Geographic, telecommunication and medical context, place this developing country in a unique position for telemedical network development (Versión en español diponible)
Beltrán Mena, MD.*, JoséBadía, MD.*, Marcelo NeiraÝ, Alejandro RíosÝ.
(* Catholic University of Chile School of Medicine. ÝCatholic University of Chile Service of Informatics and Communications)
Chile shares with the rest of the developing world problems such as scarcity of specialists, limited resources and their centralization. To these problems, the country adds others, such as its large extension, its isolation from the rest of the world, and the geographical difficulties of internal communications. Telemedicine is regarded as a solution for this kind of problems. This country finds itself in a unique position for leading this type of applications in the region.
Conscious of these circumstances, Catholic University of Chile has decided to impulse an evaluation project in telemedicine, initially focused towards radiology and pathology, and also includes applications in long distance education.
The project makes use of broad band digital networks (ATM), and in its experimental stage, establishes a connection between two hospitals located 10 miles apart. This relative closeness allows the best conditions for research in two areas:
a) comparison of traditional diagnosis and telediagnosis, and
b) the development of optimal methods in long distance collaboration.
After this stage is completed, the network will by extended to other cities, contributing to the development of a national telemedicine network by the end of the century.
Chile (Fig. 1) is separated from its neighbors by the Atacama desert, the Andes and the Pacific Ocean, all of which have historically conferred it an insular character. The country's configuration is extremely narrow, 2650 miles long from North to South, but only 110 miles wide on average. Moving from one place to another always involves a long journey. Patagonia, in the extreme South, is a disgregated territory separated into hundreds of islands and channels. With this geographic conditions, its evident that Chile's progress needs a solid development of its telecommunications. This has turned the country into the Latin American leader in this field.
Seven private companies which comprise the long distance telecommunication market have spawn a competition that has been translated into high technical standards and competitive prices. A strong annual investment has allowed the existence of more than 1.6 million lines in operation, which means 8 inhabitants per line (Latin American average is 20, while USA has 1.5), with a 100% digital telephone network.
The country has an extensive optic fiber network; at least three companies cover each the whole country. Santiago, the capital, has nearly complete coverage. In this regard, Chile's narrow configuration becomes an advantage, since a single backbone from North to South is enough to connect all main cities. No town is located more than 40 miles from a backbone. In 1997, when the Pacific cable reaches the city of Arica in the far North of Chile, the whole country will immediately be benefited.
Creating WANs is cheaper in Chile than in any other american country (Fig. 2), including USA. Complete ATM coverage of the Santiago area is expected towards the end of 1996 and by the same date, the country should complete its ISDN coverage. This means that, once evaluated, telemedicine could be immediately applied throughout the country.
In the following section we will try to explain the health situation in Chile, particularly in those aspects where telemedicine could prove useful.
Chile has 14 million inhabitants. Since the 50's, migration from the countryside to the cities has resulted in 16.6% of all Latin American population living in the capital cities. In contrast to USA where the population of its ten main cities is adds up to only 8% of the country's population.
In Chile's case, this centralization is especially serious (37.5% of the country's population lives in Santiago), being surpassed only by Uruguay (41.7%). However, Uruguay is a small country, so that centralization there could be considered an optimization of resources. Chile, instead, pays a high price for this situation.
Chile has a population density of only 18.3 inhabitants/Km2, similar to the South American average (17.4). This density is very irregular throughout the length of the country since 50% of the population lives in the central 5% of the land, and 10% lives in areas where the population density is less than 6 inhabitants per Km2.
Physicians are even more centralized than the general population. Out of the 15.451 medical doctors working in Chile, 60% practice in Santiago, where 40% of the population lives, while the rest of the regions suffer from a relative deficiency (Fig. 3).
Although Chile has a population of doctors (1/ 908 inhab) slightly better than the Latin American average of 1/1022 inhab (Fig. 4), the reality is that this relationship is much better than the Latin American average (1/ 624) in the capital and worse than it in the rest of the country. Of the 13 chilean administrative regions, only two outdo the Latin American average, while the rest do rather poorly. The VII th region, for instance, has an average of 1 physician per 2.113 inhabitants, which is as low as Bolivia and Guatemala.
Furthermore, the distribution of medical specialists throughout the country is sometimes more centralized than the general medical population (Fig. 5).
Santiago's adequate ratio of physicians per capita, does not necessarily guarantee an adequate medical coverage or a high quality clinical care since the physicians per capita ratio says nothing about the internal distribution of physicians in any given region and even less on their distribution with regard to the socioeconomic profile of the population. In fact, 90% of doctors' private offices are located in the high-income neighborhoods of Santiago.
