COVER FEATURE STATE OF TELEMEDICINE—PART I

The Evolution of
Real-Time Remote
Intraoperative
Neurophysiological
Monitoring
Jeffrey R. Balzer, University of Pittsburgh
Julia Caviness , Cadwell Industries
Don Krieger , University of Pittsburgh

Real-time consultation and monitoring of the patient’s

nervous system function is provided in more than 410,000
high-risk surgeries per year in the United States. More
than 80% of those cases are staffed by remote clinical
neurophysiologists using real-time telemedicine. The
evolution of the underlying technologies is described.

I
ntraoperative neurophysiological monitoring (IONM) occur. Most commonly these are obtained in the operat-
is an ancillary medical service used in a wide variety ing room (OR) after induction of anesthesia but before
of surgical procedures which pose risk to the patient’s positioning the patient for surgery. Recording continues
nervous system. Typically baseline IONM record- throughout the procedure and each subsequent record-
ings are obtained from the at-risk neural structure(s) ing is compared with both the preceding recordings and
either preoperatively or during surgery before the risk(s) with the baseline. When changes are recognized, their
likely cause is interpreted, documented, and reported to
Digital Object Identifier 10.1109/MC.2023.3283851
the surgeon when they represent potential compromise
Date of current version: 23 August 2023 or injury to a neural structure.

28 CO M PUTE R P U B LISHED BY THE IEEE COMP UTER SOCIE T Y 0018-9162/ 23©20231EEE

INTRODUCTION At the time, the OR suites were That early work served as proof of
The neurophysiological signals are most located in four hospitals adjacent to principle and as functional specifica-
commonly neuroelectric waveforms. each other with inside corridor access tion for both the use of telemedicine
Although automated algorithms may from one to the other. Nevertheless, in IONM and for the underlying tech-
be used to identify time variations in without remote access to the data, nology. Within a decade, commercial
these waveforms1,2,3 the responsibility IONM could be compromised when IONM devices with remote display
for those judgements and the assess- more than two cases required attention capabilities became available and were
ment of both the cause and the resultant simultaneously. In addition, typically in wide use.
imperative to inform the surgeon is car- 60+ hours/week effort was required In 2009, the American Clinical Neu-
ried by a highly trained and experienced from both clinical staff members. To rophysiology Society published rec-
oversight neurophysiologist. Qualified address these issues, ethernet local-area ommended standards for IONM which
individuals for this integral role in the
surgical team are and always have been
both essential and rare. The caseload
for many IONM services cannot be ade- THE CASELOAD FOR MANY IONM
quately managed using an in-person
SERVICES CANNOT BE ADEQUATELY
model, that is, one neurophysiologist
dedicated to one room at a time. The
MANAGED USING AN IN-PERSON MODEL,
advent and use of telemedicine has been THAT IS, ONE NEUROPHYSIOLOGIST
driven by the need to maximize the case DEDICATED TO ONE ROOM AT A TIME.
coverage provided by each such trained
and qualified individual.

Use of telemedicine in IONM network (LAN) connectivity was installed included the following statement: “… The
The development of telemedicine sup- in the ORs and IONM offices in all four monitoring physician … is responsible
port for IONM initially took place at the hospitals and a software package was for real- time interpretation of IONM
University of Pittsburgh Medical Center. developed and deployed which enabled data … should be present in the operat-
It was driven by a high and rapidly grow- real-time remote display of up to 16 neu- ing room or have access to IONM data
ing caseload. By 1987, the IONM service rophysiological modalities at a time on in real time from a remote location and
there was monitoring 1500 cases/year. any computer on the LAN.4,5 This tech- be in communication with the staff in
In addition to two clinical staff neuro- nical advance in telemedicine enabled the operating room.”8,9 Since then, the
physiologists, the service at that time a limited number of oversight neuro- demand for IONM in the United States
employed several trained and experi- physiologists to monitor multiple cases has grown to 410,000+ cases per year
enced technologists. The technolo- simultaneously, safely, and efficiently. with more than 80% handled using
gists operated IONM equipment in the Additional burdens on the limitations telemedicine. Based on utilization, lack
operating room, attached electrodes to of the technology came when surgeons of qualified providers, and efficacy
the patients, continuously surveilled who used IONM services in the univer- equivalent to the in-person model,
the IONM responses, and called for the sity hospitals requested IONM for their telemedicine arguably has become
oversight staff neurophysiologist as the procedures in nearby community hos- a standard of care for the provision of
critical parts of the case approached, pitals. In addition, several surgeons who professional IONM services by remote
communicated with the surgeon as the relied heavily on IONM took jobs else- neurophysiologists.
voice of the neurophysiologist when where. In response to these demands, a
needed, and documented the case. The wide-area-network (WAN) capable data METHODS
equipment had been assembled at the transport layer was added to the remote
university using commercially avail- display package.6,7 This expanded the Original LAN remote display
able components and integrated by an IONM telemedicine services beyond the As stated previously, the initial imple-
in-house software effort.4 University of Pittsburgh hospital campus. mentation of remote real-time display

