|INABIS '98 Home Page||Your Poster Session||Related Symposia & Posters||Plenary Sessions||Exhibitors' Foyer||Personal Itinerary||New Search|
The operator has the capability of viewing a maximum of four separate images that can derive from either the active scanner, disk file, or internet sources. Any one image can be compared with another by using a variety of user-configurable algorithms built in to the application edge enhancement, area texture analysis, Fourier, Gabor,
wavelet, and neural network tools are available for use with the image as a whole or for a user-selected rectangular region. The primary use of this tool is to enable the operator to rapidly isolate and identify interesting features and to bring out highlights in images while the medical procedure is ongoing. Once again, the tool can be used with new images that have been acquired at the point of current medical service, and this is not limited to images that are already in some large database.
A simple example illustrates the value of having in-field, real-time analytical tools for image comparison. Consider a victim of acute anterior wall myocardial infarction who is being cared for first by EMT, then in a small hospital or clinic ER, then in an ICU, then moved to a larger regional hospital for bypass surgery. During any time interval of interest, ranging from a few hours after onset to the entire period from t0 (last visit to physician) to t5 (pre-surgery) there may be benefits from viewing two or more EKG readings together, side by side. It may difficult to manage physically to obtain the desired EKGs at a moments notice, and especially difficult to work with them as physical pieces of paper.
Figure 8 : 2-day Interval EKG, Acute Myocardial Infarction ()
However, consider that all these EKGs have been input (direct-digital or by scanning) into a database. Using MediLink, the physician can call up arbitrary EKGs and drag-and-drop them into the Verité windows, set up as two or four to the screen, as shown in Figure 8 above. Interactive testing windows can be moved over two or more images and a comparison of signal patterns can be generated, with graphical results visible to the user in the "base" window of choice. In Figure 8 two comparison regions are being employed simultaneously, indicated by red and blue dashed-line boxes. In the final release version of Verité, the user will be able to obtain a pop-up table comparing selected regions by histogram analysis, wavelet, FFT, Gabor, and other measures and also will be able to overlay one signal segment over another for visual comparison of similarities and differences.
TransPAC uses a complete speech-to-text-to-database command interface that supplements rather than replaces the pen/keyboard/mouse channel. This speech interface provides for a trained vocabulary of approximately 100 words that are speaker-independent and resilient to external noise and in particular speaker accent, tone, and volume variations and also non-verbal noise such as that from operating machinery. The software is used within a widely-accepted transportation data collection product for highway and urban roadside asset data collection and has demonstrated the rigors of tests with variable speakers and noise levels. Commands include both single words and phrases, and include amongst the vocabulary the following elements (besides menu items):
All of these are intended to be speaker-independent and use a different speech algorithm than the generic speech tool for data entry, also included in the repertoire of MediLink tools. The latter requires individual training and is intended for when the medical practitioner wants to make extended notes or document entry by speech. Its functions are quite similar to those of many commercial speech recognition tools.
4.3 Global Positioning
For certain remote or emergency contexts, it may be necessary to engage the use of a global positioning system or GPS. TransPAC is designed to accommodate any one of several units, and the command speech interface software is already established to handle GPS inputs. The role of the GPS depends upon the medical application. For certain outdoor inspection tasks it may be necessary to reference the location of the patient and other objects. This could be the case in several accident or disaster related situations. A variety of GPS units with as much as sub-meter accuracy may be employed with TransPAC. One such system is the Trimble Pro XRS 1m real-time or post-process 12-channel GPS/DGPS receiver and antenna which is typically worn by the operator in a convenient backpack with no interference to physical movement or the operation of the TransPAC and other modules.
The role of the 16-bit microprocessor smart card (3KB EEPROM, 6KB ROM, 16KB on-card RAM) within TransPAC is twofold. First it serves as a compact and reliable form of access security for the system, identifying the operator and thereby setting up all access parameters for online network or internet linkages while the TransPAC is being used on an assignment. The issue of security and traceability is of paramount importance from the larger-scale systems engineering and business process perspective in that a widely-deployed, operator-intensive medical activity puts more weight and responsibility on the persons doing the tests and demands more accountability than activities which heretofore may have depended upon special planning and a special team of experts. As the operation of mobile computing and paper-free documentation becomes more ubiquitous, the risks of human organizational error increase and this form of security, already built into TransPAC for other applications, is an apt response. Currently the TransPAC has an industry-standard PCMCIA Type II interface for the smart card device. The card remains in the unit during all times of operation and its removal would constitute a security violation for which MediLink could, if desired by the hospital administrators and physicians, shut itself down automatically to prevent data security violations.