In Santiago, a 20% deficit in hospital capacity has been estimated for the year 2000. In regard to ambulatory attention, at present, only 68% of the demand concerning acute illness and only 44% of chronic illness is being satisfied. This fact highlights the profound inequalities in the distribution of medical resources among regions and within each region. It is difficult for the Public Health System to compete with the financial reward offered by the private sector. The state is thus challenged to optimize its limited resources if it wants to satisfy the demands of every sector.
The solution to the excessive centralization of medical resources in the country implies radical cultural and economical changes. Telemedicine, however, is relatively easy to implement with readiness and could contribute to the solution of the problem.
Conscious of this, Catholic University of Chile has decided to start a test project in telemedicine, initially focused on the areas of pathology, radiology, and long distance education. The first stage is strictly experimental, it does not intend to solve medical care problems, but instead it is aimed at evaluating technology and defining applications, as well as deciding the best way of long distance collaboration. Subsequently, the expansion of the project to distant cities is contemplated, with the objective of contributing to the development of a national telemedical network toward the year 2000.
The participants in the mentioned project are:
a) Catholic University of Chile, through its faculties of Medicine and Engineering, and its Informatics and Communications Service (SECICO).
b) Ministry of Health, through its South-East Health Service, in Santiago.
c) Private companies: CTC (Chilean Telecommunications Company), Coasin (Newbridge Networks), Kodak and Tandem Chile (Fore Systems).
The Catholic University Hospital has already been connected with Sótero del Río Hospital, which is located 10 miles apart. This distance is particularly useful for our evaluation purposes. The participating hospitals are close enough to allow the easy transportation of patients and medical material when necessary. On the other hand, it is conveniently apart so as not to contaminate the working method (i.e. a doctor could be tempted to talk directly to his "remote" colleague, if he is in the room next door).
The University Hospital (500 beds, 300 medical staff), is one of the most advanced in the country. Sótero del Río Hospital (1000 beds, 400 medical staff), is one of Chile's main public hospitals, covering a population of 1.200.000. Each Pathology Department reports 12.000 biopsies per year. The 12 radiologists of the University Hospital inform 10.000 exams per year. Sótero del Río Hospital has 6 radiologists, who perform 8.500 exams per year. Public hospitals do not always have staff radiologists, and x-rays are frequently informed by non radiologists.
The decision of limiting the project's first stage to radiology and pathology was based on various factors:
a) They are specialties which have always been insufficient in the country.
b) Since neither specialty deals directly with patients in order to reach a diagnosis, the patients presence will not be missed by the expert. This should make easier the introduction of telemedicine in medical practice.
c) The interaction between the specialist and the long distance operator is relatively simple in both applications. The applications considered for the next stage (ultrasonography, endoscopy and endoscopic surgery) require an important training of the operator next to the patient.
An important factor for us, is that both applications are very demanding on both processing power and bandwidth. We decided to start with them, which will make it easier to implement the rest later on.
The workspace for pathology (Fig. 6) is made up of a Silicon Graphics workstation capable of digitizing and compressing video, where the S-Video signal coming from a camera mounted to a microscope is entered. A second camera is aimed at the user's face. The workstation is used to establish communication and during the video conference, it also provides a whiteboard that allows grabbing an image and drawing on it. The main image of the biopsy is shown on a second video monitor. The operator who handles the biopsy explains the clinical data to the remote pathologist and follows his verbal instructions to run over the specimen. The system works in both directions in the same way.
In the case of radiology, the working places of each hospital are different. We use the Kodak ImageLink system. At Sótero del Río Hospital, there is a workstation that controls a film scanner. These images are sent to the radiology informing office by local network (Ethernet). At the university hospital, the images are shown in a PDS Hi-Resolution workstation, which displays 2500 x 2000 pixels, 12 bit gray scale images. This machine has a large monitor and high luminance, and is considered by Kodak to be of diagnostic quality. The workstation at Sótero del Río Hospital is the PDS Medium Resolution, considered by the company to be of "clinical reference", since the monitor is smaller (1600 x 1000) and has lower luminance.
Apart from this first two clinical applications, two auditoriums, implemented for long distance lessons and clinical meetings, will be connected.
In the case of pathology, frozen sections (60) and conventional (120) biopsies will be analyzed. The frozen sections will be diagnosed in the standard way and afterwards be shown to the remote pathologist. The same sample &endash;normally fixed&endash;will be diagnosed again later on. The diagnostic error between contemporary and deffered biopsy; and later the difference between local and telediagnosis will be compared. In this way we will distinguish intrinsic medical error from the one due to telepathology. Also, two different compression levels will be used, using a double blind method.
In the conventional biopsy group, local and remote diagnosis will be compared, and the 120 cases will be divided in two groups with different compression levels.