SEPTEMBER 2023 29
STATE OF TELEMEDICINE—PART I

in support of IONM extended only over the signal-to-noise ratio progressively remote monitoring. Data access by
the LAN spanning four adjacent uni- improved with #stimuli and was con- remote display clients was handled sim-
versity hospitals.5 The data acquisition sistently interpretable within 30–60 s. ply by reading the data files for the cur-
computers used in the operating rooms At the end of an average, the resultant rently active cases. The routine could
were diskless Apollo workstations digitized waveform(s) were saved to an also display archived cases. These capa-
(Chelmsford, MA). These machines archive file along with a time stamp bilities were enabled by the LAN-wide
included a bit-mapped grayscale moni- and an annotation when appropriate. file system provided by the AEGIS OS.
tor suitable for waveform display and an A significant percentage of cases The programming demands to imple-
Industry Standard Architecture (ISA) also required neuromuscular moni- ment the remote display required only
bus. An ISA analog interface board was toring techniques, electromyography this file access and the 2D graphics10
used to digitize the patient’s neuro- (EMG). For these, recording electrodes needed to display the waveforms.
physiological signals and also provided were placed over muscle groups on the The first iteration of the remote dis-
triggers for physiological stimulators. face and/or over the eyes for intracra- play routine required a modest effort
The processor in the machine was the nial tumor resections and pediatric to write and deploy. Each modality was
Motorola 68020 with 68881 floating mastoidectomies, and on the arms and displayed in its own window. Each of
point unit capable of 70,000 FLOPS. legs for spinal fixation case. EMG was those windows was initially controlled
The operating system was Apollo’s monitored and recorded continuously. by a separate instance of the display
proprietary Unix-like AEGIS OS which The signals could be amplified and routine. That awkward and inefficient
included a fully capable 2D graphics played in real time over a speaker in approach was eliminated in a new
package.10 Diskless machines were the OR, and most surgeons preferred software version, rdraw16, which con-
used to eliminate the risk of disk fail- having this direct access to the moni- trolled up to 16 windows at a time.
ure since the machines were routinely tored muscle groups in which case they rdraw16 included controls for each
moved from room to room. interpreted the signals themselves. In window for the range of data epochs
Network connectivity was provided those cases, the signals were still digi- to display, the gain to be applied to
by a 12-Mbit/s token ring interface tized, archived, and interpreted by the each channel, and on/off and param-
which in practice was more than dou- staff neurophysiologist as a backup to eter controls for a variety of signal
ble the speed of 10-Mbit/s ethernet. the surgeon, but interactions medi- enhancement algorithms, many of
Server machines were positioned stra- ated by remote display were typically which were developed in-house.7 In its
tegically in the OR suites and IONM not helpful. default mode, rdraw16 continuously
office spaces to support the diskless For the Apollo system, each set of polled for new epochs, displayed them
machines in the ORs and on the over- waveforms was saved in a 9 x 512-byte when they appeared, scrolled the dis-
sight neurophysiologists’ desks. Each record, that is, 4608 bytes. The first play to show the most recent 10 epochs
server was configured as a router with block was used to store stimulus rate, in a waterfall display along with two
both a local token ring interface and amplifier settings, time stamp, and so baseline epochs, and showed time
an ethernet interface for longer-range on. The waveforms were stored in the stamps and annotations (Figure 1).
communication with the other servers. remaining eight blocks as 4-byte float- For each evoked potential modality,
The most commonly used neuro- ing point numbers. This compact data a second window could be opened in
physiological modalities were aver- structure was a legacy of the original which the ongoing partially averaged
aged neuroelectric responses evoked IONM data acquisition system built data was displayed with updates every
by sensory stimuli. Repeated momen- around Digital Equipment Corporation’s 2–3 s. This useful capability was not
tary stimuli were presented to the PDP-11 computer (Maynard, MA). These enabled in the WAN version of the
ear (click), eye (flash), or to a periph- machines were limited to a 64-Kbyte software discussed next.
eral nerve in the arm or leg (electrical program space, were configured with a A command-line text interface
stimulus). With each stimulus, a time- 20-Mbyte disk, and were rack mounted controlled the application and, by
locked data segment was digitized in closets in the OR suites. default, was minimized to an icon as
and added to the running average The use of 4608-byte records mark- seen at the lower right corner of the fig-
of the preceding responses. Ideally, edly limited the network demands for ure. rdraw16 was slaved to a file polling

