Use of remote video microscopy (telepathology) as an adjunct to neurosurgical frozen section consultation

Use of Remote Video Microscopy (Telepathology) as an Adjunct to Neurosurgical Frozen Section Consultation R. L. BECKER, JR, MD, PHD, C. S. SPECHT, MD,* R. JONES, MD, M. E. RUEDA-PEDRAZA, MD, AND T. J. O’LEARY, MD, PHD We investigated the use of remote video microscopy (telepathology) to assist in the diagnosis of 52 neurosurgical frozen section cases. The TelMed system (Discovery Medical Systems, Overland Park, KS), in which the referring pathologist selects appropriate fields for transmission to the consultant, was used for the study. There was a high degree of concordance between the diagnosis rendered on the basis of transmitted video images and that rendered on the basis of direct evaluation of frozen sections; however, in seven cases there was substantial disagreement. Remote evaluation was associated with a more rapid consultation from the standpoint of the consultant, who spent approximately 2 minutes less per case when using remote microscopy; this was achieved at the expense of considerably greater effort on the part of the referring pathologist, who spent approximately 16 minutes per case selecting an average of 4.5 images for transmission to the consultant. The use of remote video microscopy for pathology consultation is associated with a complex series of tradeoffs involving cost, information loss, and This is a US timeliness of consultation. HUM PATHOL 24:909-911. government work. There are no restrictions on its use.

the consultant has control over the image information obtained. Robotic systems are relatively expensive and require high data transmission rates to achieve the required high-speed image transfer. In the second type of system, which we refer to as “selective video microscopy” (SVM), the fields to be examined are selected by the referring pathologist and transmitted at low speed via an ordinary telephone connection. If the consulting pathologist requires more images, these must be obtained by the referring pathologist and transmitted. The system is much less expensive than the robotic system, and does not require high-speed data transmission capability. This lower cost is achieved at the expense of reducing the information available to the consultant. The performance characteristics of RVM are fairly well understood because, with the exception of the image degradation associated with video image transmission, the process is virtually identical to that used in any other pathologic consultation based on the contributor’s original material. The performance characteristics of SVM are less well understood, however, both because the impact of field selection on consultant performance has not been documented and because the additional demands on both the referring pathologist’s time and that of the consultant have not been cletermined. To better define the diagnostic utility of consultation using SVM, we have used one such system (TelMed; Discovery Medical Systems, Overland Park, KS) to assist in the interpretation of frozen sections arising in the course of neurosurgery. The time required for a consultant to render a diagnosis by SVM was compared with that required to render a diagnosis under normal consultation conditions using an ordinary microscope. The accuracy of the diagnoses obtained using the two approaches also was evaluated, as was the effect of remote video microscopic consultation on the time of the referring pathologist.

The use of remote video microscopy (telepathology) to provide pathology consultation is attractive because it provides an opportunity for pathologists to obtain very timely consultation on difficult cases. As a result, the concept has attracted widespread attention,‘.” although only a few limited tests have been conducted.‘.‘,” Two competing technologies have emerged, however, with dramatically different price and performance characteristics. The first, which we refer to as “robotic video microscopy” (RVM), provides remote consultation via a robotic microscope that can be controlled by the conRobotic systems may use either sulting pathologist.‘,’ dynamic imaging technology, in which images are viewed in real time as the slide moves under control of the consulting pathologist, or static imaging technology, in which images are captured individually and transmitted as single files. However, in both types of robotic systems field selection is accomplished by the consultant. Thus,

From the Departments of Cellular Pathology, Neuropathology. and Ophthalmic Pathology, Armed Forces lnstitute of Pathology, Washington, DC; and the Department of Pathology, Walter Reed Arm) Medical Center, Washington, DC. Accepted for publication March 3,

MATERIALS

AND METHODS

Case Selection

1993. * Presentaddress:Department

in which intr.awere obtained from the tiles of the Walter- Reed Army Medical Center. The

Fifty-two consecutive operative frozen section

of Pathology, Shadyside Hospital, Pittsburgh, PA. The assertions and opinions contained herein are those of the authors and should not be construed as official or as representing the views of the Department of the Army, Department of the Air Force, or the Department of Defense. Address correspondence and reprint requests to T. J. O’Leary, MD, PhD, Department of Cellular Pathology, Armed Forces Institute of Pathology, Washington, DC 20306-600. This is a US government work. There are no restrictions on its

neurosurgical consultation

cases was used

lesions examined included a spectrum of’ ncurosurgical diseases, incluclilig nietastatic carcinoma, meningioma, medulIoblastoma. low-grade glioma, glioblastoma multiforme, vasof other cular malformation, epithelial cyst. and a variety diagnoses. The remote consuI1atton experiment was not actually carried out as a part of the intraoperative consultation, and the results of the consultation were not used in patient care. Five of tile fiftv-two cases were excluded from the analysis

USC.

0046-8 177,‘93/2408-0013$0.00/O

909

HUMAN PATHOLOGY TABLE 1.

