Professional Documents
Culture Documents
Understanding the
ImageChecker CT Lung System
®
Version 2.0
CT Lung
Understanding the
ImageChecker® CT Lung System
Version 2.0
PN 13229 Rev A
Technical Support
Before calling R2 Technical Support, refer to the user manuals for
troubleshooting instructions. For support in North America, contact:
Toll Free: +1 866.243.2533 (866.CHECKED)
Fax: +1 408.481.5607
Email: techsupport@r2tech.com
Hours: Monday through Friday, 6:00 AM – 5:00 PM, PST
For support in Europe, South America, or Asia, contact your local dealer or
distributor.
Caution: Only service engineers certified by R2 Technology are qualified to
install or repair the system.
R2 Technology, Inc.
1195 West Fremont Avenue
Sunnyvale, CA 94087 USA
www.r2tech.com
This system incorporates the Interface Definition Language Compiler Front End
(IDL CFE) and Internet Inter-ORB Protocol (IIOP) technologies, provided and
licensed by Sun Microsystems. Refer to the file 3rdPartyLicenses.txt in the
c:/r2technology/imagechecker/vendor/3rdpartylic folder on the processing unit
for licensing and warranty information.
Part 2: Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
System Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Optional Software Packages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
System Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
ImageChecker CT Lung Server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
ImageChecker CT Lung Workstation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Other Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Key Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Separate Image/CAD Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
User-Controlled CAD Markers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Findings Automatically Quantified . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Flexible Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Integrated 2D and 3D Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Frontal Projection View (Lung Map) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Selection of Review Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Complementary Data Views . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
AutoPoint Temporal Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Pulmonary Artery PE (Patency Exam) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Intended Use
The ImageChecker® CT Lung v2.0 System is a computer-aided detection (CAD)
system designed to assist physicians in softcopy review of digital computed-
tomography (CT) images of the chest. The ImageChecker CT Lung workstation
displays composite views of 2D cross-sections and 3D volumes of chest CT images,
including findings or regions of interest identified by the physician or computer-
aided detection (CAD) findings.
The system assists physicians in the characterization, classification, and tracking of
suspicious candidate thoracic abnormalities in terms of size, dimension, shape,
density, and position, and thus aids in the patient management care decision process.
The system is intended to be used as an adjunct, alerting the physician – after an
initial reading of the scan – to regions of interest that may have been overlooked.
Available Resources
The following resources are available when you are working with the ImageChecker
CT Lung System:
• Onscreen messages – The ImageChecker CT Lung System is designed to guide you
through any interaction you will need to have with the system. Informational
onscreen messages pop up when appropriate.
• R2 Technical Support and Service – For contact information, see ‘Technical
Support’ on the back of the title page of this book.
• User Manuals – Instructions for using the system are provided in this manual and
the ImageChecker CT Lung v2.0 User Manual. The user manual is a how-to guide
in the form of a simple, streamlined flip chart that describes how to use the
application. To order additional copies of the manuals, contact R2 Technical
Support. Note: The instructions in the user manual apply only when using the
ImageChecker CT Lung workstation supplied by R2 Technology, Inc.
• Algorithm Documentation – Under the Algorithm Description tab in the user
manual you will find an explanation, with examples, of how the ImageChecker CT
Lung System algorithm processes CT images. For a more general description,
without examples, see ‘CAD Algorithm Description’ on page 14 of this manual.
• DICOM Conformance Statements – The R2 Technology website
(www.r2tech.com) provides the DICOM Conformance Statements, which describe
the ImageChecker CT Lung implementation of the DICOM protocol.
• Release Notes – This document describes new product features and presents any
known issues or questions that are commonly asked.
• Customer Bulletins – We are committed to making our products safe and easy to
use. If an issue arises or new features become available, we will send you a
Customer Bulletin.
• Training – The R2 Technology Applications team is available to train your staff,
should you feel they need additional training. Contact your R2 Technology
Account Manager if you want personalized instruction.
Symbols Used
The following internationally recognized symbols are used with the system.
Symbol Description
Dangerous voltage
• The CAD software operates only on CT chest images that meet the technical
requirements outlined in ‘Part 3: Scanning Protocol’ and the DICOM
Conformance Statements for the ImageChecker CT Lung workstation and server.
• Image quality conditions such as underexposure or artifacts (e.g., motion artifacts
and streak artifacts caused by pacemakers or wires, for example) that diminish
radiographic contrast may also diminish CAD software sensitivity.