30 COMPUTER  W W W.CO M P U T E R .O R G /CO M P U T E R

script, rdraw.csh, which intermittently this system maintained a real-time dis- Remote display capability was also
searched for new active files and play of all active cases automatically enabled over standard telephone line
instructed rdraw16 to spawn a new win- and could be left running indefinitely. 19,200-baud modems. Using a Tektr-
dow with default display when one was rdraw.csh controlled rdraw16 through a onix-capable graphics monitor,11 an
found. If no new epoch had been saved scratch text file into which it wrote com- oversight neurophysiologist could
to a file for 60 min, rdraw.csh instructed mand line instructions which rdraw16 dial in to the hospital network, run a
rdraw16 to kill that window. Hence, polled, read, and deleted. rdraw16 instance from the command

FIGURE 1. A typical remote monitoring display. This figure shows a display obtainable from anywhere on the LAN. The icon at the lower
right corner contains a process that is automatically and continuously searching for active operating room or diagnostic studies. The four
small windows containing waterfall displays were spawned by this process and display one ongoing diagnostic study (upper right corner)
and three OR cases. The two text windows at the top show a manuscript in preparation. The figure in the center was produced using the
process shown in the lower left window. (Source: Krieger et al.5)

SEPTEMBER 2023 31
STATE OF TELEMEDICINE—PART I

line and see the display for one modal- is commonly used today in IONM, even naming services. The ISD was devel-
ity at a time using the Tektronix 4010 though it does not enable control of the oped in-house. It polled the local
pen plotting graphics layer built into displayed waveforms by the remote machine for active data files, pro-
the software. This capability was rou- neurophysiologist. vided notifications of their presence
tinely used from home to provide For rdraw16, a WAN-capable data across the network, and ser viced
remote oversight for a case in the hos- transmission layer was added to han- data requests from rdraw16 instances
pital and occasionally from home or dle the movement of digitized wave- anywhere on the WAN for data from
office to follow cases at nonuniversity forms across the Internet. Parallel those files.
hospitals which were staffed by uni- virtual machine (PVM)12 and message By default, PVM-enabled daemons
versity surgeons. passing interface (MPI)13 were consid- used TCP sockets to pass messages one
ered. Both packages had been devel- to the other and each process-to-process
Original WAN remote display oped to support master–slave cluster connection required an open file
At the time the original WAN remote and supercomputing applications descriptor at each end. This did not
display capability (Figure 2) was added,6,7 which require more processors than scale sufficiently to handle our typical
the underlying equipment had changed can be placed on a single shared-mem- connectivity requirements between
and improved again. The machines ory bus. MPI has since become the rdraw16 instances, ISD’s, and PVM
used in the ORs and offices were Hew- standard for high-performance com- daemons. PVM did, however, include
lett Packard workstations (Palo Alto, puting. We selected PVM over MPI User Datagram Protocol (UDP) capa-
CA) with Motorola 68040 processors because of its open architecture and bility which required only a single file
(3 MFLOPS), 1-GByte disks, 100 Mbit/s because Hewlett-Packard and others descriptor for each process. We opted
ethernet interfaces, and a fully capa- were providing manufacturer-specific to use this and modified our PVM
ble Unix operating system (HPUX) versions of MPI which raised concerns installation to use nonstandard ports
including X-Windows. The restric- that interoperability might become a for security purposes.
tions on data length had been relaxed problem among machines from differ-
in the acquisition software package ent manufacturers. Commercial remote
and the caseload had grown to 2500+ When a machine was booted in the display capabilities
cases/year. However, the substantive operating room, its presence on the net- As stated previously, the original efforts
increases in hardware capability and work was detected by a single polling at the University of Pittsburgh served
network speed more than made up for process running inside the University as a functional specification for the
the increased demands on the system. of Pittsburgh network. Each machine remote display capabilities which fol-
X-windows includes both screen was provided with a fixed IP address lowed, most notably for commercially
and window capture functionality and automatically ran a sshd daemon available machines. In 1994, Cadwell
which were tried for remote display. on a nonstandard TCP port. The sshd on Industries (Kennewick, WA) intro-
This approach proved to be unsatisfac- that port of each IP address was polled duced remote display capabilities in
tory for several reasons. It was awkward at random intervals every few min- its Excel product line. Display capture
to enable and was both slow and nega- utes. Internet Control Message Proto- and transmit technology was used
tively impacted the performance of the col (ICMP) packets14 were initially used every 15 s under Microsoft Windows.
machine in the OR. It did not enable the but when Unix utility nmap became Up to nine remote displays could be
neurophysiologist to control his/her available,15 synchronize (SYN) packets viewed from monitored cases on
waveform display, and it necessitated were used to minimize the chance that the same LAN, four simultaneously
a substantive security risk to the OR our legitimate port scan would trigger (see Figure 3).
machine since it required opening the an alarm. In 2002, Cadwell introduced “Cas-
standard X11 TCP port in the hospital The PVM daemon and an infor- cade Classic” software. Their “Remote
firewall. In recent decades, the perfor- mation services daemon (ISD) were Reader” enabled simultaneous view-
mance and security problems with this spawned on the newly connected ing of up to three waveform sets and
conceptual approach have been solved, machine using ssh. The PVM dae- included two-way chat, live video,
and screen capture for remote display mon mediated message passing and and live EMG. The remote viewer

