Grid technology in tissue-based diagnosis: fundamentals and potential developments
© Görtler et al. 2006
Received: 09 August 2006
Accepted: 24 August 2006
Published: 24 August 2006
Skip to main content
© Görtler et al. 2006
Received: 09 August 2006
Accepted: 24 August 2006
Published: 24 August 2006
Tissue-based diagnosis still remains the most reliable and specific diagnostic medical procedure. It is involved in all technological developments in medicine and biology and incorporates tools of quite different applications. These range from molecular genetics to image acquisition and recognition algorithms (for image analysis), or from tissue culture to electronic communication services.
Grid technology seems to possess all features to efficiently target specific constellations of an individual patient in order to obtain a detailed and accurate diagnosis in providing all relevant information and references.
Grid technology can be briefly explained by so-called nodes that are linked together and share certain communication rules in using open standards. The number of nodes can vary as well as their functionality, depending on the needs of a specific user at a given point in time. In the beginning of grid technology, the nodes were used as supercomputers in combining and enhancing the computation power. At present, at least five different Grid functions can be distinguished, that comprise 1) computation services, 2) data services, 3) application services, 4) information services, and 5) knowledge services.
The general structures and functions of a Grid are described, and their potential implementation into virtual tissue-based diagnosis is analyzed. As a result Grid technology offers a new dimension to access distributed information and knowledge and to improving the quality in tissue-based diagnosis and therefore improving the medical quality.
Tissue-based diagnosis includes all diagnosis procedures to analyze spatial configurations of biological functional units. Most frequently cells, cellular agglutinations such as vessels, nerves, glands, etc. are being investigated. Additional structures such as gene sequences, cellular movements, or membrane potentials are covered in advanced studies [1–7].
conventional or "classical histological and cytological" diagnosis,
"indicative" diagnosis, and
Diagnosis categories and medical application
Lymphoma, Chronic hepatitis, Lung carcinoma
Classic microscopy, Immunohistochemistry, Molecular biology, morphometry, structure analysis
TNM-stage, tumor proliferation rate, Apoptosis, Adhesion,
Immunohistochemistry, morphometry, structure analysis
Hormone receptors, Herceptin
Immunohistochemistry, Molecular biology, morphometry, structure analysis
The classical diagnosis is a prerequisite for any reliable treatment of chronic diseases such as cancer or chronic inflammatory lesions, and, by the way, is by far the cheapest diagnostic medical procedure [8, 9]. It is also quite independent from its medical environment, i.e., the specialization of a hospital or pathology institution in contrast to the other diagnosis types.
The recognition of a "risk-associated disease" such as the genetic predisposition to developing breast cancer is the duty of highly specialized (molecular genetic) institutions or departments.
Therefore, institutions involved in tissue-based diagnosis should have access to a variety of sources for data, information, and knowledge, to enable working in an efficient manner. At the same time they can provide integrated and highly abstracted information of the disease and direct the necessary treatment. This central embedding of diagnostic pathology has opened new doors in medical communication.
Another on-line technology for telepathology is the Remote Controlled Microscope. This is used by small surgical units, which do not host a surgical pathologist. The installed remote control microscopes require also "visually controlled tissue sampling and cutting tables". The systems permit intra-operative diagnosis of pathologists working with a congruent control and survey system installed in a remote pathology department or institution [24–31].
Different to these on-line telepathology systems the so-called off-line telepathology has been developed. Specific servers have been implemented to enable expert consultation, secondary advices, or to provide even a "virtual pathology institution" capability [9, 10, 32–41]. These systems are usually completely embedded into the Internet. Three main systems have been implemented so far, the iPATH [10, 18, 42] in Basel, Switzerland; the UICC-TPCC (Telepathology Consultation Center of the Union International Contre Cancer in Berlin, Germany [40, 43], and the Telepathology service of the Armed Forces Institute of Pathology (AFIP), located in Bethesda, Maryland, USA [40, 43–46]. These platforms allow sending information between distributed users; however, there is no interaction with communication systems or to grant access to computation facilities or specific data bases.