For radiology, 200 osteo-articular trauma cases will be diagnosed with both the original film and by teleradiology, to measure the technique's error. Apart from optimal conditions testing (2500 pixels, 12 bits, 150 Mbps), lower speed networks (2 Mbps and 128 Kbps) will be simulated with PCs as workstations, using good quality out-of-the-shelf monitors.
The latter test aims to establish what can be expected with low-end solutions in conventional radiology, since the Ministry of Health has shown interest in these, running on N-ISDN.
In contrast to pathology, the emphasis of our study in teleradiology will be pointed towards operational aspects and economic feasibility, as the practical usefulness of teleradiology is sufficiently accepted.
We expect preliminary results in telepathology by march 1996, and by June 1996 in teleradiology.
Initially, neither hospital had an internal optic fiber network, so more than 1000 ms had to be installed. Some segments between the two hospitals lacked fiber optics too, so it had to be installed. The construction of a new subway line in the area delayed this work for some months.
While the inter-hospital fiber optic was being installed, the system was tested, both in-campus and from one part of the Santiago to another.
The most ambicious test was one across the Andes with Buenos Aires, Argentina. This connection last four days in August 1995, during which clinical cases were discussed between the chilean dermatopathologist Dr. Sergio González and his coleague from Argentina, Dr. Ignacio Calb, thus allowing the report of 30 biopsies with 100% diagnostic accuracy. Even though this trial was not formal, it was very encouraging, as the quality of the video (composite) was inferior to the one that will be used in the final system (S-Video).
During this trial period, different pieces of equipment and technologies have been tested.
We needed to configure a system capable of testing any telemedical application, that meant real-time audio and video at high resolutions, which require broad band networks. Over such a network, less demanding systems could also be simulated.
Apart from ATM, other wide band technologies exist. FDDI is usually mentioned, but ATM presents many advantages over FDDI in this type of project. Some of this advantages are:
a) ATM makes not difference between LAN and WAN, allowing their easy integration over considerable distances (even transcontinental).
b) Because of the structure and management of the packages it sends, FDDI is a solution aimed specially towards data communication. ATM, on the other hand, easily integrates data signals with time-sensitive signals (video, audio).
c) FDDI transmits at 100 Mbps on LANs over a maximum distance of 2 Km.
The final configuration is almost taken. Details on the remote education auditoriums remain to be defined. These auditoriums should be capable of: sending the teacher's image, showing the remote public's image to the teacher, and giving the teacher the means to use on-line audiovisual material. This design is not simple, because operation of the system must be straightforward. The topology we are testing is shown in Fig.7 .
For further information contact:
Dr. Beltrán Mena
Catholic University of Chile School of Medicine Marcoleta 352 Santiago / Chile Phone: (562) 6863811 Fax: 6331457
Silicon Graphics was chosen as our standard workstation, as it is an integrated platform for collaborative work, with a design that is very oriented towards graphics and video. It has available digitizing and compression cards, which allow information to leave the workstation already in digital form. The telepathology and auditorium workstations are Indigo2 models, configured with Indigo Video digitizer, Cosmo Compress JPEG and ATM OC3 cards (Fig.6). In teleradiology we are using Indy workstations, to allow videoconference, as this function is not supported by the Kodak system (Fig. 7).
Three models of ATM switches have been evaluated: Fore's ASX-200 and Newbridge's 36150 and Vivid. The former adapts better to our needs, as it has more ports, but its approach to video management is not optimal. The 36150 is an excellent access switch, allowing conversion of different protocols (E1,E3,T1, etc.) or services (Ethernet, JPEG) to ATM format, but it has few ports. The best present option seems to be Vivid (specially the WorkGroup switch, Yellow Ridge Ethernet hub and Route Server), due to its easy management, growth capacity satisfaction of the hospitals' requirement of Internet access and integration with the present Ethernet Networks. Our final decision between Fore's ASX-200 or the Vivid family will depend exclusively on financial considerations.
The protocol used is OC3 (155 Mbits) level for the internal ATM network of each hospital, the same is used for the monomode fiber between the hospitals, which will change to OC12 (622 Mbits) as demand increases. We have preferred ATM connections with SVC, which are automatically set with the application being used, to PVC, which must be manually disconnected by a manager.
In telepathology, large compression rates can be obtained during the time the image remains still. We have tried 2 Mbps connections with acceptable quality. Nevertheless, we would recommend a compression limit of 10 Mbps. There will be a final recommendation once the protocol is finished. In ultrasonography tests, we have worked well with 2 Mb channels. We have not try endoscopic surgery yet, but will try larger channels than telepathology, allowing 30 fps, with a minimum of 10 Mbps.
The format for radiological image storage will comply with DICOM 3.