32 COMPUTER  W W W.CO M P U T E R .O R G /CO M P U T E R

FIGURE 2. A typical screen image is shown from a workstation with an rdraw16 remote monitoring display. Responses are currently
shown from seven cases running simultaneously in two hospitals, one of which is in Pittsburgh: “puh” and one of which is in Wash-
ington, D.C.: “gwh.” Each time a recording is completed, the trace appears in the remote display as well as in the operating room. Each
waterfall display window includes: 1) a window name, for example, 24; 2) a name, for example, hp71550m:*puh_b16204.msp,
indicating the name of the computer acquiring the data “hp71550m,” the name of the hospital “puh”: Presbyterian University Hospital
of Pittsburgh, the name of the case “b16204,” and the type of responses displayed “bap”: brainstem auditory evoked potentials; 3)
the names of the recording channels: “International 10/20 System placements P4/Fz, P3/Fz, Cervical C7/Fz, Erb’s point”; 4) a series
of waveforms with annotations entered by the technician. The window at the upper left labeled **rdraw-CMD** provides control over
the waterfall display windows using the mouse and keyboard. This control window was developed using the Tk Toolkit; see Apollo
Computer, Inc.10 The iconified window labeled rdraw is a text-only control window. 1) Windows 44, 34, 33, and 23 show brainstem
auditory EPs from four different cases in which a craniectomy and decompression of the trigeminal nerve are being performed. 2) Win-
dows 24 and 14 show bilateral median and peroneal nerve SEPs, respectively, from a preoperative diagnostic study in anticipation of
spine surgery. 3) Window 43 shows four channel EEG monitoring from a patient with a head injury in an intensive care unit. Windows
13, 42, and 32 show bilateral median nerve, bilateral peroneal nerve, and EEG, respectively, from a craniotomy and resection of a
tumor. (Source: Krieger et al.7)

SEPTEMBER 2023 33
STATE OF TELEMEDICINE—PART I

could control her/his data display display. The Natus Xltek Protektor32 display capability by the oversight
including channel scaling and which system (Oakville, Ontario, Canada) neurophysiologist.
epochs to display, and could view included remote viewer control of
previously recorded cases, all with- their data display as of 2020, but their Security
out affecting the data acquisition IONM equipment is now being phased The Healthcare Insurance Portability
machine’s functions. These full-fea- out and is no longer supported. Sev- and Accountability Act (HIPAA) became
tured capabilities were usable over eral other manufacturers including U.S. federal law in August 1996,16 but
t he Internet. In 2010, Medtron ic Nihon Kohden (Irvine, CA) and Cad- the code sets17 and privacy rule18 cre-
Xomed’s NetOp for Windows (Mem- well Industries (Kennewick, WA) ated by the U.S. Department of Health
phis, TN) provided remote monitor- continue to supply equipment in the and Human Services under the law
ing via LAN or Internet using screen United States and software capable did not go into effect until April 2003.
capture. By 2015, they had included of multiple-case remote display with In that interim without benefit of pub-
remote viewer control of their data full-featured data exploration and lished rules, hospital IT managers

FIGURE 3. The Cadwell Cascade remote display—1994 brochure. (Source: figure reproduced courtesy of Cadwell Industries, Kennewick, WA, USA.)