Another system, the Electronic Automated Measurement User System (EAMUS™, ) automatically measures the staining intensities and derived features of images acquired from immunohistochemically stained glass slides. It is an open system and can be accessed via the Internet [48, 49].
Obviously, these systems are all build on a specific purpose and cannot interact with each other. They can be considered to be precursors of more advanced and broader designed networks meeting the characteristics of a virtual network, a Grid.
All these systems require digital images acquired from a histological glass slide that are a prerequisite to using these tools. Today, still images of limited size (SVHS, or other formats of approximately 1000 × 1000 pixels) serve for these purposes. The glass slides are still archived in the conventional manner. However, since about two years glass slide scanning technologies are available, which acquire a complete glass slide and also provide interfaces for digital archives and support advanced Internet Communication between pathologists for interactive remote consultation [48, 49]. In a next step diagnostic pathology would move on from image acquisition generating "Digital Slides", into Virtual Networking, i.e. – using a Grid.
Grids are based on open standards like PACS (Picture Archiving and Communication System) for Medical Imaging and provide a simple, fast, resilient and open framework. They are designed to generate an easy to use platform for delivering intra- and interdisciplinary collaborative medicine. Images would be one core, and Health Care Systems can share pathology, cardiology, radiology and other digital images across sites.
Grid Technology enables physicians to access and use all compute and storage resources available in a virtual network. Users are granted physical freedom from the underlying technology, enabling fast remote access. Healthcare providers can leverage computing and storage resources across multiple departments and sites. By sharing resources, Grid technology will help to eliminate hardware vendor 'lock-in' via vendor agnostic architectures.
Obviously, immediate access to different diagnostic resources will improve the patients' care and physicians' diagnosis ability. Naturally, the network has to provide security and privacy to protect the patients' confidentiality.
What are the features of a Grid? Which Grids related to tissue-based diagnosis do already exist, and which specificities can be implemented in computational diagnostic pathology? Is the design of the existing telepathology services appropriate to be migrated into an advanced Grid system? Which features are promising, which ones have to be modified, or even neglected?
This article tries to give some answers from the technological and medical point of view to these questions. In addition, we want to describe the basics of Grid technology in relation to future changes in tissue-based diagnosis, which will most likely occur, in our opinion.
Basically, the components of a Grid include the end users or clients, the distribution and control nodes, and the servers, anyone able to perform the requested tasks. The concept of Grid computing was primarily developed to make use of the installed compute power, which was not fully utilized (e.g. office equipment during the off-hours). The benefits are improving the execution time for a compute intensive job in linking – even geographically dispersed computers – in order to combine their computational power for this individual job. As more users might be interested using this approach all their workload has to be managed to optimize the offered capacities and services. The infrastructure of a Grid is a computer-based collaborative environment using a management software layer (Middleware). This software layer again requires computation nodes, the so-called brokers. A Grid sourced broker administers the workload, potential problems, discovers free resources, and controls the
A client uses a Grid to solve his specific tasks, and to receive a solution independently where and by which individual systems, called resources, it has been generated. The Grid manages the accessibility of the combined distributed resources and their services. Therefore, it is adequate to analyze the implemented types of services from the end-user's point of view. These include computational, data, application, information, and knowledge services, which can be described as follows:
Computational services deal with secure distributed computational resources for executing application jobs and are provided by so-called resources brokers. They serve for the set up and analysis of high energy experiments, and are also a useful tool in astrophysics. Computational services solve tasks that require high computational power, for example to solve recursive formulas. In its simplest manner, a computational task is transferred to one of the distributed supercomputers. This computer takes the job as long as it is not busy with or overloaded by other tasks. Once this happens, the task and its computational stage are transferred to another included supercomputer, etc. as long as the task is not finished. Examples of computational Grids are: NASA IPG , the World Wide Grid [52, 53], and the NSF Tera-Grid [54, 55].
