Clinical Communication

Low-cost teleradiology for dentistry

Yoshihiko Hayakawa*/Allan G. Fannan*/Francisco E. Eraso*/Kmya Kuroyanagi**

Abstract The performance of a low-cosi teleradiology system was studied The time neededfor radiographie transmission between Computers and the image integrity followingtransmission were measured. Tlie image resolution was analyzed with a line-pairtest chart, ¡mages were transmitted through computers that had a video-processingboard and a modem ai both the transmission and receiving sites. Intraoralradiographs \iere captured with a black and white charge-coupied devicecamera. The time required for image transmission was less than 1 minute (46 to 56seconds), an effective transmission speed of J. 73 kbyte/s (13.9 kbit/s). Nochanges were obser\-ed in pixel value distributions; hence, there was no loss ofimage derail. The maximal resolution of ¡he system was 4 line-pairs/mm. Theperformance of the teleradiology system demonstrated its potential as an effective,low-cost telenetwork for dentistry (Quintessence Int ¡996:27:175-¡7S.)

Introduction

Like other areas of telecommunication, "teledentistry"is at the beginning of a period of rapid growth.'"* Thisrapid expansion of installation and utilization is beingdriven by increasingly low costs, improved hardware,and innovations in software design that have greatlyimproved both the speed and convenience of image-transmission systems.

In the dental office, the speed of adoption ofdigitized images will be dictated by convenience andthe increasing desirability of simplified storage andretrieval of whole patient charts, including dentalradiographs.'"* A local area network can interconnectmultiple sites to allow access to such images. There isalso the possibility of teledentistry via a dedicated orstandard switched wide area network.'"*

Among the uses of telecommunications in dentistryare diagnostic consultations, quality-assurance evalua-

' Division of Radiology and Imaging Sciences, University of Louis-vilie. School of Dentistry, Louisville. Kentucky.

* Department of Oral and Maxillofacial Radioiogy. Tokyo DentalCollege, Chiba. Japan.

Reprint requests: Dr Yoshihilto Hayakawa, Division of Radiology andlinaging Sciences, University of Louisville, School of Dentis(r>', Louis-ville. Kentucky 40292.

This project was funded by the WHAS Crusade for Children grant94-140.

tions, and long distance learning. A telecommunica-tion system could link private dental offices in remoteand semiremote areas with expertise in more developedsites. Digital images are easily exchanged betweenusers of a telecommunication system. The network alsoprovides the ability to connect general practitionersand specialists with large third-party dental insurancecarriers to expedite prior approval of treatment andreimbursements following proof of treatment rendered.

The purpose of the present study was to assess theperformance of an inexpensive teledentistry system.Dental Link (EScan). The time needed for imagetransmission was measured, and the pixel valuedistribution in images before and after transmissionwas compared.

Method and materials

System configuration

Two software programs. Dental Link for Windows,version 3.1, and Qmodem Pro for Windows, version1.11 (Mustang Software), were installed ¡n an IBM-compatible personal computer operated by a Pentiumprocessor (Intel). A digital video-processing board,(SnapPlus, Cardinal Technologies) and a high-speedfax and modem (Zoltrix) were installed. Images weredisplayed on a monitor (CTX International) using avideo-display card.

Oiiintessence interna'j'^nrll " " i ' " " " '^^ Nin 175

Hayakawa et al

Fig 1 Sampie of the digitized images on tine monitor.These images are composed of a 635 x 475 or 475 x 635pixel matrix, images a, b. and c correspond to histograms inFig 3.

Fig 2 (right) Reiationship between image-tile volume andthe time required for image transmission.

O)

Ë- 5 0o'(/)

i 45cH /

/

7^

70 75 80 85 90 95 100

Image-file volume (kbyte)

Several kinds of templates were available for input-ting intraoral radiographs into the computer withDental Link software.̂ The template could be in theformat of either a singie radiograph or several radio-graphs "mounted" together The one template used inthis study was the complete-mouth series. This formatis composed of 20 images, namely eight periapicalimages from the molar and premolar regions, eightperiapical images in the incisor and canine regions,and four bitewing radiographs.