34 COMPUTER  W W W.CO M P U T E R .O R G /CO M P U T E R

were understandably overcautious in AES 256 encrypted using a Microsoft and 2) real-time data exploration and
handling requests for movement of Enhanced Cryptographic provider display capabilities on those office
patient health information (PHI) out of that is FIPS 140-2 compliant. Data is machines. Secondary emphasis was
their networks. encrypted prior to network transmis- placed on both security and documen-
With negotiation and with growing sion over SChannel-based TCP using tation capabilities and on intercommu-
experience and understanding of the the highest-level available security nications between technologist and
protections and limitations of newly of TLS 1.2. A private certificate is pro- staff neurophysiologist. For security,
installed firewalled routers, it became vided that can be evaluated by both heuristics were applied which fit the
possible and then routine for Univer- sides of client/server communication. demands of each hospital’s data secu-
sity of Pittsburgh oversight neurophys- HL7 transmission is supported for rity group. Documentation was limited
iologists to provide IONM services to electronic medical record interoper- to the notes entered into the remotely
remote hospitals. Identifying PHI was ability, and wireless, wired, and dedi- viewable data record acquired in the
excluded from all transmissions across cated cable connections are supported operating room by the technologist in
the network by the data transport for client-server network communica- attendance with a hand-written note
layer. IT managers at remote hospitals tion. When the client is connected to entered in the patient’s chart at the end
placed exceptions in their firewalls for a central CadLink server, the patient of each case. Intercommunication was
a nonstandard transmission control data and reports are streamed to the handled using a generic chat program or
protocol (TCP) port for ssh to enable CadLink server’s storage setup. When in some cases, by land-line telephone.
spawning pvmd3 on a newly connected directly connected to another client It was the efforts of commercial
OR node, and both a primary and two system for remote monitoring, the equipment manufacturers and the
alternate nonstandard UDP ports for patient data are streamed from one development of industry wide med-
pvmd3 message transfers. system to the other system. Only regis- ical and legal standards which drove
Today, at least one manufacturer tered client machines are able to send advances in both security and docu-
has introduced a central server element patient data to the CadLink Server or mentation. The Health Level Seven
into their data transport layer. This is other registered client machines via (HL7) standard for transfer of clinical
an incremental but significant innova- an encrypted connection. Audit logs and administrative data by health-care
tion to the crossbar data flow architec- are encrypted, can be monitored by the providers was initially published in
ture used in the past. It enables better owner/operator, and show the history 1989.19 As mentioned previously, the
scaling of data transport from many of actions performed with the CadLink U.S. Congress’ HIPAA regulations came
operating rooms to many observers. It system on all clients connected to the into force in 2003. The Distributed Com-
simplifies and enhances the security server. Server databases can be backed ponent Object Model (DCOM) for secure
of the system. It simplifies the man- up and restored.” communication between software com-
agement of the data archive. The engi- ponents on networked computers was
neering group at Cadwell Industries, RESULTS AND DISCUSSION publicly launched for Windows 95 in
Kennewick, WA, provided the follow- At every stage of its evolution, IONM 1996 with full documentation availabil-
ing statement on the functionality of has been driven by the demands of ity beginning in 2006.20
their approach. The central server in surgeons performing high-risk pro- Today, IONM documentation includes
their system is called “Cadlink.” cedures and by the responsiveness of contemporaneous annotations which
CadLink serves as the interface for hospital administrators. In its infancy, ultimately construct the narrative of the
database storage and remote moni- IONM was limited by the financial case. This is written and time-stamped
toring support. Access to the system resources and technical capabilities electronically by the technologist in the
requires a username and password resident at a few academic medical operative room. The time-stamped nar-
that must meet minimum require- centers. The primary technical efforts rative can then be viewed at any time
ments, and role-based access controls were aimed 1) at enabling efficient during the procedure by the oversight
can be assigned to tailor end-user data transfer from the computers in neurophysiologist so that consideration
access to system settings. Both local the ORs to the computers in the offices of physiologic variables, anesthetic
and server managed databases are of the oversight neurophysiologists variables, and the flow of the surgical

