Dentomaxillofacial Radiology (1998) 27, 358 362 1998 Stockton Press All rights reserved 0250 832X/98 $12.

2.00 http://www.stockton-press.co.uk/dmfr

TECHNICAL REPORT

Implementation of an oral and maxillofacial radiology image management and communication system
S-K Chen1,2, KM Yu3, R-S Chen1,2 and C-C Hsieh1
1

Department of Dentistry, College of Medicine, National Taiwan University; 2Department of Medical Informatics, College of Medicine, National Taiwan University; 3Department of Computer Science and Information Engineering, Chung Hua University, Taipli, Taiwan

Objectives: To implement an oral and maxillofacial radiology image management and communication system (OMFR IMACS) in the School of Dentistry, National Taiwan University. Methods: An OMFR IMACS has been implemented over 3 years in three phases: (1) Establishment of an OMFR image processing laboratory for research and graduate student training; (2) Installation of network infrastructure and (3) Total system integration. Results and conclusions: Intraoral, panoramic, cephalometric, and TMJ radiographs are directly processed with digital image acquisition devices or are converted into a digital format for viewing via UTP cable through FastEthernet in the eleven dierent divisions of the school. The system archives about 400 Mb of imaging data daily, representing 80% of the workload of the OMFR Division. The system is now used parallel to the conventional lm-library system and has been integrated as part of the oral health service. Keywords: radiology information systems; radiographic image enhancement; radiography, dental; digital radiology, dental Introduction Since early 1980s the potential of Picture Archiving and Communication Systems (PACS) for medical imaging has been widely investigated.1 17 The principal subsystems of a PACS are the infrastructure, the image acquisition system, the image display and output system, the database and storage system, and the communication and networking system. With the development of higher-speed computers, networking facilities, database structures and image display systems, PACS has been implemented in several hospitals, resulting in lmless medical radiology departments.18,19 While the reasons for developing PACS in medicine and dentistry are similar, the requirements dier significantly.20 We therefore propose to use the term oral and maxillofacial radiology image management and communication system (OMFR IMACS) as a better description of the system. Although the number of examinations performed in OMFR is usually larger than in most medical imaging procedures, the average image size is much smaller, resulting in less digital data. In this respect, OMFR IMACS should be easier to implement. On the other hand, OMFR images require high resolution and therefore are used for detecting subtle changes. These factors complicate the design. There are three possible methods for implementing a PACS: one is to purchase a turnkey system, the second is to contract with a system developer, while the third is to develop one's own.21 During the past 3 years, we have implemented our own OMFR IMACS in three phases and shifted from developing our own system to contracting with a system developer. First, starting in December 1994, we established an OMFR image processing laboratory for research and graduate student training, experimented with various types and models of equipment and contacted manufacturers. During this initial phase we evaluated various kinds of hardware and software and had the opportunity to examine a range of interface problems. Second, the network infrastructure was constructed between July and December 1996 with installation of image acquisition software and hardware. Finally, total system integration took place between January and March 1997. This paper describes the implementation of the entire system.

Correspondence to: S-K Chen, Department of Dentistry, National Taiwan University Hospital, No. 1, Chang-Te Street, Taipei, Taiwan Received 1 May 1998; accepted 23 June 1998

OMFR IMACS S-K Chen et al

System implementation The OMFR IMACS consists of ve major components: (1) infrastructure, (2) image acquisition station, (3) image display and output station, (4) image database and (5) networking system. Infrastructure The architecture of the infrastructure of our system is depicted in Figure 1. Open architecture, reliability, security and standardization are the key characteristics. We used an optical ber to connect two 100-Base Fast Ethernet networks located in two separate buildings. Two-star network topology is used in each building to provide a fast response. The length of optical ber is 2000 m, providing a rate of data transfer of 200 Mb/s in the duplex connection mode. To integrate the system, we used NetBEUI (NetBIOS Enhanced User Interface) as well as the TCP/IP protocols to accommodate the Windows NT (Microsoft Corp., Redmond, WA, USA) operation system which guarantees the shortest communication time. For reliability data integrity is emphasized. To preserve data integrity, once an image has been acquired it is rst categorized and then stored in the Redundant Arrays of Inexpensive Disks (RAID). Both the image data and the patient database have a mirrored architecture using RAID (level 5); therefore, a single hard disk crash will not aect the data integrity. The infrastructure was designed as an open, dynamic modular structure so that it can be upgraded in the future without becoming obsolete. Provision was made for possible system expansion. Industrial standards (Windows NT 4.0 OS, TCP/IP communication protocol, SQL database query language, ACR-NEMA standard system reliability) were used whenever possible21 to stress data integrity and minimize system downtime. Image acquisition stations Digital image acquisition devices frequently used in an OMFR clinic can be classied into two categories: direct (CT, MRI, US, PPCR (Phototstimulate Phosphor Computed Radiography) and CCD-based intra-oral and extra-oral X-

