Conference PaperPDF Available

Fiducial Marker Based Probe Depth Estimation Method for Efficient Transesophageal Echocardiography Examination

Authors:

Abstract and Figures

The invent of transesophageal echocardiography(TEE) has brought significant value addition to heart diagnosis and related surgical procedures. Ultrasound probes used in TEE are primarily designed for monitoring human heart during a surgical procedure or visualizing it for diagnosis purposes. Unlike conventional ultrasound probes, here the echo source is embedded inside the distal tip of a long shaft, which is designed to pass through a patient’s mouth. During a TEE procedure, the relative position of the distal tip within the human body from the patient’s mouth is visually interpreted from the numeric markings printed on the shaft surface. Lack of focus and slow procedure are inherent to this type of examination because of the need of a constant visual interpretation from the probe. This work propose to use a fiducial marker based depth estimation technique using computer vision. Continuously interpreted depth values enables automatic workflow triggers and measurement selection based on the anatomical view. A software based simulation is done as a proof of concept and the results were observed positive for random probe movements which includes both advancing and withdrawal.
Content may be subject to copyright.
Fiducial Marker Based Probe Depth Estimation
Method for Efficient Transesophageal
Echocardiography Examination
Sherin Sugathan
Siemens Healthcare Pvt. Ltd.
Bengaluru, India.
Email: sherin.sugathan@siemens.com
Lavanya Kumari Munsiff
Siemens Healthcare Pvt. Ltd.
Bengaluru, India.
Email: lavanya.munsiff@siemens.com
Abhai Easwar Nag
Siemens Healthcare Pvt. Ltd.
Bengaluru, India.
Email: abhai.nag@siemens.com
Abstract—The invent of transesophageal echocardiography
(TEE) has brought significant value addition to heart diagnosis
and related surgical procedures. Ultrasound probes used in TEE
are primarily designed for monitoring human heart during a
surgical procedure or visualizing it for diagnosis purposes. Unlike
conventional ultrasound probes, here the echo source is embedded
inside the distal tip of a long shaft, which is designed to pass
through a patient’s mouth. During a TEE procedure, the relative
position of the distal tip within the human body from the patient’s
mouth is visually interpreted from the numeric markings printed
on the shaft surface. Lack of focus and slow procedure are
inherent to this type of examination because of the need of a
constant visual interpretation from the probe. This work propose
to use a fiducial marker based depth estimation technique using
computer vision. Continuously interpreted depth values enables
automatic workflow triggers and measurement selection based
on the anatomical view. A software based simulation is done as
a proof of concept and the results were observed positive for
random probe movements which includes both advancing and
withdrawal.
KeywordsTransesophageal echocardiography, fiducial mark-
ers, real-time processing, 2D pattern recognition.
I. INTRODUCTION
Transesophageal echocardiography (TEE) [1], [2] is an
ultrasound based scanning method that generates a 2D/3D vi-
sualization of a human heart. On the other hand, Transthoracic
echocardiogram (TTE) [3] is a conventional echocardiography
method which is commonly used. The difference between TEE
and TTE lies with the way the heart is imaged. TEE tries to
get the heart view from inside the body by inserting a tube
like shaft through easophagus whereas TTE always scans the
heart from outside by placing a conventional probe on patient’s
chest. TEE is particularly useful when the patient is having
properies like a thick chest wall, obesity, bandages on chest
etc. An example illustration of TEE based procedure is shown
in Figure. 1.
It is also important to note the intraoperative uses of TEE in
monitoring the heart during procedures like valve replacement
[4], [5].
A. Generic Scanning Procedure
The scanning procedure generally takes 30 to 60 minutes
and is done by specialized doctors. Since the procedure in-
volves inserting the probe into the esophagus as shown in
Fig. 1. An example illustration of a TEE procedure.
Figure 1, an operator sprays a medicine into the patient’s throat
so as to make it numb. In addition, to help the patient remain
calm, a mild sedative is also injected into the body through the
intravenous line obtained from the patient’s arm. For recording
the electrocardiogram (ECG), small metal electrodes will be
placed on the patient’s chest. A bite guard will be placed into
the patient’s mouth to prevent the patient accidentally biting
and damaging the probe. The probe will be inserted through
the mouth guard and will go down the throat as shown in
Figure. 1. As the clinician gently advances the probe through
esophagus, the patient will be asked to swallow the probe to
enable an easy advancing of the probe. The probe advancement
will be done to a desired depth required by the clinician based
on the type of procedure. The clinician needs to constantly
monitor the probe for achieving the desired depth value. The
echo signals for imaging the heart is generated from the distal
tip of the probe and the clinician will get a real-time B-Mode
imaging on the computer screen.