There is a second role to the smart card, again one of enhancing the system of operation. The 16K of application-accessible memory on the card is divided into two sections: (1) Upload and (2) Download. The Upload section acts as a memo pad for instructional data and pointers for the operator regarding the specific testing assignment at hand. It stores the URLs of reference files, generally large (> 1 MB) files that may need to be accessed by the operator while performing procedures in the field. These can be obtained through either modem, LAN, or more often that not, the wireless connectivity over the internet to a central server. TransPAC is designed to interface specifically with Bentley Systems ModelServer Discovery, an application expressly designed for managing the retrieval and use of Microstation and other CAD format files over intranets and the commercial internet. However, TransPAC will also interface to any standard web server that will provide natively or through plug-ins JPEG, CGM, SVF and other format files for display, without application interaction, on a web browser provided as part of the MediLink software.
This capability of accessing a large MRI or Ultrasound file that is either already loaded onto TransPACs hard disk or else obtained ad hoc during a task via the internet allows the care provider to have a remarkably more versatile control over how tests are performed. If it is necessary to refer to a particular x-ray or MRI image while performing a procedure, it is only a matter of seconds away. LAN-based access tests show a response time averaging 5 sec. for drawings obtained via ModelServer Discovery. Wireless connectivity has come of age for the purposes needed by mobile medicine. Operators can make annotations to an image file for reference by other real-time or at a future date. A user makes use of reference points that are accessible in tabular form to the operator and can be entered, through the speech interface, into the transaction database recording all inspection activity. The reference points are read by the operator from the TransPAC display and appropriate entered verbally (or by pen input) into the data record for each imaging task.
Whereas the Upload section of the smart card memory serves to bring useful data to the operator during a task, the Download section is reserved for transaction recording that will preserve on the card with without possible erasure during a session. Each session record that is entered into the database on the TransPAC with images, location data, operator comments, and so forth, has an encapsulated summary record created at the time the MediLink application writes the record into its onboard database. This encapsulation includes a time stamp, location information, and a compressed-text summary of the recorded information about the action performed by the care provider but without any actual large text or image data. This encapsulation record has a format similar to that shown in Table 3 below. Each transaction record will on average occupy less than 50 bytes due to the encoding scheme employed that uses one-byte and two-byte codes for a variety of words and strings. There is only one primary transaction record (PTR) per work session. Numeric data is accommodated by integer and real representation. Future smart cards will have additional memory of upwards of 64K for Upload and Download purposes.
PTR (Primary Transaction Record)
|Unique key||alphanumeric string|
|TCR Field List||alphanumeric string; field pointers separated by delimiters|
|optional other task-defining fields||(optional)|
TCR (Transaction Content Record)
|PTR key||alphanumeric string||pointer to the associated PTR record|
|file list||linked list; e.g. (see below)||(link/field ID + file pointer + locator in file)|
|---||link/field1||text memo in MEMO.XXX, loc 001|
|---||link/field2||still photo in PHOTO1.YYY|
|---||link/field3||text memo in MEMO.XXX, loc 002|
|---||link/field4||sketch in DRAW01.ZZZ|
|---||link/field (n)||video clip in VID01.XXX|
Table 3 --- Basic Data Structure with IDEA Prototype
There are two main purposes underlying the Download operation of encapsulated data. First this provides a secure record of the work performed which cannot be altered on the smart card once entered, except by an optional override that itself stamps the smart card with a recording of that override. This is for data security and consistency. Second this offers a fast-track access for a colleague or expert who may be evaluating the tests and procedures done in the field on a real-time basis. A fast and concise record of what was inspected can be gained by anyone connected via a network to a server receiving the primary transaction and transaction content records from the inspectors smart card once it is entered into the reader at the hospital or clinic. An evaluating physician could easily browse through this data to ascertain if all tests were performed and if additional data should be collected, even before examining in detail the images collected. Such a quick review could determine if it is necessary to look at all images, or which ones should be reviewed or forwarded.