• Scans grossly distorted due to motion, and scans with numerous streak artifacts,
are likely to be repeated, if possible, since either condition would affect both
human and CAD performance.
• Individual practice patterns may influence the results obtained when using the
CAD software. Therefore, each facility and physician should carefully monitor the
result that this CAD software has on their practice in order to optimize its
effectiveness.
• Prior to review, ensure that multiple thoracic CT series being compared by the
software are from the same patient.
• If the automatic alignment tool fails to align the cases appropriately, the user
should use the manual alignment tools for accurate alignment.
• All comparative volumetric statistics such as volume change and doubling time
presented by the software must be assessed by the interpretative skills of the user.
• Primary diagnosis of the patient should not be based solely on volume change and
doubling time information presented, or the low-density regions highlighted by
the software.
hazard. Ensure that the system is connected to a power receptacle that is properly
grounded and provides voltage and current within the specifications of the system
to prevent the possibility of electrical shock or fire hazard.
• Do not place liquid containers on the device. In the event of a spill, shut down
power to all components prior to cleaning to prevent the possibility of electrical
shock. Do not operate the device if internal components are exposed to liquid.
Contact R2 technical support.
• This equipment has been tested and found to comply with the limits for a Class A
digital device, pursuant to Part 15 of the FCC Rules. These limits are designed to
provide reasonable protection against harmful interference when the equipment is
operated in a commercial environment. This equipment generates, uses, and can
radiate radio frequency energy and, if not installed in accordance with the
instruction manual, may cause harmful interference to radio communications.
Operation of this equipment in a residential area is likely to cause harmful
interference in which case the user will be required to correct the interference at
the user’s own expense.
Part 2 describes the ImageChecker CT Lung System and key system features.
System Overview
The ImageChecker CT Lung System is an advanced thoracic 3D application designed
to assist physicians in the detection of thoracic abnormalities during review of multi-
slice CT (MSCT) exams of the chest. It is to be used as an adjunctive reader,
providing the physician with ‘a second set of eyes’ with which to find, assess, and
track abnormalities that may represent clinically significant disease.
The system does this by applying proprietary signal processing CAD algorithms to the
large digital datasets generated during scanning. These algorithms analyze the
complete set of images and search for findings with features suggestive of thoracic
abnormalities such as pulmonary nodules. The system conveys information regarding
candidate nodules of current and prior studies to the application for consideration by
the reader, including a marker noting each candidate nodule’s location.
The system provides fast, interactive nodule segmentation and volume measurement,
coronal 3D lung projection with stack navigation and nodule localization, as well as a
lung 3D view port that displays views of nodules and related anatomy with automatic
vessel, airway, and chest wall segmentation. After performing the initial read of the
exam, the user activates the display of the markers to review the computer-identified
findings.
The system facilitates case analysis by providing a complete toolkit for the measure-
ment and management of candidate nodules. Findings and measurements are
summarized in a database for subsequent review. This summary can be printed to any
Windows-compatible printer. The key images in the summary can be exported as
separate graphics files.
The system reduces the chance of performance lapses by the reader due to fatigue,
distraction, satisfaction of search, or other recognized causes of a suboptimal review;
it thereby increases the likelihood of identifying a clinically significant abnormality at
an earlier, more treatable stage.
Note that the ImageChecker CT Lung System is not intended to be used principally in
a screening environment, where the vast majority of scans would be expected to be
System Components
The ImageChecker CT Lung System consists of two computers, a review workstation
and a server. The workstation may either be provided by R2 Technology or be any
DICOM compliant PACS workstation. The graphic shows the principal system
components and actions.
After the initial study acquisition by the scanner, images typically are sent to a PACS
archive (1). Then, the ImageChecker CT Lung application either queries the PACS
archive for images (at which point the images are routed to the ImageChecker CT
Lung server for CAD processing) or the images are automatically routed from the
PACS archive to the ImageChecker CT Lung server (2).
After the server processes the image data and completes the CAD segmentation
processing, it sends the results to the workstation in the form of a DICOM structured
report. You can then choose to display the CAD findings (suspected nodules and/or
filling defects) (3), and use the application’s many tools to assist in your review. In
addition, when configured to do so, the system can store a secondary series of images
with CAD findings or an overlay of CAD findings to the PACS archive for later
retrieval by a PACS workstation. (4)
When you complete your review, the system makes a summary report available that
lists any findings and measurements associated with the study. You can print this
summary as well as any key images (5).