B. Numeric Depth Markers
To help the clinician or the operator who is advancing
the probe through esophagus, numeric depth markings will be
printed over the probe shaft surface. The numeric markings on
a Siemens V5Ms TEE probe in shown in Figure. 2.
Fig. 2. An example TEE probe shaft showing the depth markings.
This numbers on the probe will help understand the cur-
rent depth of the distal tip with reference to the mouth of
the patient. As shown in Figure 2, major depth values are
represented using numbers (representing 20cm and 30cm in
Figure 2) and the minor markers (representing 25cm in Figure
2) are represented by a line marking. For cardiac imaging using
a TEE probe, there are around 20 standard views. The desired
depth value depends on the desired view, which is chosen
based on the pathology under study or the procedure to be
done. The 20 standard views are further classified into five
broader view names. Table I shows the broad classification on
the views based on the depth range in which those views are
expected.
TABLE I. CLINICALLY RELEVENT STANDARD TEE VI EW S
TEE Depth Range (cms) Standard TEE View
25-28 Upper or high esophageal
29-33 Mid-esophageal
34-37 Gastro-esophageal
38-42 Trans Gastric
>42 Deep Trans Gastric
As mentioned earlier, the operator or the clinician will
have to keep an eye on the probe to track the current depth
value. There are two main drawbacks associated with this
kind of manual numeric inpterpretation of the probe markings.
Firstly, while trying to obtain a window, the operator needs
to constantly look back to the TEE probe for reading the
current depth. In fact, this becomes a distraction and adds
to the difficulty of quickly acquiring a desired scan window.
Secondly, the operator needs to manually activate measurement
options in the software based on different views. Here, the
software system in the ultrasound machine is unable to bring
an auto-measurement option mainly because it does not have
any information on the probe’s current depth or the current
view. The clinician or the operator has to manually verify
the current depth and needs to decide upon the appropriate
measurement workflow.
C. Automatic Depth Estimation
One of the main advantages of having an automatic depth
estimation is that the focus of the operator can be maintained to
the computer screen during the procedure and have the mea-
surements auto-populated or workflows auto-trigerred based
on the current view which is inferred from the current depth
value. A mechanical solution in proposed in [6], where a roller
based solution is used to detect the current depth of the TEE
probe. The solution in [6] carries a major drawback as it
introduces frictional force on the probe or even an obstructive
force when the rollers are damaged. It is important for the
clinician to receive a constant haptic feedback from the probe
as the clinician need to know whether the TEE is getting
obstructed by a tissue or not. Furthermore in [6], a different
marking system is illustrated where the signals are totalized.
The drawback of the marking system in [6] is that if one single
line gets erased or broken due to wear and tear, then the entire
depth calculation is compromised. This happens because the
depth value is calculated based on signal accumulation.
II. METHODOLOGY
A visual tracking based solution using fiducial markers
printed on the probe is proposed to enable automatic detection
of the probe depth and make it available to the system for easy
and quick procedures. The proposed system comprises of three
main modules viz. a sensor appended bite guard, a specially
patterned TEE probe and an pattern analysis algorithm to
compute the depth value.
A. Sensor Appended Bite Guard
The vision sensor for the proposed system needs to be
appended in bite guards which are commonly used in TEE
procedures. A 3D model of the sensor appended bite guard is
shown in Figure 3.
Fig. 3. Example model of the mouth guard or similar ring structure integrated
with imaging module.
The vision sensor is also accompanied by a set of diffused
LEDs for making the system invariant to illumination.
B. Patterned TEE Probe
The proposed solution incorporates a novel binary marking
scheme on the probe replacing the existing numeric textual
markings. This enables an automatic detection of probe depth
which can be sent to the ultrasound machine for further uses.
An example model of the probe shaft with the proposed
changes incorporated is shown in Figure 4.
Fig. 4. An example 3D model showing the proposed pattern on TEE probe.
The proposed marker distribution for a length of 10 cm
is illustrated in Figure 5. As shown in Figure 5, the proposed
patterns are classified into three viz. major tick, minor tick and
directional markers.
