Posted on by & filed under Content - Highlights and Reviews.

Screen name: Ash
Title/Functional role: Lead Developer
Project title: Automated TB nodule detection in chest X-rays
Time to complete : a year
Project details: Pulmonary Tuberculosis (TB) is a common and deadly infectious disease which can infect the lungs. One of our natural defences against TB is to try to ‘wall’ the foreign bacterium off, resulting in nodular densities that can calcify, these nodules indicate ‘inactive TB’ is present in the patient and the nodular densities are seen on chest X-rays. Medically it is important to know whether a person has inactive TB because this can indicate that the patient is now more susceptible to contract TB again. The aim of my project was to compare the use of morphological image processing techniques, support vector machines, statistical texture representation and correlation matching as methods to create an automated detection system for these nodules within a chest X-ray. A number of algorithms were tested, firstly the ‘Modified Grayscale Top Hat Morphological Transform,’ (MGTH) a technique that uses morphological image processing techniques to enhance calcified regions of the lung. Secondly an Investigation into using Support Vector Machines, a relatively new machine learning tool, to a classify calcified nodules was carried out. Eventually I defined a new algorithm – the MGTH-SVM-Transform, used to detect calcified parts of the lung I also suggest a new algorithm for the detection of image artefacts is introduced and evaluated. The algorithms under investigation were evaluated using the receiver operator curve, where the MGTH-SVM-transform achieves its highest performance of 0.8902 (as the area under the ROC), and the MGTH achieves 0.9010. One of the areas I also investigated during this project was the use of Principle Component Analysis…during this investigation i discovered and coded a library which allows application developers to perform Singular Value Decomposition (SVD) on matrices above 4gb within a 32bit machine, This is probably the probably the only framework allowing users to perform SVD on datasets that are too large to fit into physical memory, this is a problem that is very relevant to many scientific applications such as numerical weather prediction, astronomical data modeling and many industrial image processing applications. info can be found on
Resources used: [1] Tuberculosis, 2003. Publisher: Lippincott Williams. William N Rom, Stuart M Garay [2] The White Death: A History of Tuberculosis, 2001. Publisher: Hambledon & London. Thomas Dormandy [3] World Health Organization (WHO). Tuberculosis Fact sheet N°104 – Global and regional incidence. March 2006. [4] Interpreting Chest X-Rays. Cambridge Press. Philip Eng, Foong-Koon Cheah [5] Tuberculosis: Current Concepts and Treatment, Second Edition. Publisher: Informa Healthcare (2000) Lloyd N. Friedman [6] Digital chest radiography. Clin. Chest Med., 12:19–32, 1991. H. MacMahon and K. Doi. [7] Image quality and the clinical radiographic examination. Radiographics, 17:479–498, 1997. Cj Vyborny. [8] Expert System with an Embedded Imaging Module for Diagnosing Lung Diseases. Retrieved from IEEE Xplore 02 January 2007 Kavitha Shaga Devan, P.A Venkatachalam, Ahmad Fadzil Mohd Hani. [9] Digital Image Processing using Matlab. Publisher: publishing Rafeal C Gonzalez, Richard E. Woods, Steven L.Eddins, [10] Study of Computer Diagnosis of X-ray and CT Images in Japan – A Brief Survey Department of Information Engineering, Faculty of Engineering, Nagoya University NAGOYA, 464-01, JAPAN, Retrieved from IEEE Xplore 02 January 2007 Jun-ichiro TORIWAKI, Furo-cho, Chikusa-ku, *11+, “Screening for cancer,” in selected topic in Chest Radiology, A. B. Miller, Ed. New York, Academic, 1985, pp. 163-191. F. P. Stitik, M. S. Tockman, and N. F. Khouri *12+ “Non small cell lung cancer: Results of the New York screening program,” Radiology, vol. 151, pp. 289-293, 1984. R. T. Heelan, B. J. Flechinger, and M. R. Melamed et al. [13]Artificial Convolution Neural Network Techniques and Applications for Lung Nodule Detection Shih-Chung B. Lo, Shyh-Liang A. Lou, Jyh-Shyan Lin, Matthew T. Freedman, Minze V. Chien, and Seong K. Mun, [14]Statistical Learning Theory –Wiley-Interscience Publication. Vladimir N. Vapnik. [15] A Practical Guide to Support Vector Classification Department of Computer Science and Information Engineering National Taiwan University Chih-Wei Hsu, Chih-Chung Chang, and Chih-Jen Lin *16+ “A tutorial on support vector machines for pattern classification” Tech Rep, Bell Laboratories, Lucent Technologies. 