Data services offer secure access to distributed datasets. They manage access, retrieval, storage, replication, or catalogues of individual or distributed libraries. In a more simple structure their services can be implemented by so-called links, which has been realized by several search machines. These so-called Data Grids are used in the area of high-energy physics  or drug design [57, 58]. Another derivative is a Storage Grid as applied for Medical Imaging or data analysis in neurophysiology .
Application services manage Grid application and give access to remote software, libraries and Web services. They represent the next higher level built on computational and data services provided by the Grid. They combine the computation of specific formulas with access to prerequisite data sets. As an example, the user might be interested to viewing the shape of a new macromolecule that has some structural similarities to a known one. The application services provide the adequate formulas, and, in addition, the necessary databank of parameters etc. to fulfill this task. In tissue-based diagnosis, the EAMUS™ [9, 48, 49] can be considered as a simple, one node implementation of this service. A well known Grid application service is, for example, created by NetSolve .
Information services are at an advanced level of application services. They try to extract and present information provided by data of computational, information, and/or application services, and to put these into relationship. In tissue-based diagnosis, a simple implementation could be created by combining the EAMUS™ services with an existing telepathology information system such as UICC-TPCC, or iPATH. At low-level information services handle the way that information is represented, stored, accessed, shared, and maintained (Meta Data). An example of this service is the EU-sponsored Virolab Grid, a project that addresses the problem of HIV drug resistance. Its service offers the integration of biomedical information, advanced applications, patients' data, and intelligent literature access .
Knowledge services are the most advanced Grid services from the viewpoint of informatics. They are designed to supporting users in achieving their particular goals or objectives. They offer tools to improve with the way that knowledge is acquired, used, retrieved, published, or maintained. Knowledge is understood in a broad sense or as information applied to achieve a goal, solve a problem, or execute a decision. A characteristic example is data mining for automatically building a new knowledge. In tissue-based diagnosis it would be an appropriate tool in screening and evaluating virtual slides prior to be viewed by the pathologists, or to direct the clinician to providing pathologists with mandatory clinical information [8, 23, 62–64].
The Grid architecture consists of hardware and software that provide, control, and actualize the required functionality. It presents globally distributed resources, called the Grid fabric, as well as the Grid Middleware. Grid applications and portals to be accessible by the user can be considered the third element of the environment. Derived from these compartments four main aspects characterize a Grid in general:
Grid resources are geographically distributed and usually belong to different administrative domains and organizations. The autonomy of resource owners, their local resource management and usage policies have to be acknowledged. Their primary local function has not to be touched or even disturbed. It is quite rare that Grid resources only serve for an individual Grid; usually they provide primarily services which they have been designed to.
Grid resources are heterogeneous in nature and encompass multiple technologies. The more can be incorporated the more attractive the Grid becomes;
Open communication networks are dynamic. They might grow or shrink. The physical and functional communication channels can cause remarkable delay in information transfer and speed if a communication network expands to fast. The growth of a Grid cannot be foreseen, and might raise the problem of potential performance degradation as the size of Grids increases. Consequently, applications that require a large number of resources must be adequately designed.
The Grid Middleware provides capabilities to dynamically identify vacant and non-accessible resources and Workload Balancing ensures the efficient use of the accessible capacities.
Designing a Grid environment requires consideration of various designs to ensure the workflow and the long-term stability. For example, the definition of the information flow, supported communication protocols, file transfer technologies, networking technologies and bandwidths limitations, security and access control management etc have to be defined.