SEPTEMBER 2023 35
STATE OF TELEMEDICINE—PART I

procedure can be used as part of the inter- insurers have recognized the safety, display control is a choice driven by
pretation and differential diagnosis. legitimacy, and reliability of vir- many factors. There are several advan-
When the case is complete, all saved data tual attendance via telemedicine and tages to fully remote-viewer controllable
should be moved from the local machine routinely reimburse the work. Since data display. 1) All data are transmit-
to a central server for storage. This allows 2013, telemedicine oversight of IONM ted to the remote viewer, so the raw
for the data to be saved for appropriate has been defined and reimbursed in dataset exists on the remote viewer’s
periods of time and reviewed in total for 15-min increments under CPT code computer or on a central server. 2) The
quality assurance and control purposes GO435 by Medicare21,22 and in 60-min remote viewer is able to manipulate
or should a question arise concerning increments under CPT code 95941 by the data transparent to the user in the
changes which occurred during the case. private party payers.22 The number of operating room including sizing of
More sophisticated intercommuni- IONM cases in the United States with individual windows, manipulation of
cation between technologist and clin- the neurophysiologist in attendance history/waterfalls, individual manip-
ical neurophysiologist via chat, voice, virtually via telemedicine has risen to ulation of data amplitude and time
and video were added by commercial more than 330,000 per year as of 2022. bases as well as minimizing data win-
equipment manufacturers as network- For many cases, large commercial dow(s). 3) In screen capture cases, the
ing, and software infrastructure was IONM service providers rely on screen technologist is responsible for making
added by the computing and network- capture technology for remote access sure that data can be viewed remotely
ing industries. Universal accessibility and display rather than on full featured by having the data and event logs vis-
of cell phones has provided a backup real-time data exploration and dis- ible on their desktop as well as being
intercommunication capability which play capabilities. For EMG monitoring solely responsible for changing gains,
is occasionally useful. For most cases, of many otolaryngological and spinal time-bases, and so on, should the over-
two-way chat is the fastest and most instrumentation cases, this limitation sight neurophysiologist request them.
efficient way for the technologist and is not problematic since the data qual- This can present a screen “real-estate”
oversight neurophysiologist to com- ity in many cases is quite good and most issue for both the technologist in the
municate. Moreover, the technologist surgeons can effectively interpret the OR and for the oversight neurophysi-
acts as the interpretive mouthpiece for signals themselves. For some IONM ser- ologist. 4) If personal health informa-
the oversight neurophysiologist, thus vices, data quality and interpretability tion is displayed on the IONM machine
allowing for communication between of evoked potential monitoring depend in the OR, screen capture and trans-
the oversight neurophysiologist and more heavily on the remote observer, mission of the window containing it
surgeon (via the technologist) in real and the observer requires more capable is a potential HIPAA violation. 5) Data
time. Two-way audio or multiway con- data exploration and display capabili- review is often required during or
ferencing are typically not needed as ties, for example, open or endovascular after the procedure to facilitate quality
the technologist is able to deliver good approaches to intracranial aneurysms, assurance or to resolve questions con-
or bad news concerning the IONM data. open or endoscopic approaches to intra- cerning sensitivity of the IONM.
That being said, should specific ques- cranial mass lesions, fixation of spinal The display of current data along
tions arise that can only be handled by fractures and spinal deformities, trau- with baseline waveforms from hours
direct consultation with the oversight matic hip replacements, carotid endar- or days before is most often viewed in a
neurophysiologist, higher bandwidth terectomies, aortic coarctation repairs, “stacked” or “waterfall” format which
communication may be a necessary surgeries for spinal cord tumors and typically requires full featured control
alternative to relayed communication spinal dysraphism, and open surgery to optimize the interpretability of the
via the chat. for cranial nerve neuropathy. This display. The choice to deploy an IONM
To enable meeting the steadily is also dependent on the complexity system without these capabilities
increasing surgical demand for IONM, of the monitoring approach since for is effectively made by the oversight
neurophysiologist oversight via tele- many of these cases, multiple modali- neurophysiologist in consideration of
medicine has grown to account for ties are required. 1) the case mix for which the system
more than 80% of all IONM cases The use of screen capture display will be used, 2) the work style she/he
in the United States. Health-care versus full-featured data access and uses for oversight, and 3) the training

36 COMPUTER  W W W.CO M P U T E R .O R G /CO M P U T E R

and experience of the technologist(s) was personnel. In response to a lack of medical environment,” Computer, vol.
who will be in the OR and 4) the train- personnel to provide in-person, one on 24, no. 3, pp. 45–55, Mar. 1991, doi:
ing and experience of the over- one oversight of surgical procedures, 10.1109/2.73512.
sight neurophysiologist viewing the new technologies were developed and 6. D. Krieger, B. Simon, T. Chay, and R.
case(s) remotely. matured. Since its inception, numerous Sclabassi, “A WAN surgical monitor-
advances have occurred, but the tech- ing application built on PVM,” in Proc.

