A Model Community Information System for Postmortem Psychiatry

Letter of Intent for a P20 proposal to the Human Brain Project
July 1, 1994 for October 15, 1994 proposal submission

Stanley Watson, PI

Mental Health Research Institute
Department of Psychiatry
University of Michigan Medical School
205 Zina Pitcher Place
Ann Arbor, MI 48109-0720
(313) 763-3725 Fax (313) 747-4130
email: Stanley.Watson@med.umich.edu

Bruce Schatz, Co-PI

National Center for Supercomputing Applications
Graduate School of Library and Information Science
Department of Computer Science
Program in Neuroscience
University of Illinois, Urbana-Champaign
schatz@canis.uiuc.edu

Clifford Saper

Beth Israel Hospital
Department of Neurology
Harvard Medical School
csaper@bih.harvard.edu

Thomas Huang

Beckman Institute for Advanced Science and Technology
Department of Electrical and Computer Engineering
University of Illinois, Urbana-Champaign
huang@ifp.uiuc.edu

Dr. Michael Huerta


National Institutes of Mental Health
5600 Fishers Lane, Room 11-95
Rockville, MD 20857
(301) 443-3948
June 30, 1994

Dear Michael,

In response to the request for a letter of intent for the Human Brain Project, we outline here a proposal we plan to submit as a P20 for the October 15, 1994 round of the HBP. The four of us (S. Watson PI, University of Michigan; B. Schatz Co-PI, University of Illinois; C. Saper, Harvard University; T. Huang, University of Illinois) intend to submit a four project proposal with a budget in the range of $650-$700 K (direct cost) or $1.1 M (total) each year for 5 years.
The overall project is to develop and deploy the Psychiatric Postmortem Community System (PPCS), a model information system for the biochemistry of neuroanatomical maps, as well as performing underlying research into encoding of neuroscience knowledge and design of information technology. The neuroscience research covers clinical analysis of mental illness and drug effects, along with basic science of molecular genetics, cellular biochemistry, and neuroanatomy. The information research covers practical construction of digital libraries and information systems, along with basic technology of information retrieval, network transmission, and image processing.
We propose to develop brain maps based on the anatomy of postmortem human brains. The project emphasizes the organization and integration of experimental data from the community of scientists who directly study the brains of the severely mentally ill (the "Psychiatric Postmortem Community" or PPC). These studies correlate the behavioral and clinical analysis of patients with mental illness, such as schizophrenia or depression, with the postmortem and scientific analysis of their neuroanatomy, at the cellular biochemistry and molecular genetics levels. The concentration on chemical analysis of sections of human brains, as opposed to the animal models or functional images of existing HBP projects, forces a concentration on new and difficult technological and sociological challenges. In particular, the technology must handle a diverse community of geographically distributed laboratories and the maps must handle individual variations in physical structures.
The project consists of both experimental construction and theoretical investigtion. The efforts are focused around a national testbed of neuroscientists who will be using the evolving PPCS information system. The bulk of the resources will be devoted to developing this system and deploying it as a working tool for the community. At the same time, research will be performed on more advanced functions relevant to both the neuroscience and the information technology, which will be incorporated into the system testbed in the later years of the project.
PPCS is scaled to provide a national testbed of human brain mapping. Some 30 laboratories distributed across the country, whose investigators participate in the PPC, have agreed to actively participate in this experiment, including entry of their data as well as use of the system. The PI of this project is a senior member of this community and has a close working relationship with its members. The software system being used as a basis for this project, the Worm Community System (WCS), is one of the few systems capable of supporting an electronic scientific community across the Internet with both data retrieval and data entry. The Co-PI of this project is the developer of the WCS, which was referenced in the original Institute of Medicine report on Mapping the Brain as the best current model for what a brain system could be and which has become a national model for computer support of scientific collaboration.
The project is organized as 4 individual subprojects, using the P20 mechanism. Each individual project investigator is a national leader in his own area. Two of the subprojects form the components of the PPCS itself. Project 1 is the neuroscience effort to encode the knowledge of the community and deploy the system to the laboratories. Project 2 is the information technology effort to develop the network information system which will enable interaction with and distribution of the community knowledge. Two of the subprojects are the longer-term investigation of fundamental issues required for a successful system in this area. Project 3 is the neuroscience effort to develop a brain map for actual human neuroanatomy, with semantic retrieval of labelled regions. Project 4 is the information technology effort to develop processing techniques for effective network transmission of neuroanatomy images.
Together, these projects form a synergistic whole, focused around the testbed and evolving its technology and usage in a working neuroscience community. While this proposal has the aim of designing and implementing a shared scientific information system for a small community of scholars, the strategies proposed should be transferable to other neuroscience communities. This proposal may thus be seen as a model for the HBP and a forerunner of the new information systems most likely to underlie major gains in the neurobiology of the future.

