PERFEX Project

Knowledge Based Interpretation of Myocardial SPECT Imagery

1. Automatic Determination of the orientation of the left ventricular myocardium; This is being presented as a separate sub-project, called DISHA
2. Extension and enhancement of a knowledge base for interpreting myocardial perfusion imagery and other relevant information;
3. Prediction of resting perfusion from resting thickening distributions through connectionist methods;
4. Integration of connectionist and symbolic methods;
5. Implementation and automation of the methodology into a fully integrated system;
6. Clinical testing and evaluation of this system.

A demo of a simple prototype user interface is available here.

The PERFEX system

The overall goal is to assist in the diagnosis of coronary artery disease. The approach employs knowledge based methods to process and map the 3D visual information into symbolic representations, which are subsequently used to infer structure (anatomy) from function (physiology), as well as to interpret the temporal effects of perfusion redistribution, and assess the extent and severity of cardiovascular disease both quantitatively and qualitatively. The knowledge based system presents the resulting diagnostic recommendations in numerical, textual and visual forms in an interactive framework, thereby enhancing overall utility.

The PERFEX knowledge base currently contains over 250 rules. Present rules are improved upon and new rules added as the research within this project continues. PERFEX is currently implemented in an object oriented environment using Neuron Data's Nexpert Object v3.0. This object oriented framework provides some advantages, including portability inheritance properties and C code. This software, however, has been extensively modified to incorporate the Certainty Factor (CF) model used to handle likelihoods. After the creation of a simple User Interface (Sun version) to enable initial testing and evaluation, a more elaborate work environment (SGI version) has been created. This environment provides various images: planar, short / horizontal long / vertical long axis slices and Emory Bullseye plots. The Emory Bullseye plots are polar perfusion images, with the Apex as center.

Stress and Rest Bullseye Image

The system also provides an automatically generated report in English and an English language justification of that report. An HTML-translated (and severely simplified) demo of the system is available.

The system is undergoing extensive clinical evaluation. The system itself has been already ported to a commercial clinical system and will be released as a demo or extension package soon.

The latest development within this project is the use of an Artificial Neural Network (ANN) to predict defect reversibility based on stress and thickening information. The advantage of being capable of doing so is that performing a rest study wil no longer be strictly necessary.

Project Members:

• Dr. Norberto F. Ezquerra
• Levien de Braal
• Dr. Ernest V. Garcia
• C. David Cooke
• Rakesh Mullick
• Russell D. Folks
• Elizabeth G. Krawczynska

References:

• "PERFUSE: An Interactive Knowledge-Based System for the Interpretation & Explanation of Cardiac Imagery,"
  L. de Braal, N. Ezquerra, E. Garcia, C. Cooke, E. Krawczynska. IEEE Engineering in Medicine and Biology Society 1996 (EMBS 1996) Conference, Amsterdam, Netherlands, October 31 - November 3, 1996.

• "Expert System Interpretation of Technicium-99m Sestamibi Myocardial Perfusion Tomograms: Enhancements and Validation,"
  E. Garcia, D. Cooke, E. Krawczynska, R. Folks, J. Vansant, L. de Braal, R. Mullick, and N. Ezquerra, Circulation, Vol. 92, No. 8, October 1995.

• "Inteligencia Artificial en Medicina (Artificial Intelligence in Medicine, published in Spanish),"
  ISBN 84-88051-42-5, Coleccion Informatica No. 3-1994, Fund. A. Bra*as, Pub., Santiago de Compostela, Spain, 1994.

• "PERFEX: An Expert System for Interpreting Perfusion Images,"
  N. Ezquerra, R. Mullick, D. Cooke, E. Krawczynska, and E. Garcia. invited paper, Expert Systems With Applications, Vol. 6, pp. 459-468, 1993.

• "Myocardial Ischemia Detection by Expert System Interpretation of Thallium-201 Scintigrams,"
  M. Herbst, E. Garcia, D. Cooke, N. Ezquerra, R. Folks, and G. DePuey. Cardiovascular Nuclear Medicine and MRI, (J. Reiber and E. Van der Wall, eds.), Kluwer Academic Publishers, 1992.

• "PERFEX: An Expert System for Interpreting 3D Myocardial Perfusion,"
  N. F. Ezquerra, R. Mullick, E. V. Garcia, C. D. Cooke and E. Kachouska. Expert Systems with Applications, Pergamon Press, 1992.

• "A Knowledge Based System to Assist in the Diagnosis of Coronary Artery Disease,"
  R. Mullick, N. F. Ezquerra, E. V. Garcia and C. D. Cooke. Proceedings of the Tenth Southern Biomedical Engineering Conference, 107-9, October 1991.