The implementation of a Grid is often the joint efforts several industrial partners and scientific institutions. These include, for example, NetSolve , Globus , or Legion . In diagnostic medicine, aspects of diagnostic accuracy and reliability have been in focus of Grid applications. For example, an Age-Related Eye Disease Study system for classifying age-related macular degeneration from stereoscopic color fundus photographs has been published in 2001 [67, 68]. Live imaging applied for functional brain analysis by magnetic resonance technique (MRI)  is also undertaken with Grid technology. Grid systems to compute patients' dose, image quality and system performance in cancer screening have been described . Bioinformatics Grids to be applied for analysis of genes and NDA sequences  are additional examples. In therapy, a new term called radio-surgery has been introduced to describe potential applications of Grid technology in surgical procedures . These examples indicate that Grids have emerged as a promising technology to handle large amounts of data and compute the specific medical requirements in radiology, bioinformatics, dermatology, and neurosurgery. Especially, digital medical image processing is a promising application area for Grids that try to fill the gap between the Grid middleware and the requirements of clinical applications. A Grid system (Grid Medical Archive System, GMAS) directed to share the access, storage and retrieval of digital images obtained in radiology, cardiology, and other medical live imaging departments enables the application of the common Picture Archive and Communication System (PACS) standard and other documentation systems to access fixed-content data including medical images and documents. An extension called Grid Medical Archive Solution Entry Edition (GMAS EE) has been designed for Regional Hospitals or live imaging departments within larger hospitals to reduce the entry price point for providers while still offering all the advantages of original Grid Medical Archive Solution (GMAS) solution . GMAS EE will allow Hospital Information Systems to share cardiology, radiology and other digital images across sites, and to safely store patient cases for years. This Grid is powered by IBM, and based upon Bycast StorageGRID software, a standard in grid-based fixed-content storage .
Recently, the European Community released a new Grid project, called ViroLab  or . This Grid is a joined venture of the following institutions: Universiteit van Amsterdam, Institute Universitair Medisch Centrum Utrecht, Institute of Computer Science AGH, Academic Computer Centre Cyfronet, Universita Degli Studi di Brescia, Universita Cattolica del Sacro Cuore, Institute de recerca de la SIDA, Katholieke Universiteit Leuven, Eotvos Lorand Tudomanyegetem, University College London, Virology Education B.V, and Universitaet Stuttgart. Its infrastructure has been designed and built by GridwiseTech, a company specialized in Grid computing . The official starting date of the project was March, 1, 2006. Virolab has been designed as virtual laboratory focusing on viral infections, especially HIV/AIDS. The Virtual Laboratory will include tools to submit data for statistical analysis, visualization, modelling and simulation. Access to patients' data and genetic information will allow clients to prognosticate the temporal virological and immunological response of viruses with complex mutation patterns to drug therapy.
Grid-powered image storage and retrieval systems based upon Picture Archive and Communication System (PACS) applications have been developed for live imaging, neurosurgery, or dermatology. Examples have been reported in [51, 72, 77–80]. In contrast to these reports, implementations of Grids to be applied in tissue-based diagnosis have not been published to our knowledge. There are descriptions of systems that automatically evaluate cytology smears [81–83], or automated measure DNA content or expression of antigens [9, 84, 85], however, these tools can only be considered as precursors and do not meet the performance of a Grid in general, as they are designed for one analysis system with open access.
The proposed Grid realizes a virtual pathology institution. It acts simultaneously as data source, data processing, and posting (i.e., diagnosis releasing) system. The released diagnosis depends significantly on small image areas that contain the "diagnosis clue". To create a reliable Grid resource to selecting these small image compartments is probably the most difficult task of the proposed Grid. Whether this algorithm can be based upon numerical procedures or has to rely on predefined image examples still remains an open question.
In aggregate, we are convinced that Grid technology will be implemented in diagnostic surgical pathology in the near future. The process of glass slide digitalization will open the door to combine all available information resources in order to furthermore establish tissue-based diagnosis in the medical environment as it is the most reliable and even cheapest diagnostic procedure in numerous and social important diseases, such as cancer or chronic inflammatory lesions.
The financial support of the International Academy of Telepathology and of the Verein zur Förderung des biologisch – technologischen Fortschritts in der Medizin e.V. are gratefully acknowledged.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.