4.4.1 Patient Medical Cards
TransPAC is designed to be configurable for either of two types of operation with respect to smart cards:
In the former instance, a single card reader occupies one of the PC card slots. In the latter case a dual reader is used. In this case the Active Session Card is inserted and remains in the TransPAC for the duration of the sessions for one or more patients. However a patient card similar to the will be inserted and removed for each patient that is seen.
MediLink is configurable at installation time to handle either type of configuration. The basic sequence of operations is as follows:
For each Session, prompt user for Patient Card
If no Patient Card detected, prompt user to verify and continue
(some patients may not have cards assigned or present)
If Patient Card detected, load medical records data into system and proceed
Operate in single-card (Active Session Card only) mode.
5. Applications for Emergency and Remote-Site Medicine
Equipped with MediLink on the TransPAC, a medical practitioner has the basic toolset for enhanced communications between any remote, in-the-field site where either scheduled or unscheduled (i.e., emergency) treatment may be necessary, and resource centers that can be reached through standard and cellular communications for voice telephony and data transfer. The practical usage may be illustrated through the following scenario.
Patient MK lives in a small village approximately 50 km from a regional hospital (A) equipped for full birthing services. The roads are secondary quality and there are poor driving conditions. Her family physician lives in another town 30 km equidistant to MKs home and to the hospital. Her obstetrician lives 30 km from the hospital and 80 km from MKs home. A smaller clinic with ER facilities is 25 km away. MK is in her 32nd week of pregnancy and has been experiencing a variety of discomforts that increase to the point of concern for her health and that of the fetus. MK has been advised to maintain bed rest at home and to avoid any unnecessary or strenuous movement including travel by automobile.
MK experiences a fainting spell and her neighbor dials 911 for emergency assistance. Upon arrival the EMT has concern about the best procedure to follow since there are several mitigating factors, the treatment of one possibly causing aggravation and life-threatening danger for the fetus. Immediate movement may be ill-advised. It is believed possible from initial examination that MK is going into labor and there is concern over the position of the fetus in the birth canal.
The TransPAC is in the EMT vehicle, normally in a docking station cage where it can be used within the vehicle. With the TransPAC set up in the home at MKs bedside, an EMT can record all data that is being collected from the patient, including not only electronic data but verbal responses as well. All of this information can be made available real-time or "VNRL" (very-nearly real-time) to the obstetrician and/or family physician or their most available staff assistant. The EMT can use the MediLink phone service to make the call directly to one of the physicians.
Consider that the EMT has a portable fetal heart monitor at the patient site. The exact image pattern of the data acquired can be transmitted directly to a web server site that is at the hospital, thence accessible to the obstetrician or her on-call assistant, while being visible to the EMT on the TransPAC display. Other recorded data (e.g., MKs pulse, blood pressure) can be streamed via the internet without distracting interaction steps by the EMT. The only element that is clearly not reproducible (and the value here of course is immensely important) is the actual physical content touch, sight, smell, etc. that can only come from the remote physician and the patient being literally in the same room. However, the EMT is personally there and can describe things in answer to the right questions from the obstetrician or the family physician, and those questions can be more intelligently tailored and phrased given the direct access to so much more data through the MediLink connection.
In this example scenario, a portable ultrasound may also be present. The video image taken by the instrument is input directly to the TransPAC and compressed by the wavelet algorithms into nearly lossless streams that encode one or more video clips. These can be automatically transmitted as "clips" over the internet connection, aided significantly in speed if the TransPAC can be plugged in to a standard phone line and use a PCMCIA standard modem; otherwise the wireless link will suffice at approx. 24K 28.8K bps. One further feature of the TransPAC is the seamless "invisibility for making and breaking connections the user does not need to get involved in reconfiguring anything about the network, modem, or dial-up features. This is all programmed in advance and through the data that is on the user smart card, the Active Session Card.
The monitor and ultrasound data, in the hands of the obstetrician as things are proceeding, can assist in the evaluation and decision process. It can mean a difference between an emergency trip to the nearest hospital, or a further bed rest and a home visit by the obstetrician. Even while enroute in the ambulance, real-time fetal heart monitor data transmitted to the consulting physician can aid in determining whether to take MK to the nearest clinic, suitable for critical emergencies, or to the more distant hospital for a guided labor and natural birth.