Image capture and transmission

Twenty intraoral radiographs were digitized for inputinto the computer with a black and white charge-coupled device ÍCCD) solid-state camera (Cohu),which has 301,625 pixels (635 x 475). The distancefrom the CCD camera to the film was determined toobtain an imaging area of 41 x 31 mm. As a result, eachpixel was 65 X 65 |j.m. The image that was captured bythis CCD camera was converted into an 8-bit digitizedimage using a Tagged Image File Format (TIFF) imageformat. According to the User's Guide,^ in DentalLink version 3,1, the compression scheme used isLimply-Zig-Welsh, which is lossless and yields com-pression rates between 1:1 and 3;1; in the averageimage, the rate is in the neighborhood of 1,9:1.

Images that were ready for transmission were copiedto the export directory with the Dental Link and

transmitted to the other computer with the QmodemPro,*̂ The images were sent to a sitnilar software andhardware setup.

Image analysis

Before and after transmission images were convertedinto Macintosh (Apple) format to allow analysis of thehistogram, average, and standard deviation of pixelvalues with a Photoshop version 2,5 (Adobe Systems),

Image resolution was analyzed with a line-pair testchart. Funk Type 1 (Kasei Optronics), which couldmeasure from 0.5 to 10,0 line-pairs/mm. The visibilityof test chart images was recorded. Maximal resolutionwas visually decided in vertical, horizontal, anddiagonal alignments.

Results

Figure 1 shows a sample of the digitized imagesdisplayed on the monitor. These images were com-posed of a 635 x 475 or a 475 x 635 pixel matrix, Iftheimage were saved without compression, the filevolume would be more than 300 kbyte. Ail image fileswere in the range of 78 to 97 kbyte in a TIFF format.

Figure 2 shows the relationship between the timerequired for image transmission and the volume of theimage file. The length of transmission increasedproportionately to the volume of the file. The time

176 Quintessence International Volume 27, Number 3/1996

H aya ka wa et al

needed for image transmission ranged from 46 to 56seconds, depending on the file's volume. For theimages shown in Fig la to c, the elapsed titnes were 50,49, and 55 seconds, respectively, resulting in atransmission speed of 1,73 kbyte/s (13.9 kbit/s).

The pixel value distributions in Fig 3 correspond tothe images in Fig I, No differences were obsetredbetween before- and after-transtnission distributions ofany of the radiographs.

Different orientations of the resolution test chartresulted in slightly different fmdings. The maximalresolution of the system was 4 line-pairs/mm at thevertical alignment. Visibility was slightly better at thevertical alignment than at the horizontal alignment.

Discussion

The performance of the telecommunicadon systemstudied was remarkably good, especially in view of theinexpensive technology employed. This network usedswitched dial-up telephone lines, avoiding the expenseof dedicated lines, a PC-based platform, and a modem,A time of less than 1 minute to transmit a single dentairadiograph on analog lines is probably sufficient formost dental practices. All images were composed of635 X 475 pixels, and the grays scale was 8 bit,

Farman et al' reported that images were transferredboth transcontinentally and intercontinentally in 4,6seconds through an integrated service digital networkfiber-optic line at switch-56 rate, but these were32-kbyte Radiovisiography (RVG) 320OÛ tmages(Trophy Radiologie), They also reported that thefastest transmission rate obtained when File TransferProtocol and the Internet were used was 20 seconds;however, transmission speeds varied with the traffic inthe system.

The current development of digital intraoral im-aging systems is rapid, and file volumes continue toincrease. The image file volumes of the ComputedDental Radiography system (Schick Technologies)and RVG-4 system (Trophy Radiology) are 399 kbyteand 419 kbyte, respectively The Dental Link softwaresupports input of images obtained by digital intraoralimagitig systems (for example, Sens-A-Ray, RegamMedical Systems, which produces image files of 223kbytes without compression, and the RVG-4). It willalso accommodate inputs from S-video and NTSCcolor intraoral videocameras but these files are stillIaiger, With use of standard analog switched-telephonelines, it is evident that image compression tools will beessenrial for the practical use of teledentistry.