Fig. 5. A pattern group representing a unit measurement.
All the major tick markers used in the probe are designed
to be unique in terms of shape and encoded information.
However, all the minor ticks deployed between the major
ticks are same and they represent only the amount of incre-
ment/decrement required with reference to a recently visited
major tick.
Apart from the commonly used square shaped markers [7],
we’ve proposed a rectangular marker for encoding numeric
values representing the depth and white vertical lines to
incorporate directional information. As shown in Figure 5, the
white vertical line markers will leave black spacings a,band c
and the band spacing will be in such a way that the inequality
1 is maintained.
(a+b)/2<(b+c)/2(1)
This type of space encoding is sufficient to distinguish between
advancing and withdrawing motion of the probe. This helps in
obtaining a sense of direction and based on the direction, the
computation module performs either a delta decrement or a
delta increment on the last observed major tick depth value.
The standard square markers were replaced with rectangular
ones so as to increase encoding capacity and also to ensure
the visual availability of at least one marker to the imaging
sensor all the times.
The major markers are represented by a 2×3binary matrix
as shown in Figure 5. The size of the matrix is determined
by factors like the width of the tubular shaft, distance from
camera to the shaft, resolution of the camera etc. Suppose a
generalized M×Nbinary matrix of a marker blob is given
by
B=
x0,0x0,1x0,2. . . x0,N1
x1,0x1,1x1,2. . . x1,N1
.
.
..
.
..
.
.....
.
.
xM1,0xM1,1xM1,2. . . xM1,N1
The binary matrix Bis required to satify the following
constraints in order to preserve the dimensions of the white
binary blob.
N1
X
i=0
xj,i 6= 0 j∈ {0,1, ..., N 1}(2)
M1
X
i=0
xi,j 6= 0 j∈ {0, N 1}(3)
Equation 2 and 3 will help maintaining a constant height
and width respectively. The constraints in Equation 2 and 3 will
also make an impact in the encoding capacity of the matrix.
For our TEE application, the 2×3matrix offered sufficient
binary encoding capacity.
C. Algorithm to Read Patterns
The use of visual markers [7] is a well known technique in
computer vision applications especially for object localization.
Commonly used fiducial markers are square shaped and we
cannot use them directly on the probe due to various con-
straints like the camera view limitations due to the curved
nature of the probe, the need for a white background for
localization etc. As a solution, we propose a rectangular marker
for which we can ascertain that at least one full marker will
be always available to the image sensor.
Fig. 6. An example depiction of the probe shaft moving through the bite
guard.
Fig. 7. An abstract view of the data flow and the steps involved in the
proposed system.
The proposed system is intended to function in a real-
time environment. A real-time high frame rate image capturing
sensor is required to capture the TEE probe while it is passing
through the mouth guard/ring structure. The captured raw
frames can contain both the TEE probe as well as some back-
ground data. Obviously the background will be of a uniform
color as we are capturing data from the inner surface of the ring
region. This can be understood from the depiction in Figure
6. This helps to easily segment out the probe shaft region
from the background. The registration process only involves
marginal clockwise or anti-clockwise rotations so that we are
left with perfect vertical or horizontal edges in the image. Only
marginal rotational registration is required because the probe
shaft can only have a highly restricted movement through the
bite guard. In addition, the normalization process will adjust
the scale by taking any of the vertical markers as a reference.
The presence of major and minor tick markers are identified
by sensing the horizontal edges appearing at the horizontal
center of the image. Similarly, the presence of directional
marker can be detected by analyzing the vertical edges.
The pattern data received over time is tracked and con-
verted to a depth value and is then classified into one of
the standard views shown in Table. I. The depth information
computed in the previous step is displayed on the computer
screen at the convenience of the operator. This can include
Algorithm 1 Depth Estimator
1: direction flag ← −1,
2: minor tick code 12,
3: depth value 0
4: lef t band size LEF T S IZ E,
5: right band size RIGHT SI ZE
6: current frame Read frame from the sensor.
7: width Width of current f rame
8: height Height of current frame
9: segmented image Subtract background from
current f rame
10: registered image Rotated segmented image to pre-
serve only vertical and horizontal edges.
11: vertical edges Vertical edges in
scale normalized image.
12: horizontal edges Horizontal edges in
scale normalized image.
13: vertical strip Sub-image (from vertical edges) of
size width/2×1.
