
Abstract:
Details of the design, fabrication, ground and flight calibration of the High Energy Transmission Grating, HETG, on the Chandra X-ray Observatory are presented after five years of flight experience. Specifics include the theory of phased transmission gratings as applied to the HETG, the Rowland design of the spectrometer, details of the grating fabrication techniques, and the results of ground t…
▽ More
Details of the design, fabrication, ground and flight calibration of the High Energy Transmission Grating, HETG, on the Chandra X-ray Observatory are presented after five years of flight experience. Specifics include the theory of phased transmission gratings as applied to the HETG, the Rowland design of the spectrometer, details of the grating fabrication techniques, and the results of ground testing and calibration of the HETG. For nearly six years the HETG has operated essentially as designed, although it has presented some subtle flight calibration effects.
△ Less
Submitted 1 July, 2005;
originally announced July 2005.
-
The Formal Underpinnings of the Response Functions used in X-Ray Spectral Analysis
Authors:
John E. Davis
Abstract:
This work provides an in-depth mathematical description of the response functions that are used for spatial and spectral analysis of X-ray data. The use of such functions is well-known to anyone familiar with the analysis of X-ray data where they may be identified with the quantities contained in the Ancillary Response File (ARF), the Redistribution Matrix File (RMF), and the Exposure Map. Start…
▽ More
This work provides an in-depth mathematical description of the response functions that are used for spatial and spectral analysis of X-ray data. The use of such functions is well-known to anyone familiar with the analysis of X-ray data where they may be identified with the quantities contained in the Ancillary Response File (ARF), the Redistribution Matrix File (RMF), and the Exposure Map. Starting from first-principles, explicit mathematical expressions for these functions, for both imaging and dispersive modes, are arrived at in terms of the underlying instrumental characteristics of the telescope including the effects of pointing motion. The response functions are presented in the context of integral equations relating the expected detector count rate to the source spectrum incident upon the telescope. Their application to the analysis of several source distributions is considered. These include multiple, possibly overlap**, and spectrally distinct point sources, as well as extended sources. Assumptions and limitations behind the usage of these functions, as well as their practical computation are addressed.
△ Less
Submitted 2 November, 2000;
originally announced November 2000.