No differences were found between before- and

350003000025000200001500010000 -5000 -

00 50 100 150 200 Z50

3500030000250O02000015000100005000

0100 150 200Pixel value

250

Fig 3 Pixel value distributions corresponding to theimages in Fig 1. The x-axis is the gray level in the range of 0(white) to 255 (black) The y-axis is the number of pixels ateach gray level. The total number of pixels is 301,625,

after-transmission pixel value distributions, indicatingthe potential reliability of this system. This discretedistribution indicates that the image is in a 6-bit mode;hence only 64 distinct gray values are distributedevenly over the 256 gray values," This is essentially thesame as the method used with the Visualix/VIXA(Gendex Dental Systems), a typical digital intraoralimaging system,' which uses an 8-bit analog-to-digitalconverter but displays the image in a 6-bit mode. Passet al" reporied that no statisrically significant differ-ences were found between 6-bit and S-bit displays ofperiapical lesions.

Quiniessence Infprnpitinnal Voiume27, Number 3/1996 177

Hayakawa et al

Because each pixel is square (65 x 65 |im), the testchart visibility of the vertical and horizontal align-ments should be equal. The visibility obtained at thevertical alignment was, however, slightly better thanthat at the horizontal aiignment. The difference in thevisibility could be caused by monitor characteristics,such as horizontal scanning lines. In the case ofthediagonal alignment, the image resolution degeneratedas would be expected because ofthe square pixel.

A videocamera was used to acquire images in thepresent study. The contrast and resolution of imagescaptured by a videocamera can be altered detrimentally.''''The image quality, however, has been found to bestLÍflcient for diagnosis.'

The Dental Link system used supports the directacquisition of images with either the Sens-A-Ray orRVG-4 system.̂ '* We recommend the use ofthe directdigital radiography system with the image-transmis-sion system. With use of such a system, images withhigher resolution can be acquired.'"

Picture archiving and communicating systems andimage (or information) archiving and communicatingsystems are commonplace within large hospitals, and itis possible to communicate any necessary information,provided that the band of the connecting channel issufficiently wide. Problems of cost and channelavailability have, however, precluded connection be-tween hospitals and individual practices in remoteareas.

In Kentucky, a network serving pédiatrie carepersonnel has been effected among the University of

Louisville, the Kosair Children's Hospital, and mdivi-diial practices in Appalachia and southern and westernregions. The network uses the EScan system descnhed

in this communication. '

References

1. FarmanACFaraeA, Yeap p. Communication in digital radiûlûBy,Dentomaxillofac Radiol I992;2h?l3-215.

2. Farman AG, Farag A, Yeap P. Teleradiology of direct digital denialimages for consultations and treatment prior approval. SympComput Assisted Radiol 1992.1 M93-I99.

3. Farman AG, Farag A. Teleradiology for dentistry. Dent Clin NorthAm i993iÍ7:67-81.

4. Farman AG. Digital imaging in craniomandibular practice. JCraniomandib Pract 1993;II;161-16J.

5. Dental Link for Windows, version 3.0. LJser's Guide. Santa Rosa.CA; EScan, 1994.

6 Däntal Link for Wndows. version 3.1, User's Guide Addendum.Santa Rosa, CA: EScan. 1995.

7. Molteni R. Direct digital dental x-ray imaging with Visualix/VIXA.Oral Surg Oral Med Oral Pathoi 1993^76:235-243.

8. Pass B, Furkart AJ, Dove SB. McDavid WD. Gregson PH. Six-bitand H-bi[ digital radiography for detecting simulated periodontallesions. Oral Surg Oral Med Oral Pattioi I994;77:4O6-411.

9. Weniel A. Influence ofcompulerized in&irmaliun technologies onimage quality in dental radiographs. Dan Dent J 1991;95T527-559.

10. Wenzel A, Gröndalil H-G. Direct digital radiography in the dentaloRlce. lnt Dent J l995;4;:27-34.

11. Farman AG. Scarfe WC, Hunier NL. Kentucky pédiatrie teledentis-iry network. In: Lemke HLI. Inamura K, Jaffe CC. Vannier MW(eds). Proceedings ofthe International Symposium on Computerand Communication Systems for Image Cjuided Diagnosis andTherapy. Berlins Springer. 1995:13-24. D

26th International Meetingon Dental Implants and Transplants

Bologna (Italy), June 7-9, 1996

Information: G.I.S.I. c/o Prof. G. Muratori1, ViaS. Gen/asio, 40121 Bologna (Italy), Tel. 39 51/22 75 05-23 75 16, Fax 39 51/26 00 31

178 Quintessence International Volume 27, Number 3/1996