Diagnoses Associated With Selective Video Microscopy, Frozen Section, and Permanent Section When a “Significant” Diagnostic Discrepancy Occurred

Case No.

FS Diagnosis

SVM Diagnosis Metastatic

2

carcinoma

21

Vascular

28 29

Favor infarct Spindle cell hypocellular tumor; defer to permanent sections Probable pituitary adenoma Consistent with glioma

38 49

malformation

Cystic mass, defer to permanent sections

52

Abbreviations:

Volume 24, No. 8 (August 1993)

SVM, selective

High-grade

glial tumor

Hemorrhage and necrosis; need more tissue Infarct TJnecrotic tumor v abscess Not tumor; consistent with but not diagnostic of vascular malformation Pituitary adenoma v meningioma; defer to permanent sections Anaplastic astrocytoma; request more tissue to rule out glioblastoma Cystic tumor; defer to permanent sections

Acquisition

High-grade glioma with gemistiocytic component Arteriovenous malformation Glioblastoma multiforme Fibrinous clot surrounded connective tissue Consistent Astrocytoma,

with pituitary grades

by fibrous

adenoma

3 and 4

Glial proliferation

video microscopy; FS, frozen section; PS, permanent section

of the time required to acquire, transmit, and interpret the images because incomplete data were obtained. However, these cases were included in the analysis of the diagnostic accuracy.

Image

PS Diagnosis

and Transmission

Image acquisition, transmission, and display were carried out using the TelMed system. Briefly, the system consisted of a color RGB camera (Hitachi VR-C360, Lindhurst, NJ) with a 2/3“ single layer MOS 760 (H) X 485 (V) pixel image sensor attached via C mount and camera tube to a Leitz microscope and interfaced with a TelMed image acquisition, storage, and display (1,024 X 856 pixel) unit. A similar Telmed computer and display unit was present at the receiving end. The color images were transmitted with or without overlays drawn to indicate areas of special interest. Overlays drawn during an interactive telephone session at either TelMed unit were transmitted for display over the same color image shown on the other TelMed unit. Image compression was carried out by the TelMed unit; transmission occurred at 14.4 kbyte/s unless telephone line noise necessitated a slower transmission speed. Image transmission equipment was located at the Walter Reed Army Hospital; receiving equipment was located at the nearby Armed Forces Institute of Pathology. Selection of fields to be transmitted was left to the discretion of the single referring pathologist (R.L.B.), who digitized and transmitted those images as a batch before beginning an interactive telephone session with the consulting pathologist (R.J. or C.S.S.). The consulting pathologist received patient history and intraoperative findings by voice while viewing the transmitted images. The referring pathologist was informed if the consultant believed additional fields to be required for diagnosis; the referring pathologist then carried out field selection according to the consulting pathologist’s directions. Immediately after rendering diagnoses by SVM on a batch of cases, the consulting pathologist viewed the frozen section slides by direct microscopy and rendered diagnoses for comparison with the SVM diagnoses. Data accumulated included the number of pictures transmitted in batch mode, the number of images transmitted singly, and the number of overlays transmitted in each mode. The time required for batch image capture and transmission was recorded, as was the time required for the interactive portion of the TelMed consultation and the time required for direct microscopic evaluation of the material available at frozen sec-

tion. The diagnosis obtained from the TelMed consultation was recorded, as was the frozen section diagnosis made by direct microscopy and the diagnosis reached after examination of all permanent sections. Analysis of data was carried out using the Epistat computer program (Tracy Gustavson, MD, Round Rock, TX).

RESULTS The time required for the referring pathologist to examine the specimen and select images for transmission ranged from 2 minutes to more than 1 hour, with an average of 16.5 minutes per case (standard deviation, 11.6 minutes). The number of images transmitted per case varied from 1 to 11, with an average of 4.5 (standard deviation, 2.1). In most cases only images used in a batch mode were used to achieve a diagnosis. In only four cases were additional images transmitted during interactive consultation; these cases were all obtained within the first third of the study. Image transmission time varied correspondingly from 1.9 to 20.9 minutes, with an average of 10.5 minutes. The time required for the consulting pathologist to examine the frozen sections varied from approximately 30 seconds to 47 minutes (mean, 9.30 minutes; standard deviation, 8.93 minutes). This exceeded somewhat the time required for the consulting pathologist to view and interpret the images transmitted across the SVM system (mean, 7.5 minutes; standard deviation, 5.1 minutes) by slightly less than 2 minutes per case. In 28 of the 52 (54%) cases in which diagnostic accuracy and precision could be analyzed there was absolute concordance between the SVM, frozen section, and permanent section diagnoses. In an additional 10 cases there was exact concordance with the frozen section diagnosis, but some discrepancy with the diagnosis rendered after examination of permanent sections. In five cases there were extremely minor discrepancies between the SVM and frozen section diagnoses. Finally, in seven cases (Table 1) there were discrepancies between RVM and frozen section diagnoses that might, by some measure, be considered major. In none of these 910