ray devices), and indirect (primarily the lm-scanner system). An OMFR IMACS should be able to acquire images from these dierent modalities. The current system can meet this requirement by integrating the following devices: (a) A VXR-12 roller-type lm scanner. (Vidar System Corp., Herndon, VA, USA). (b) A RVG-PC (Trophy Radiologie, Vincennes Cedex, France) direct digital intra-oral system. (c) Two Digora (Soredex Orion Corporation, Helsinki, Finland) PPCR intra-oral X-ray systems. (d) An Orthophos DS Ceph (Siemens AG Dental Systems, Bensheim, Germany) direct digital panoramic and cephalometric system. A personal computer with Pentium 166 MHz CPU, 2.1 GB HDD, 32 Mb RAM hardware conguration forms the image acquisition station. The acquisition station acquires images from the devices listed above by either direct or indirect input. For direct input, the acquisition station invokes a daemon to acquire and compress images and then sends the patient data and image les to the database server and le server. For indirect input from the lm scanner, the acquisition station has a graphic user interface (GUI) which allows the user exibility for determining the type of template and which slot(s) in the template to ll (Figure 2). It then automatically cuts the image le. All the images acquired by present devices are automatically imported into the database with system application programs which were developed by the vendor (FirstInfo Corp., Hsing-Chu, Taiwan) with the specications dened by the OMFR division. In the lm-cutting process, a program is used to sort out the lm pockets lled with images. This program will rst calculate the left and bottom edges of the lm mounts. This is achieved rst by deciding the bottom left of the lm mount and then using rotation to t the edges. The positions of the lm slots are decided once these two edges have been determined. Many pixels are

359

Figure 1 Diagram showing the design of the infrastructure for the OMFR IMACS at the Dental School. Viewing stations are located in major clinical areas covering more than 100 dental cubicles (VXR-12 : lm scanner)

OMFR IMACS S-K Chen et al

360

Figure 2 Diagram showing the lm cutting areas enclosed in the dotted lines. Because position and orientation of lm mounts vary in each scan, positions of the lm slots need to be individually judged. Dotted lines surrounding each lm slot are decided by rst nding the left and bottom edges of the lm mount which are also shown

sampled for each slot and the mean is used to evaluate whether a lm is inserted into a specic slot. After the lm cutting procedure is complete, the acquisition software will compress the image les and insert the patient's data and image le into the database server and le server. A lossless compression algorithm with a compression ratio of about 1 : 3 is used. Because of the amount of data being processed, images are compressed immediately after acquisition and archived in compressed form. Image display and output station Image display is the most frequently-used interface in an OMFR IMACS. Most users interact with and evaluate the system through this function. Therefore, important design considerations include the speed of display, ease of operation of the display program, and image quality of the hardcopy output. We have developed a generalpurpose user-interface for all the present users in the school. The programmer should work with the enduser from each dental specialty to develop the most user-friendly interface for their needs. Video display monitors must be appropriate for the mode of viewing in each location. The CRT is less luminous than a viewbox, and its display homogeneity is worth noting. Although a CRT with a high resolution has a viewing area capable of displaying most dental X-ray images, ickering becomes a problem with a large CRT. We therefore chose a high luminance 21 in. One K monochrome monitors (M21LMAX, Image Systems, Hopkins, MN, USA) for use in the Oral Diagnosis, OMF Surgery, Operative Dentistry, Periodontic, Orthodontic and In-patient Departments where diagnostic quality is important.