We propose to pursue the following specific aims:

Project 1. (S. Watson- P.I.) Neuroscience Knowledge Encoding :


We will organize a prototype neuroscience community (the Psychiatric Postmortem Community), with emphasis on developing a shared database of published experiments available across the Internet.


This project will design the knowledge representation for the experiments in this community, which include patient descriptions, biochemical concentrations, and anatomical images. The data sets will be organized around providing more detailed information on the experimental results referenced in the published literature. The data sets will be entered and submitted into the integrated database directly from the laboratories, using the same system (described in Project 2) used for browsing and searching the existing database.
Implementation of the Psychiatric Postmortem Community system (that is, the integration of Projects 1 and 2), will occur in three stages over the life of the grant: the pilot phase of 5-7 test labs (1-1.5 years), followed by the first stage of full use by the core of the community (10-15 labs; 1.5-3 yrs), and finally the entire community (30 labs, 3-5 yrs). Issues such as a community organizational structure, database design, strategies for organizing image sets and linking them to experiments, data ownership, data sharing, means of collaboration, and access to the database are issues to be addressed in the early phases. After implementation of the community software (Project 2) and active influx of large data sets, we anticipate addressing questions of the nature and structure of the larger community functioning, including extended private collaborations and the means of implementing high level retrieval of cross-study data sets.


Project 2. (B. Schatz- P.I.) Systems Technology Research:


We will develop a prototype network information system (the Psychiatric Postmortem Community System), which will enable browsing and sharing of the community experimental knowledge.


This project will evolve the Worm Community System (WCS) technology into functionality suitable for the PPCS. This technology already supports interaction with existing biology databases across the Internet and data entry into the databases directly from the laboratories across the Net. The first release is running on a production basis in some 30 laboratories studying the nematode worm C. elegans and the second release is running on an experimental basis in the initial 5 laboratories. (Note this is the same scale of community as the PPC.) For the worm community, WCS already supports a full complement of molecular biology data, such as genes displayed as forms and clones displayed as graphical maps, and a wide array of biology literature, such as journal abstracts and newsletter articles both displayed as formatted documents with embedded references to database items. For the postmortem community, in the early phases, the data types must be extended to those defined in Project 1, such as patient descriptions and anatomical images, including new displays for linking the neuroanatomy to the biochemistry. In the later phases, the results of the research in Projects 3 and 4 must be incorporated to provide a graphical brain map as a query interface and to support interaction with large image datasets across the Internet.

Project 3. (C. Saper- P.I.) Neuroscience Brain Mapping:


We will develop an interactive digital atlas for human neuroanatomy.


The Human Brain Atlas Digital Interface (HBADI) is a response to the burgeoning literature mapping the connections, neurotransmitters, and functional capabilities of the human brain. Many scientists, who are limited by the boundaries of the disciplines in which they trained and work, find it difficult to cross those boundaries in accessing the latest and most important results of scientists who may work on the same functional systems in the brain, but using different approaches. The HBADI will bridge this gap by providing a set of atlas-quality photographs of sections through the human brain, intended for use at a cellular resolution. This subproject will be coordinated by this investigator, who is the editor of the Journal of Comparative Neurology. The digital images representing these sections will be overlaid with designated regions indicating neuroanatomical structures. To handle different structural maps used by different experimenters, several different map overlays will be available with anatomical information provided by different anatomists expert in those regions of human brain. To handle individual variations, region selection will be made by pointing to a region in a map or by specifying the name of the region containing the desired cells (rather than giving idealized coordinates). In the early phases, the HBADI will be used in a standalone fashion to provide a query engine for searching Medline. In the later phases, it will be incorporated into the PPCS as a graphical interface to the extended literature datasets and integrated with the semantic thesaurus technology that is part of the WCS which supports terminology mapping across subject domains.

Project 4. (T. Huang- P.I.) Image Technology Research:


We will develop processing techniques for network transmission of neuroanatomical images.


High-resolution images are a major component of the databases forming the community knowledge of the PPC. These images must display cellular resolution and be displayable within the system rapidly enough to be interactive. Current end-to-end network bandwidth is adequate for text and tables, as shown by WCS, but only for images in specialized circumstances. Although the bandwidth will increase over the course of the project (and we will run tests using faster experimental networks since the co-PI is located at a supercomputer center), significant image processing will be necessary to make interaction across the Internet a reality for the PPCS and other neuroscience projects. The necessary research will be coordinated by this investigator, who is the chair of the IEEE Conference on Processing of Biological Images. In the early phases, the concentration will be on theoretical analysis of different algorithms for coding and compression, with their effect on network transmission of neuroanatomical images. In the later phases, a practical implementation of these and other techniques will be used in the PPCS to provide network interaction with large image repositories.