14: horizontal strip Sub-image (from
horizontal edges) of size width/2×1.
15: edges vertical edges horizontal edges
16: if non-zero values in edges then
17: if non-zero values in horizontal strip > 0then
18: major tick value Call
Code Reader(registered image)
19: if major tick value == minor tick code then
20: return depth value +(minor tick value ×
direction f lag)and goto step 3
21: else
22: depth value major tick value
23: return major tick value and goto step 3
24: end if
25: end if
26: if non-zero values in vertical strip > 0then
27: if left and right bands available then
28: average band size Average size of left &
right black bands
29: if |average band size lef t band size|>
|average band size right band size|then
30: direction flag 1and goto step 3
31: end if
32: direction flag ← −1and goto step 3
33: end if
34: end if
35: else
36: goto step 3
37: end if
38: depth value depth value +(minor tick value ×
direction f lag)
39: return depth value and goto step 3
alerting the operator while going outside a preset view. The
depth value when available to the system opens up a new
line of efficient ultrasound workflow which offers a faster
and distraction free examination. The processing logic can
be executed either within the ultrasound scanner or using an
external processor for faster adaptation. The ring structure that
houses the imaging module will have to run wires to power
the imaging module.
Algorithm 2 Code Reader
1: blobs Array of all connected components in
binary image
2: blob count Number of elements in blobs
3: maximum area 0,code string ””
4: largest blob NU LL
5: for i= 1 to blob count in steps of 1do
6: current blob area Area of blobs[i]
7: if current blob area > maximum area then
8: maximum area current blob area
9: largest blob blobs[i]
10: end if
11: end for
12: largest blob size Width of largest blob×Height of
largest blob
13: Write largest blob data to code image of size
largest blob size
14: code image width Width of code image
15: code image height Height of code image
16: for i= 1 to code image height 2in steps of
code image height/2do
17: for j= 1 to code image width 2in steps of
code image weight/3do
18: sub image Sub-image from code image(j, i)
of size code image height/2×code image weight/3
19: if non-zero pixels in sub image > 0then
20: code string =code string + ”0”
21: else
22: code string =code string + ”1”
23: end if
24: end for
25: end for
26: code code string mapped to a numeric value.
27: return code
III. RES ULTS A ND DISCUSSION
In this study, we’ve demonstrated through simulation, the
advantages of an automatic depth estimation technique for
TEE probe examinations. The software prototype developed
for the experiments can take synthetic pattern data which can
be moved in advancing or withdrawing directions using a slider
control. The simulated image sensor will take a limited view
of the patterns which comes right beneath it. A screenshot of
the developed software is shown in Figure 8.
As shown in Figure 8, only a particular portion of the
patterns will be visible to the sensor at a time. The other
portion of the long shaft data can be slided in using a slider
control and the processing logic will compute and update the
depth value. The current anatomical context can be inferred
from the depth value and can be used for automated workflow
triggers and helps in faster procedures without any distraction.
The proposed solution advocates integration of an imaging
sensor directly into the mouth guard. Even though there are
reusable bite guards, the solution becomes a problem in case
of disposable bite guards as they are meant to be disposed after
use. A detachable imaging module would still raise a question
of hygiene. As an alternate solution, we can also integrate the
imaging sensor to another ring structure which can be then
safely appended to the disposable original bite guard. The
Fig. 8. Software prototype developed for processing TEE shaft data.
depth measurements would then simply use an offset based
on the size of the appended object.
IV. CONCLUSION
A marker based automatic depth estimation for TEE exam-
ination is proposed to improve the existing slow and inefficient
workflow. In contrast to the prior-art solutions like [6], the pro-
posed one aims to provide a non-contact solution purely based
on computer vision. There is no need for doing unnecessary
probe manipulation for just reading the numbers printed over
the surface of the probe as every data is automatically available
to the system and on the computer screen. The solution thus
helps a clinician to do a seamless visual tracking on the
reconstructed B-Mode image data. The same technique can be
further extended to fit any depth reading problem for tubular
structure movements through a restricted space similar to a
bite guard.
ACKNOWLEDGMENT
The authors would like to thank their reviewers in the
intellectual property department of Siemens and also like to
express their appreciation to their colleagues in the Siemens
Ultrasound Division for providing comments, technical support
and other materials.