T
nology remains partially at the mercy of PVM Users’ Group Meeting, Pittsburgh,
he work described previously regulatory (HIPAA) and security (fire- PA, USA, May 1995.
from the 1990s established the wall) rules. That being said, IONM deliv- 7. D. Krieger, S. Onodipe, P. J. Charles,
technological bases for data ery via telemedicine and the continued and R. J. Sclabassi, “Real time signal
transmission and communication on advancement of the technology that processing in the clinical setting,”
which IONM via telemedicine depend. supports it has become a standard in the Ann. Biomed. Eng., vol. 26, no. 3, pp.
Those foundational platform technolo- field and, as long as caseload vastly out- 462–472, May 1998, doi: 10.1114/1.95.
gies have matured since then but have numbers providers, will be relied upon 8. M. R. Nuwer, “Overview and history,”
not substantively changed. Since those for the foreseeable future as the safest in Intraoperative Monitoring of Neural
early papers, advances in data trans- and most efficient method for remote Function. Amsterdam, The Nether-
mission and distribution capabilities oversight of IONM. lands: Elsevier, 2008, pp. 2–6.
are only found in the published litera- 9. Guideline 11A: Recommended Stan-
ture of IONM equipment providers. REFERENCES dards for Neurophysiologic Intraopera-
The signal processing algorithms 1. D. Krieger and R. J. S. Sclabassi, “Time- tive Monitoring – Principles, American
used to enhance the interpretability varying evoked potentials,” J. Med. Eng. Clinical Neurophysiology Society,
of the physiological signals were and Technol., vol. 18, no. 3, pp. 96–100, 1994, Bloomfield, CT, USA, 2009.
continue to be active areas of research. doi: 10.3109/03091909409030238. 10. “Domain graphics primitive resource
Automated alarms which embody 2. D. Park and I. King, “Application call reference,” Apollo Computer,
computational algorithms for identi- of machine learning in the field of Chelmsford, MA, USA, Order
fying “significant” changes in the phys- intraoperative neurophysiological No. 007194, Rev 02, 1987.
iological signals are a growing research monitoring: A narrative review,” Appl. 11. VT240 Series Owner’s Manual, Digital
area both in the academic literature and Sci., vol. 12, no. 15, Aug. 2022, Art. Equipment Corp., Maynard, MA, USA,
inside the closed doors of commer- no. 7943, doi: 10.3390/app12157943. 1984, pp. 86–88.
cial equipment providers. 23 These 3. M. Kin, S. Park, Y. Kubota, S. Lee, and 12. A. Geist, A. Beguelin, J. Dongarra,
methods include neural network 3 D. S. Kong, “Applying a deep convo- J. Weicheng, R. Manchek, and V. S.
and machine learning2 classifiers. By lutional neural network to monitor Sunderam, PVM: A Users’ Guide and
far, the largest source of research find- the lateral spread response during Tutorial for Network Parallel Computing.
ings on IONM are not technological microvascular surgery for hemifacial Cambridge, MA, USA: MIT Press, 1994.
but rather are reports of clinical appli- spasm,” PLoS One, vol. 17, no. 11, Nov. 13. L. Clarke, I. Gendinning, and R.
cations. 24,25 This latter reflects the 2022, Art. no. e0276378, doi: 10.1371/ Hempel, “The MPI message passing
facts that 1) IONM is of significant and journal.pone.0276378. interface standard,” in Programming
growing importance in minimizing iat- 4. D. N. Krieger, R. M. Lofink, E. L. Environments for Massively Parallel Dis-
rogenic injury during high-risk surgery Doyle, G. Burk, and R. J. Sclabassi, tributed Systems, K. M. Decker and R.
and 2) the technologies on which it is ­“NEURONET: Implementation of an M. Rehmann, Eds. Basel, Switzerland:
based are reliable and stable. integrated clinical neurophysiology Springer-Verlag, 1994, pp. 213–218.
In conclusion, real-time remote IONM system,” Med. Instrum., vol. 21, no. 6, 14. E. Conde, “The history of ping – By
capability was born and has grown pp. 296–303, Dec. 1987. Terry Slattery,” Computerworld, Apr.
from a perceived need and was based on 5. D. Krieger, G. Burk, and R. J. Scla- 2009. [Online]. Available: https://
creating efficient and safe delivery of bassi, “NEURONET: A distributed www.computerworld.com/article/
services. In this case, the service was a real-time system for monitoring 2771333/the-history-of-ping---by
clinical one, IONM, and the unmet need neurophysiologic function in the -terry-slattery.html