For general purpose viewing in the Special Needs Clinic and the Endodontic, Prosthodontic, and Pedodontic Departments we chose 20 in. Super VGA monitors (20C, Philips Electronics Corp., Amsterdam, the Netherlands). To simplify the user interface, we selected only the most frequently used functions, image rotation, magnication, ltering, histogram equalization, contrast/brightness adjustments and length and area measurements. The viewing program automatically decompresses the images before viewing. For user convenience and ease of reference, our display stations provide seven templates for viewing. Currently, there are templates for full mouth surveys, periapical and bitewing radiographs, occlusal radiographs, various size of extra-oral radiographs and pedodontic images. Users can easily search for the desired images by selecting the appropriate template. This function can help users retrieve images in the same format as they are accustomed to conventionally. For ease in displaying the image and communicating with the database server, the display software was developed using Delphi 2.0 (Borland International Inc., Scotts Valley, CA., USA) with the NT workstation V4.0 operating system. The display workstation is built from a Pentium 133 MHz (Intel Corp., Hillsboro, OR, USA) personal computer with 32 Mb RAM, standard 104 keyboard logic, mouse, display card with non-interlaced 10246768 resolution and simultaneous display of 1.44 M colors, and a network communication card. Each display workstation has a local hard disk with 2.1 Gb memory that serves for short-term storage of the most recent examinations.

OMFR IMACS S-K Chen et al

Because of the limited availability of the image workstation at present, it is important to use a hardcopy output system to distribute the images. Thermal prints from video copy processor (P68U, Mitsubishi Electronics America Inc., Cypress, CA, USA) are used for most images. In some cases, digital images are output with a dry laser imaging system (Helios Model 810, Polaroid Medical Imaging Systems, Newton, MA, USA) to make them available on lm. Image database Requirements in the design of an OMFR image database are that the original data is unchanged, condential, and secure. Basic image data includes the patient's name, chart number, ID number, date of birth, date and modality of image acquisition, number of images, position of imaged area, exposure conditions, clinical ndings, radiological ndings and other reports. The digital image database is designed to have a owchart similar to that of the lm library system with which users are already familiar. It is important that the system is designed to make loss of image data impossible in view of the potential legal problems. Networking system Several networking protocols are available for image transmission, including Ethernet at 10 MbS71 (MBPS), FastEthernet at 100 MBPS, Asynchronous Transmission Mode (ATM) and Fiber Distributed Data Interface (FDDI). When digital dental X-ray images are transmitted, Ethernet is adequate. However, Fast Ethernet should be used if other data are to be transmitted with the digital image. System costs, maintenance and operation Total cost of the hardware was approximately USD 100 000, including the networking facilities and installation, twelve viewing stations, one server, and one archiving device. Software costs run at approximately USD 20 000. The system has been in operation since April 1997. The server is shut down for monthly maintenance but otherwise it runs continuously for 24 h a day, 7 days a week. The maintenance team checks the integrity of the data weekly by comparing it with the SQL data and makes a weekly round of the imaging stations to make sure they are operational. Maintenance costs are 3 h/ man/week for system maintenance and 2 h/man/week for image uploading. The installers of the system provided training for the clinicians on how to operate the imaging stations during the rst week the system was in operation. Thirty minute orientation sessions were conducted with ten users at a time. The short training periods were made possible by the simplied user interface. Discussion The system has been in daily use for over 6 months and we have been working on several problems in an eort to optimize the system.

Film scanning speed At present, an average of 315 images are being archived daily, that is 6615 a month. The upload time for a full mouth survey of 20 images is about 120 s, which represents a processing bottleneck for a busy OMFR clinic. This long wait results mainly from data compression and communication with the server database. We intend to change shortly from direct image uploading to batching to speed up the acquisition process. Space available for viewing stations The imaging stations are installed in the major clinical areas. Because of budgetary limitations, only one station can be installed among an average of ten cubicles. Because of their large size, it is dicult to select a convenient site for them. Also, since overly intense ambient light aects the reading of the image in our experience, the viewing station should be shielded. However, it is not always possible to nd a place where this can be accomplished. Since Liquid Crystal Display (LCD) technology has improved signicantly during the past few years, LCD monitors may be the best solution once they become aordable in a few years time. Hardcopy and output devices Although the soft copy images are available throughout most of the clinical areas, they can not be accessed in several sites where they are often needed such as classrooms, group meeting rooms and operating rooms. This problem will be solved if networking can be extended into these areas when LCD projectors become available. There is also a need to design special control devices to replace the mouse and keyboard so that viewing stations can be operated during surgery. Maintenance Although timely service is very important, no dedicated sta are currently available. Two radiographic technologists have to take part in rst-line maintenance, but they are not responsible for handling most of the system problems. Although sta from the vendor (FirstInfo Corp., Hsing-Chu, Taiwan) provide weekly maintenance, some system problems are not solved the same day that they occur. Therefore, dedicated sta within the department should be allocated for maintenance. The rst image management and communications system for dentomaxillofacial radiology was reported 5 years ago.20 This established the principles for the design and implementation. Because of technological advances in all areas of the OMFR IMACS, we have encountered new challenges during implementation of our system despite the use of more up-to-date equipment.