REFERENCES
[1] E. Foster and N. B. Schiffer, “The role of transesophageal echocardiog-
raphy in critical care: Ucsf experience,” Journal of the American Society
of Echocardiography, vol. 5, no. 4, pp. 368–374, 1992.
[2] J. B. Seward, B. K. Khanderia, J. K. OH, M. D. Abel, R. W. Hughes,
W. D. Edwards, B. A. Nichols, W. K. Freeman, and A. J. Tajik,
“Transesophageal echocardiography: technique, anatomic correlations,
implementation, and clinical applications,” in Mayo Clinic Proceedings,
vol. 63, no. 7. Elsevier, 1988, pp. 649–680.
[3] M. Jensen, E. Sloth, K. M. Larsen, and M. B. Schmidt, “Transthoracic
echocardiography for cardiopulmonary monitoring in intensive care,
European journal of anaesthesiology, vol. 21, no. 9, pp. 700–707, 2004.
[4] S. G. Memtsoudis, P. Rosenberger, M. Loffler, H. K. Eltzschig,
A. Mizuguchi, S. K. Shernan, and J. A. Fox, “The usefulness of
transesophageal echocardiography during intraoperative cardiac arrest in
noncardiac surgery,Anesthesia & Analgesia, vol. 102, no. 6, pp. 1653–
1657, 2006.
[5] H. K. Eltzschig, P. Rosenberger, M. L¨
offler, J. A. Fox, S. F. Aranki, and
S. K. Shernan, “Impact of intraoperative transesophageal echocardiogra-
phy on surgical decisions in 12,566 patients undergoing cardiac surgery,”
The Annals of Thoracic Surgery, vol. 85, no. 3, pp. 845–852, 2008.
[6] A. F. Bolger and C. Tacklind, “System and method for monitoring
intraluminal device position,” Aug. 1 1995, uS Patent 5,437,290.
[7] S. Garrido-Jurado, R. Mu˜
noz-Salinas, F. J. Madrid-Cuevas, and M. J.
Mar´
ın-Jim´
enez, “Automatic generation and detection of highly reliable
fiducial markers under occlusion,” Pattern Recognition, vol. 47, no. 6,
pp. 2280–2292, 2014.
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
This paper presents a fiducial marker system specially appropriated for camera pose estimation in applications such as augmented reality and robot localization. Three main contributions are presented. First, we propose an algorithm for generating configurable marker dictionaries (in size and number of bits) following a criterion to maximize the inter-marker distance and the number of bit transitions. In the process, we derive the maximum theoretical inter-marker distance that dictionaries of square binary markers can have. Second, a method for automatically detecting the markers and correcting possible errors is proposed. Third, a solution to the occlusion problem in augmented reality applications is shown. To that aim, multiple markers are combined with an occlusion mask calculated by color segmentation. The experiments conducted show that our proposal obtains dictionaries with higher inter-marker distances and lower false negative rates than state-of-the-art systems, and provides an effective solution to the occlusion problem.
Article
Full-text available
To evaluate the feasibility of an abbreviated focus assessed transthoracic echocardiographic protocol, consisting of four standardized acoustic views for cardiopulmonary screening and monitoring. The protocol was applied in 210 patients in a 20-bed multidisciplinary intensive care unit in a university hospital. When inconclusive, an additional transoesophageal echocardiographic examination was performed. Diagnosis, indication, acoustic window, position and value were recorded. Significant pathology, load, dimensions and contractility were assessed. Two-hundred-and-thirty-three transthoracic and four transoesophageal echoes were performed. The protocol provided usable images of the heart in 97% of the patients, 58% subcostal, 80% apical and 69% parasternal. Images through one window were obtainable in 23%, through two windows in 41% and through three windows in 34%. In 227 patients (97.4%) the focus assessed echo protocol contributed positively. In 24.5% of cases the information was decisive, in 37.3% supplemental and in 35.6% supportive. By means of an abbreviated, focus assessed transthoracic echo protocol it is feasible to visualize the haemodynamic determinants for assessment and optimization. One or more useful images are obtainable in 97% of critically ill patients.