6. Future Directions
TransPAC and MediLink have been designed and both are in the implementation and testing stages. Custom-tailored systems based upon the Mentis and ViA platforms along with some components of MediLink are being developed for clients. A survey concerning the use of different elements found in the TransPAC and MediLink designs within the global medical community is planned for 1999, and availability of MediLink, Version 1.0 for beta testing is targeted for the second quarter of 1999. There are several studies that can be performed to evaluate the effectiveness of the TransPAC-MediLink system in particular for emergency medicine, military medicine, home health care, travelers medical emergencies, and rural medical practice. It has been an aim of the present undertaking to lay the groundwork for a collaborative project involving medical practitioners from several disciplines and regions, worldwide. The aim of this project will be to evaluate the use of such mobile technology under diverse population and cultural conditions in order to refine the functions and the devices, including both hardware and software, for more convenient and practical use by the maximum number of medical professionals.
Sample MRI image in figure 4:
Courtesy Harvard Medical School, Whole Brain Atlas, www.med.harvard.edu/AANLIB/home.html. Site administered by Drs. Keith Johnson, M.D. (Harvard) and J. Alex Becker (MIT).
Sample EKG images in figures 5, 9:
Courtesy Emergency Medicine and Primary Care web site, www.embbs.com. Images contributed by Dr. H. Nussbaum and Dr. G. Fink; site administered by Drs. Ash Nashed, M.D. and Glenn Fink, M.D.
Sample ultrasound images in Figure 5:
Courtesy General Electric Ultrasound. Produced with the LOGIQ 700 MR 3D Ultrasound. www.ge.com/medical/ultrasound/msul7im3.htm. Site maintained by GE Medical Systems.
This work was conducted as internal research and development by Silicon Dominion Computing, Inc. of Richmond, Virginia.
Allen, A., Cristoferi, A., Campana, S., and Grimaldi, A. (Dec. 1997) An Italian telephone-mediated home monitoring service, TeSAN personal emergency response system & teleservices : 25,33
Dudziak, M. and Koval, T. (Nov. 1998), Developing a Practical Wearable Telemedicine System for Emergency and Mobile Medicine, G7 Global Impact on Health Care of Telemedicine Symposium, Univ. of Regensburg, Germany, Nov. 21-23, 1998
Dudziak, M. and Koval, T. (July 1996) CommonHealthNet as an Architecture for Practical Global Medicine, Global Telemedicine and Federal Technologies Conference, Williamsburg, VA
Dudziak, M. and Koval, T. (Oct. 1995), IMEDNET : Information Gateways and Smart Databases for Telemedicine and Distance-Based Learning Using the Internet and the WWW, Internet Medicine Conference, Chicago, Oct. 23-24, 1995
Grad, O., Hassol, A., Gaumer, G., Mintzer, C., Puskin, D., and Grigsby, J. (Apr/May 1997) Rural teleradiology: a snapshot, Administrative Radiology 16(4-6) : 28-33
Kukkonen, C.A. (June 1995) NASA high performance computing, communications, image processing, and data visualization-potential applications to medicine, Journal of Medical Systems 19(3) : 263-273.
Nordberg, M. (1996) Remote control. Telemedicine revolutionizes EMS in rural America, Emergency Medical Services 25(8) : 39,41,43-45.
Otto, C., and Pipe, A. (1997) Remote, mobile telemedicine: The satellite transmission of medical data from Mount Logan, Journal of Telemedicine and Telecare 3(Suppl 1): 84-85
Rosen, E. (Oct 1997) Mobile telemedicine arrives, Telemedicine Today 5(5) : 14, 42,44
Takizawa, M., Sone, S., Takashima, S., Feng, L., Maruyama, Y., Hasegawa, M., Hanamura, K., and Asakura, K. (1998) The mobile hospital--an experimental telemedicine system for the early detection of disease, Journal of Telemedicine and Telecare 4(3) : 146-51
Thorborg, S. (1990) Mobimed; a telemedicine system for mobile monitoring of physiological parameters, Proceedings of SPIE; The International Society for Optical Engineering 1355 : 32-33
Shimizu, K., Matsuda, S., and Miyanaga, T. (1994) Optical biotelemetry.(Yamamoto K , ed.) IEEE Instrumentation and Measurement Technology Conference, IEEE Press, New York, NY: 579-582
Stevens, L. (Apr 1998) Has Internet-based telemedicine finally come of age?, Medicine on the Net : 8-14
| Discussion Board | Back to Beginning Page | Your Poster Session |