When configured to do so, the application archives the DICOM structured report
along with any key image series for nodules, pulmonary emboli, measurements and
3D captures (6).
• Aids image review – The application presents a single screen that shows various
display representations:
2D Review Window – The original 2D axial images (slices). You will be able to
perform functions that are commonly available in other review workstations,
such as Window and Level, Pan, Zoom, Manual paging through the images,
and distance measurements. You can also choose to switch this view to a thick
slice or maximum intensity projection (MIP) view of the data.
Lung Map (Frontal Projection View) – An anterior-posterior coronal
projection view with the mediastinum and chest wall removed automatically.
You can interact with this image, which acts like a localizer. When your cursor
touches a point on the map, the 2D Review Window will display the slice from
the corresponding location.
3D Review Window – A neighborhood around a point that is selected by the
user in an axial slice. This 3D image presents an image that shows lungs, vessels,
airways and other objects within the lung parenchyma. You can rotate and
magnify this image to further analyze a suspicious finding.
• Allows region of interest and finding selections by user – This phase can occur in
conjunction with the image review or immediately after. During this phase, the
physician reviews the 2D axial images (slices) and selects a point considered to be a
region of interest. If this point is part of a composite feature found during the
preprocessing stage, the characteristics of the region (e.g., volume, average density)
are displayed. Based on experience, the reviewer can decide to mark the region as a
finding. If the selected point is not part of a composite feature, the reviewer can
perform distance measurements on it as well as obtain density values. This
information may help the reviewer decide whether or not to mark the feature as a
finding.
• Permits temporal comparison – With the optional AutoPoint temporal
comparison tool, the application displays the current and selected prior study
simultaneously and aligned. This screen provides all the 2D review tools, enables
you to link, either manually or automatically, the findings from the current and
prior, as well as add and discard findings.
• Aids in detection of pulmonary arterial filling defects – With the optional
pulmonary artery PE (Patency Exam) tool, the application highlights regions of
low density within the vasculature of the lung when the image is enhanced through
the use of contrast media. The screen presentation provides 2D and 3D review
tools plus reporting capability.
• Creates a summary – After reviewing the exam, the physician can review the
summary of the findings. Depending on the optional software packages you have,
this summary can include the summary of any temporal comparison or
pulmonary arterial filling defect detection performed. A physician or other user
can review all of the findings in a tabular form that includes all measurements
made for that point. For each finding you can also see the corresponding axial
image (slice) with an indicator showing the finding. You can print the summary
report on paper and attach it to the exam record or save the summary on the
workstation for later review.
Other Components
The ImageChecker CT Lung System also comes with the following:
• Documentation (user and service)
• Licensing keys (one for each computer received from R2 Technology)
• Ethernet cables (for connecting the system to the LAN)
Key Features
While this section provides a list of the most prominent system features, it is not
exhaustive. For information on how to use the features described below, refer to the
ImageChecker CT Lung System User Manual.
Flexible Algorithm
The R2 Technology CAD algorithm supports multiple multi-detector row computed
tomography vendors and acquisition protocols.
Controlled Access
Everyone who uses the application must have a user ID and password in order to log
on to the workstation. This is a security measure to ensure that only authorized
personnel can use the application enhancing patient privacy.
Queue Manager
The ImageChecker CT Lung server provides a queue manager that enables clinical
staff to monitor and manage the CAD segmentation processing of studies. You can
reprioritize, resend, and delete studies. The queue manager is available on the
workstation Study List, Server View screen.
For more details about the algorithm, see ‘Part 3: Scanning Protocol’ and the
Algorithm Description tab in the ImageChecker CT Lung User Manual.
System Maintenance
The ImageChecker CT Lung System is a software system and as such does not require
regular maintenance. On a regular basis, summary files are automatically saved to the
configured default archive.
Note: The specifications in Part 3 are subject to change as new applications are
available and/or the ImageChecker CT Lung System supports additional acquisition
protocols.
The ImageChecker CT Lung System consists of a processing unit server and a
software application for image display on a workstation. The system provides
flexibility in the acquisition of MSCT chest exams acceptable for processing and
visualization. The flexibility includes:
• Support of chest CT scans from multiple MSCT vendors with four or more
detectors.