361

Acknowledgements This study was supported by research grants NSC-85-2331B-002-306 and NSC-86-2314-B-002-218.

OMFR IMACS S-K Chen et al

362

References
1. Hynes DM, Stevenson G, Nahmias C. Towards lmless and distance radiology. Lancet 1997; 350: 657 660. 2. Osteaux M, Van den Broeck R, Verhelle F, de Mey J. Picture archiving and communication system (PACS): a progressive approach with small systems. Eur J Radiol 1996; 22: 166 174. 3. Dwyer SJ, III. Imaging system architectures for picture archiving and communication systems. Radiol Clin North Am 1996; 34: 495 503. 4. Horii SC. Quality assurance for picture archiving and communication systems (PACS) and PACS technology applications in radiology. J Digi Imaging 1995; 8: 1 2. 5. Honeyman JC, Frost MM, Huda W, Loeer W, Ott M, Staab EV. Picture archiving and communications systems (PACS). Current Prob in Diagn Radiol 1994; 23: 101 158. 6. Meyer-Ebrecht D. Picture archiving and communication systems (PACS) for medical application. Int J Biomed Comput 1994; 35: 91 124. 7. Wiltgen M, Gell G, Graif E, Stubler S, Kainz A, Pitzler R. An integrated picture archiving and communications systemradiology information system in a radiological department. J Digit Imaging 1993; 6: 16 24. 8. Boehme JM, II, Choplin RH. Systems integration: requirements for a fully functioning electronic radiology department. Radiographics 1992; 12: 789 794. 9. Horii SC, Mun SK, Elliott LP et al. PACS clinical experience at Georgetown University. Int J Biomed Comput 1992; 30: 275 280. 10. Naylor AF. A strategy for the progressive introduction of picture archiving and communication systems for a radiology department. J Digit Imaging 1992; 5: 82 88. 11. Haynor DR, Smith DV, Park HW, Kim Y. Hardware and software requirements for a picture archiving and communication system's diagnostic workstations. J Digit Imaging 1992; 5: 107 117. 12. Huang HK, Taira RK. Infrastructure design of a picture archiving and communication system. AJR 1992; 158: 743 749. 13. Choplin RH, Boehme JM, II, Maynard CD. Picture archiving and communication systems: an overview. Radiographics 1992; 12: 127 129. 14. Horii SC, Mun SK, Levine B et al. PACS clinical experience at Georgetown University. Comput Med Imaging Graph 1991; 15: 183 190. 15. Stewart BK, Lou SL, Wong WK, Huang HK. An ultrafast network for communication of radiologic images. AJR 1991; 156: 835 839. 16. Taira RK, Huang HK. A picture archiving and communication system module for radiology. Computer Methods Programs Biomed 1989; 30: 229 237. 17. Didden HW, de Valk JP, Bakker AR. Top-down design of a picture archiving and communications system (PACS) by means of simulation. Computer Methods Programs Biomed 1988; 26: 85 95. 18. Mosser H, Urban M, Hruby W. Filmless digital radiology feasibility and 20 month experience in clinical routine. Med Inf (Lond) 1994; 19: 149 159. 19. Siegel E, Brown A. Preliminary impacts of PACS technology on radiology department operations. Proc Annu Symb Comput Applic Med Care 1994; 917 921. 20. Dove SB, McDavid WD, Welander U, Tronje G, Wilcox CD. Design and implementation of an image management and communications system (IMACS) for dentomaxillofacial radiology. Dentomaxillofac Radiol 1992; 21: 216 221. 21. Huang HK, Taira RK, Lou SL et al. Implementation of a largescale picture archiving and communication system. Comput Med Imaging Graph 1993; 17: 1 11.