• "Visual Protocol Collection for the Enhancement of the Radiological Diagnostic Process,"
  E. Rogers, R. Arkin, M. Baron, N. Ezquerra, and E. Garcia. IEEE Proc. Visualization in Biomedical Computing Conference, pp. 208-215, May 1990.

• "Visualization of Cardiovascular Nuclear Medicine Tomographic Perfusion Studies,"
  C. D. Cooke, E. Garcia, R. D. Folks, J. W. Peifer, and N. F. Ezquerra. Proc. Conf. on Visualization in Biomedical Computing, Atlanta, GA, pp. 185-189, May 1990.

• "Knowledge-based Visualization of myocardial perfusion tomographic images,"
  E. V. Garcia, M. D. Herbst, C. D. Cooke, N. F. Ezquerra, B. L. Evans, R. D. Folks and E. G. DePuey. Proceedings Visualization on Biomedical Computing 1990, 157-161, May 1990.

• "Artificial Intelligence in Nuclear Medicine Imaging,"
  N. F. Ezquerra and E. V. Garcia. American Journal of Cardiac Imaging, 3, 2, 130-41, 1989.

• "Artificial Intelligence in Medical Imaging,"
  N. Ezquerra and E. Garcia. Journal of Cardiac Imaging, Vol. 3, No. 2, pp. 130-141, June 1989.

• "Techniques and Artificial Intelligence in Cardiac Imaging,"
  E. DePuey, E. Garcia, and N. Ezquerra. Am. Journal of Roentgenology, Vol. 152, pp. 1161-1168, June 1989.

• "Artificial Intelligence Comes to The Clinic,"
  N. F. Ezquerra and E. V. Garcia. Annual Meeting of the Southeastern Chapter of the Society of Nuclear Medicine, Orlando, Florida, October 1987.

• "Feature Extraction as a Means of Consistent TI-201 SPECT Image Interpretation,"
  H. Hise, E. Garcia, and N. Ezquerra. J. Nuc. Med. Vol. 28, No. 4, P. 618-619, 1987.

• "Artificial Intelligence Interpretation of Myocardial Tomograms,"
  E. Garcia, N. Ezquerra, E. DePuey, H. Hise, and W. Robbins. American Heart Assoc., Part 2, Vol. 74, No. 4, p. II-295, 1987.

• "An Expert System for Interpreting Thallium-201 Tomographic Images,"
  E. V. Garcia and N. F. Ezquerra. 1987 IBM Academic Information Systems (ACIS) University Conference, p. 40, June 28 1987.

• "Knowledge-Based Cardiac Image Interpretation and Display,"
  N. F. Ezquerra, M. Zerbi, E. V. Garcia, H. L. Hise, and E. G. DePuey. Proceedings 1987 SOUTHCON Electronic Show and Convention, Atlanta, Georgia, March 1987.

• "A Knowledge-Based System for Interpreting Thallium-201 SPECT Images,"
  N. F. Ezquerra, H. L. Hise, E. V. Garcia, and E. G. DePuey. J. Nuc. Med., Vol. 27, No. 1 1, p. 1798, November 1986.

• "Artificial Intelligence Interpretation of Thallium-201 Myocardial Tomograms: Method and Pilot Study,"
  E. V. Garcia, N. F. Ezquerra, E. G. DePuey, H. L. Hise, and W. L. Robbins. Proc. of the Annual Meeting of the American Heart Association, Part 2, Vol. 74, No. 4., p.II-295, October 1986.

• "Development of an Expert System for Interpreting Medical Images,"
  N. F. Ezquerra, E. V. Garcia, E. G. DePuey, and W. L. Robbins. Proc. IEEE Int. Conf. on Systems, Man, and Cybernetics, Vol. 1, pp. 205-210, October 1986.

• "An Artificial Intelligence Approach to Interpreting Thallium-201 Three-Dimensional Myocardial Distributions,"
  N. Ezquerra, E. Garcia, and W. Robbins. Journal Nuclear Medicine, Vol. 27, No. 6, p. 1005, 1986.

• "A Knowledge-Based System for Interpreting Cardiovascular Nuclear Medicine Images,"
  N. F. Ezquerra, E. Garcia, E. DePuey, L. Hise, and S. Shapiro. Proc. of IEEE Computers in Cardiology Conf., Boston, MA, pp. 3-8, October 1986.

• "Artificial Intelligence for Imaging Problem-Solving,"
  N. Ezquerra and E. Garcia. Diagnostic Imaging, Vol. 7, No. 11, pp. 195-200, November 1985.