TELEPATHOLOG

cases, however, is it likely that the SVM diagnosis would have resulted in different treatment than the frozen section diagnosis. DISCUSSION The use of RVM as an aid to pathologic diagnosis is attractive because it provides rapid and relatively painless access to consultants. We have demonstrated that the still transmitted images selected by a referring pathologist are, in most cases, adequate to make a diagnosis that is similar to that which would be rendered by such a consultant based on neurosurgical frozen section material. From the standpoint of the consulting pathologist, the time required for such consultation is decreased, presumably because he or she does not have to view the entire specimen before rendering an opinion. The cost in time to the referring pathologist is substantial, however, since great care must be used in selecting images for transmission to assure that a meaningful consultation can be rendered. Consultation by video microscopy relies on the interpretation of less information than is usually available to a consultant. First, there may be loss of information associated with selective sampling. In addition, all currently available video microscopy systems transmit images that demonstrate considerable loss of detail when compared with images viewed directly in a light microscope or images that are captured on film. This loss of detail, which results from digitization and (for some systems) image compression, causes significant difficulty in interpretation of high-resolution cytologic and tissue features. Additional difficulty in image interpretation results from the necessity of relying on the single fixed focal plane of SVM images, which conveys less information than that obtained by direct microscopy, in which easy and continuous focus adjustment through a range of focal planes is routine. Optimal interpretation of chromatin features, refractility, and possible bacterial forms benefits from visual integration of information covering a neighborhood of focal planes. For brain tissue studied by SVM, distinction of neuropil from bland necrosis was often difficult. Since images representing only a subsampling of tissue could be transmitted, consultants were cautious in rendering diagnoses. In one case failure of the referring pathologist to include images from an area of relatively high cellularity held the consultant back from a diagnosis of astrocytoma (rather than the differential diagnosis of glioma u infectious process that was rendered by SVM). The diagnosis from permanent sections in this case was anaplastic astrocytoma. In another case the prominence of necrosis in a glioblastoma multiforme was not fully appreciated from the transmitted images, resulting in an SVM differential diagnosis of “high-grade glioma u metastatic tumor; defer to permanents.” In several other cases the consultant rendered less definite SVM diagnoses because of the inability to screen large areas of tissue for focal features, such as micronecrosis, pseudorosettes, or dysmorphic cells. Finally, the range of information that may be provided by immunohistochemical stains and molecular biologic assays ordinarily 911

(Becker et al)

available to a consultant cannot be used when evaluating a video image. The net result of all these limitations is that SVM diagnoses often include slightly wider differential diagnoses than those rendered after direct microscopy. Consultants seem to be more cautious when rendering an opinion based on a video itnage than they are when they have evaluated the tissue in their own laboratories. Given the lack of a “standard of care” for interpretation of transmitted video images, this caution may well be appropriate. The limitations associated with the use of SVM in consultation require that the consulting pathologist be well oriented to the video image appearance of pathologic changes. The referring pathologist must be adept at acquiring high-quality video images that are not simply “representative” of the lesion in question, but that include the specific architectural and cytologic features necessary for making a video diagnosis. Effective communication between the referring and consulting pathologists will be facilitated by a uniform format for this type of consultation and is absolutely essential for useful consultations to be rendered on diagnostically difficult cases. Our experience suggests that establishing effective communication requires a number of consultation sessions. The application of remote video microscopy to frozen section diagnosis, in which the patient is on the operating table, might be justified given the relatively small (approximately $25,000 per unit) cost for equipment purchase. Although our results do not suggest that the use of remote consultation would have improved patient care, it might have improved the confidence of a pathologist ill at ease with neurosurgical frozen section diagnosis. For other consultations, however, a l-day delay in diagnosis (caused by use of express courier services) may be justified by the additional information available to the consultant and by the fact that the cost of the remote video microscopy equipment alone is enough to ship approximately 1,000 specimens by express courier. Acknowledgment. The authors wish to thank Dr James Nelson for valuable comments on the clinical significance of several of the diagnostic discrepancies observed in this study, Dr Vernon Armbrustmacher for his critical reading of the

manuscript, and Discovery Medical TelMed units used in this study.

Systems

for loan of the

REFERENCES 1. Nordrum service:

I, Engum B, Rinde E, et al: Remote A telepathology project in northern Norway.

22514-518,

frozen

section

HIIM PATHOL

1991

2. Weinstein RS: Telepathology comes of ;~ge in Norway. HUM PA-rHOL2251 1-513, 1991 3. Eide TJ, Not-drum I, Engum B, et al: Use of telecommunications in pathology and anatomy services. Tidsskr Nor Laegeforen 111: 17-19. 1991 4. Weinstein RS, Bloom KJ, Rozek LS: Telepathology. Longdistance diagnosis. Am J Clin Pathol 9 1 :S39-S42, 1989 5. Weinstein RS: Prospects for telepathology. HUM PATHOL 17: 433-434, 1986 6. Weinstein RS, Bloom KJ, Rozek LS: Telepathology and the networking of pathology diagnostic services. Arch Pathol I.ab Med 111:646-652,1987