Article
The applications of transesophageal echocardiography (TEE) now encompass the arena of critical care. We reviewed the indications, findings, and outcome of TEE in the critical care setting in a tertiary care hospital and report our experience. Eighty-three transesophageal echocardiograms were performed in 69 critically ill patients for the following indications: endocarditis (43%), embolic source (13%), hypotension (10%), mitral regurgitation (10%), left ventricular function (6%), aortic dissection (4%), prosthetic valve dissection (4%), and other (10%). Findings were unexpected in 21 of 83 studies and led to a change in management in 17%. On the basis of the results of TEE, 22% of patients had further evaluation by a more invasive technique, and 19% of patients had surgical intervention without further study. No significant complications attributable to TEE were noted. We conclude that TEE is a safe and useful diagnostic technique with fairly broad applications in the critical care setting.
Article
According to guidelines established by the American Society of Anesthesiologists and the Society of Cardiovascular Anesthesiologists, life-threatening hemodynamic disturbances are classified as a category I indication for the intraoperative use of transesophageal echocardiography (TEE). However, the usefulness of TEE during intraoperative cardiac arrest and its impact on patient management have not been rigorously investigated. Using our departmental TEE database, we identified a population of 22 patients who underwent noncardiac surgical procedures and experienced unexpected intraoperative hemodynamic collapse requiring the initiation of Advanced Cardiac Life Support procedures between the time of induction of general anesthesia and the termination of the surgical procedure. Results of TEE examinations, patient records, detailed operative records, and outcome of patients were reviewed for the utility of TEE to diagnose the etiology of the hemodynamic collapse. Furthermore, the impact on subsequent patient management was evaluated. A primary suspected diagnosis of the underlying pathological process was established in 19 of 22 patients with TEE, including 9 with thromboembolic events, 6 with acute myocardial ischemia, 2 with hypovolemia, and 2 patients with pericardial tamponade. A definitive diagnosis could not be made in 3 patients with TEE. In 18 patients, TEE guided specific management beyond implementation of Advanced Cardiac Life Support protocols, including the addition of surgical procedures in 12 patients. Fourteen patients survived to leave the operating room, and 7 of these patients were eventually discharged from the hospital. Thus, TEE may provide additional diagnostic information in patients with intraoperative cardiac arrest and may directly guide specific, potentially life-saving therapy.
Article
The utility of intraoperative transesophageal echocardiography (TEE) for different types of cardiac surgical procedures has not been thoroughly investigated despite its increasing popularity. Therefore, we retrospectively evaluated the impact of before and after cardiopulmonary bypass (CPB) TEE on surgical decisions in 12,566 consecutive patients undergoing cardiac surgery at a single institution. We analyzed all patients undergoing cardiac surgical procedures who had an intraoperative TEE examination between 1990 and 2005 at the Brigham and Women's Hospital. Results of the TEE examinations were entered into a database. Previously undiagnosed TEE findings from the pre- and post-CPB examinations that directly impacted surgical decisions were evaluated. Before and after CPB TEE examinations influenced surgical decision making in 7.0% and 2.2%, respectively, of all evaluated patients (n = 12,566). In patients undergoing only coronary artery bypass graft surgery (CABG [n = 3,835]), surgical decisions were influenced by 5.4% of the pre-CPB and 1.5% of the post-CPB TEE examinations, and in 6.3% and 3.3%, respectively, of those patients undergoing isolated valve procedures (n = 3,840). In combined CABG and valve procedures (n = 2,944), surgical decisions were influenced by 12.3% of the pre-CPB and 2.2% of the post-CPB TEE examinations. Intraoperative TEE influences cardiac surgical decisions in more than 9% of all patients in the presented study population, with the greatest observed impact in patients undergoing combined CABG and valve procedures.
Transesophageal echocardiography: technique, anatomic correlations, implementation, and clinical applications
  • J B Seward
  • B K Khanderia
  • J K Oh
  • M D Abel
  • R W Hughes
  • W D Edwards
  • B A Nichols
  • W K Freeman
  • A J Tajik
J. B. Seward, B. K. Khanderia, J. K. OH, M. D. Abel, R. W. Hughes, W. D. Edwards, B. A. Nichols, W. K. Freeman, and A. J. Tajik, "Transesophageal echocardiography: technique, anatomic correlations, implementation, and clinical applications," in Mayo Clinic Proceedings, vol. 63, no. 7. Elsevier, 1988, pp. 649-680.
System and method for monitoring intraluminal device position
  • A F Bolger
  • C Tacklind
A. F. Bolger and C. Tacklind, "System and method for monitoring intraluminal device position," Aug. 1 1995, uS Patent 5,437,290.