2000 C. J. C. Burges [17] Morphological classification of medical images using nonlinear support vector machines Section for Biomedical Image Analysis, Department of Radiology. University of Pennsylvania, Philadelphia, Pennsylvania Christos Davatzikos, Dinggang Shen, Zhiqiang Lao, Zhong Xue, Bilge Karacali [18] Kernel Methods for Pattern Analysis. Cambridge Press. John Shaww-Taylor, Nello Cristianini [19] An algorithm for the enhancement of chest X-rays in TB patients *S. Hariharan,* * A. K. Ray and * M. K. Ghosh [20] Feature Selection in the Pattern Classification, Problem of Digital Chest Radiograph Segmentation Michael F. McNitt-Gray, Member, IEEE, H. K. Huang, Senior Member, IEEE, and James W. Sayre [21] Neural network image analysis and classification in Hybrid Lung Nodule Detection (HLND) system. Caelum Research Corporation Y. S . Peter Chiou”, Y. M. Fleming Lure* and Panos A. Ligomenides’ [22] lung segmentation in digitized posteroanterior chest radiographs. Academic radiology, 5:245–255, 1998. S.G. Armato, M.Giger, and H.MacMahon. [23] Knowledge-based method for segmentation and analysis of lung boundaries in chest x-rays images. Computerized Medical Imaging and Graphics, 22:463–477, 1998. M.S. Brown, L.S. Wilson, B.D. Doust, R.W. Gill, and C.Sun. [24] Automatic calculation of total lung capacity from automatically traced lung boundaries in postero- anterior and lateral digital chest radiographs. Medical Physics, 25:1118–1131, 1998. F.M. Carrascal, J.M. Carreira, M. Souto, P.G. Tahoces, L. Gomez, and J.J. Vidal. [25] A fully automatic algorithm for the segmentation of lung fields in digital chest radiographic images. Medical Physics, 22:183–191, 1995. J. Duryea and J.M. Boone. [26] Automated Segmentation of Multiple Sclerosis Lesions Shahrum Nedjati-Gilani Supervisor: Dr. Daniel Alexander, September 2004 [27] Kernel Methods for Pattern Analysis (Hardcover) , Publisher: Cambridge University Press (June 28, 2004) John Shawe-Taylor, Nello Cristianini [28] Pattern Classification: A Unified View of Statistical and Neural Approaches, Publisher: Wiley-Interscience March 1996 Jürgen Schürmann [29] Medical Imaging 2002: Image Processing (Progress in Biomedical Optics and Imaging,) Milan Sonka , J. Michael Fitzpatrick [30] Detecting Cognitive States from fMRI Images by Machine Learning and Multivariate Classification, University of Pennsylvania, USA presented in MMBIA 2006: IEEE Computer Society Workshop on Mathematical Methods in Biomedical Image Analysis. 2006 Yong Fan, Dinggang Shen, and Christos Davatzikos, [31], Neural Networks, A comprehensive Foundation, 2nd edition. Prentice Hall International, Inc. 1999. Simon Haykin [32] Image Processing Analysis, and Machine Vision, 1999 Milan Sonka, Vaclav Hlavac, Roger Boyle [33] Matlab Technical Help and support documentation, 2007 The MathWorks, Inc
Parties involved: University College London – Computer Science dept.
Any other details you would like us to know: For the above projects i mainly programmed in MATLAB and C++, I am currently working as a Lead Developer at Credit-Suisse First Boston as a C# WPF Framework developer
What do you read?: Concurrent Programming in Windows – Joe Duffy
CLR via C# – Jeff Richter
Framework Design Guidelines – Brad Adams
An introduction to The Theory Computation – Sipser
Expert F# – Don Syme
Computer Organization and Design – David A. Patterson and John L. Hennessy
Win32 Multithreaded Programming by Aaron Cohen and Mike Woodring
GPU Gems 3 by Hubert Nguyen
modern operating systems – Tanenbaum
Foundations of Genetic Programming by William B. Langdon and Riccardo Poli
An Introduction to Support Vector Machines and Other Kernel-based Learning Methods by Nello Cristianini and John Shawe-Taylor


Tags: Automated TB Nodule Detection in Chest X-rays, Contests, Challenges and Sweepstakes, dynamite coder contest, win safari books online subscription,

7 Responses to “*Dynamite Coding Skills Winner ~ Automated TB Nodule Detection in Chest X-rays”

    • Ashley Aberneithy

      Good to know my paper is useful – I’ve sent you a mail, would be good to touch base.



  1.  DYNAMITE CODING SKILLS Challenge Winners Selected! « Safari Books Online's Official Blog
  2.  DYNAMITE CODING SKILLS Challenge Winners Selected! « Safari Books Online's Official Blog
  3.  Innovative Tech Professionals and Coder Challenge Winners From Around the Globe Honored at Safari Books Online Event in Sebastopol, California « Safari Books Online's Official Blog
  4.  Innovative Tech Professionals and Coder Challenge Winners From Around the Globe Honored at Safari Books Online Event in Sebastopol, California « Safari Books Online's Official Blog
  5.  Press Release: Safari Books Online Announces Exciting “Learn Something New Team Challenge” « Safari Books Online's Official Blog