SEPTEMBER 2023 37
STATE OF TELEMEDICINE—PART I

ABOUT THE AUTHORS

JEFFREY R. BALZER is an associate professor in the De­­ providers and strategies that help deliver patient care across
partment of Neurological Surgery, University of Pittsburgh, multiple IONM provision models. Caviness received an MBA
Pittsburgh, PA 15261 USA. His research interests include the in healthcare administration and analytics from Colorado
utilization of genetic biomarkers for the prediction of delayed State University. Contact her at juliac@cadwell.com.
cerebral ischemia in subarachnoid hemorrhage and defin­
ing the utility of intraoperative neuromonitoring in a variety DON KRIEGER is a research scientist in the Department of
of surgical procedures that place the central nervous system Neurological Surgery, University of Pittsburgh, Pittsburgh,
at risk. Balzer received a Ph.D. in behavioral neuroscience PA 15261 USA. His research interests include advancing
from the University of Pittsburgh. Contact him at balzjr@ the understanding, diagnosis, and treatment of traumatic
upmc.edu. brain injury and includes extensive work in high-throughput
supercomputing. Krieger received a Ph.D. in neuro ­
JULIA CAVINESS is the intraoperative neurophysiological science from Maharishi International University with
monitoring product manager for Cadwell Industries, Ken­ post-doctoral fellowship training in clinical neurophysiol­
newick, WA, 99336 USA. Her research interests include ogy at Children's Hospital of Pittsburgh. Contact him at
the technologies and clinical solutions that support IONM kriegerd@upmc.edu.

15. G. Lyon. Nmap Network Scanning. Ann Arbor, MI, USA, 2023. [Online]. reimbursement policy CPC032,” Blue
(1997). Nmap. [Online]. Available: Available: https://www.hl7.org/ Cross and Blue Shield of Illinois,
https://nmap.org/book/toc.html implement/standards/product Chicago, IL, USA, Apr. 2022. [Online].
16. U.S. House. 104th Congress. (1996). _brief.cfm?product_id=185 Available: https://www.bcbsil.com/
H.R.3103 – Health Insurance Porta- 20. “[MS-COM]: Distributed com- docs/provider/il/standards/cpcp/
bility and Accountability Act of 1996. ponent object model (DCOM) cpcp032-04222022.pdf
[Online]. Available: https://www. remote protocol.” Microsoft, Jun. 23. “Autonomous neuromonitoring AI
congress.gov/bill/104th-congress/ 2021. [Online]. Available: https:// to prevent paralysis,” Nervio, Misgav,
house-bill/3103/text learn.microsoft.com/en-us/ Israel. Accessed: Mar. 2023. [Online].
17. “HIPPA code sets overview,” openspecs/windows_protocols/ Available: https://www.nervio.ai/
U.S. Centers for Medicare and ms-dcom/4a893f3d-bd29-48cd-9f43 24. “NLM Pubmed search on ‘intraopera-
Medicaid Services, Baltimore, -d9777a4415b0 tive neurophysiological monitoring.’”
MD, USA , Apr. 2022. [Online]. 21. “Billing medicare for remote intra- PubMed. Accessed: Mar. 2023. [Online].
Available: https://www.cms. operative neurophysiology moni- Available: https://pubmed.ncbi.nlm.
gov/Regulations-and-Guidance/ toring (HCPCS Code GO453),” U.S. nih.gov/?term=intraoperative+
Administrative-Simplification/Code-Sets Centers for Medicare and Medic- neurophysiological+monitoring
18. “Health information privacy,” U.S. aid Services, Baltimore, MD, USA, 25. “Google Scholar search on ‘intraopera-
Department of Health and Human Sep. 2020. [Online]. Available: tive neurophysiological monitoring.’”
Services, Washington, DC, USA, Mar. https://www.cms.gov/medicare/ Google Scholar. Accessed: Mar. 2023.
2022. [Online]. Available: https:// medicare-fee-for-service-payment/ [Online]. Available: https://scholar.
www.hhs.gov/hipaa/for-professionals/ physicianfeesched/downloads/ google.com/scholar?hl=en&as
privacy/index.html faq-remote-ionm.pdf _sdt=0%2C39&q=intraoperative
19. “HL7 international version 2 22. “Intraoperative neurophysiology +neurophysiological
product suite,” HL7 International, monitoring (IONM) coding and +monitoring&btnG=

38 COMPUTER  W W W.CO M P U T E R .O R G /CO M P U T E R