• Support for a range of high-resolution CT protocols (see parameters below).
• Support of exams performed without or with intravenous contrast agents.
• For use with all indications for chest CT studies.
• Most MSCT chest exams can be reviewed on the workstation as soon as they are
completed/processed.
Scanning Specifications
The following statements define the acceptable acquisition protocol for applying the
CAD algorithms for lung-nodule detection and filling-defect analysis. Exceptions are
noted.
Input Data
ImageChecker CT Lung can display a wide range of chest CT exams, but to be eligible
for CAD processing, the acquisition protocols must meet the following requirements:
• Images acquired with a 4- or more detector MSCT scanner.
• Collimation of 3 mm or less that results in image reconstruction with slice
thickness of 3 mm or less (2 mm or less for filling-defect analysis) as reported by
DICOM Slice Thickness parameter (0018, 0050).
• Image spacing less than or equal to slice thickness (image overlap is acceptable, but
gaps are not).
• Slice thickness is the same throughout the series.
• Slice spacing is the same throughout the series.
• Tube voltage no less than 20 keV.
• Exposure no less than 15 mAs.
• Series containing lung should not have more than 1,000 images.
Reconstruction Filters
ImageChecker CT Lung accepts MSCT exams acquired with common acquisition
protocols. In particular, GE ‘Standard’, ‘Bone’ and ‘Lung’ reconstruction algorithms
(and their equivalent for other manufacturers) are allowed.
When comparing studies, always compare series that have been processed with the
same reconstruction filter, as automatic volume measurements may vary with
different filters.
Preliminary investigation has shown that the reconstruction algorithm does not affect
the CAD sensitivity; however the false-positive marker rate may vary with edge-
enhancing reconstruction algorithms
Contrast Support
ImageChecker CT Lung accepts exams with and without intravenous contrast media.
The CAD algorithms perform equally well in exams with and without intravenous
contrast media. Note that contrast medium is required for Perfusion Exam analysis.
Merged Exams
ImageChecker CT Lung supports continuous volumetric reconstructions. It does not
support series that are the result of merging two acquisitions that result in two or
more slices on the same location or nonuniform image spacing.
Gated Exams
ImageChecker CT Lung does support MSCT exams acquired with cardiac gating.
CT lung scans should be completed during a single held breath to avoid motion
artifacts and ensure optimal processing.
Gantry Tilt
Indications for chest MSCT exams with gantry tilt are rare and ImageChecker CT
Lung will reject such exams for processing or review.
DICOM Fields
ImageChecker CT Lung will reject exams where the following DICOM fields are
empty. This is an uncommon scenario for patient scanning but the customer is
reminded that these fields must be filled in either by the technologist or by the
scanner:
Compressed Images
ImageChecker CT Lung does not accept images that have been subjected to
compression (lossy or lossless).
Can a CTA exam be used for CAD lung nodule detection in ImageChecker CT Lung?
Yes. The presence of intravenous contrast media is taken into consideration by the
CAD algorithm in the automatic detection process as long as the other scanning
protocol criteria are met.
Can ImageChecker CT Lung be used with MSCT lung cancer screening exams?
Yes, as long as the exams meet the input parameters described above.
What about patients with staples or other metallic objects in the thorax or
mediastinum?
These cases are acceptable for review and will be preprocessed for 3D and CAD if they
are acquired with scanning techniques that minimize the presence of artifacts.
What is the minimum vessel diameter in which the CAD algorithm will detect
filling defects?
Four (4) mm
Part 4 describes the clinical studies submitted to the United States Food and Drug
Administration (FDA) for approval of the ImageChecker CT Lung System.
Objectives
R2 Technology, Inc. has conducted three pivotal clinical studies to evaluate the safety
and effectiveness of the ImageChecker CT Lung System. The studies were based on a
retrospective case collection project involving multiple clinical sites in various regions
across the United States. The clinical studies had the following objectives:
• The first study was designed to generate a ‘truth’ set of cases containing solid
pulmonary nodules, as well as cases with no nodules, to be used as a reference
truth for subsequent studies.
• The second study, conducted in two phases, was an Observer/ROC (Receiver
Operating Characteristic) Study designed to measure the performance
enhancement of radiologists using the system.
• The third study was a retrospective study to characterize the stand-alone sensitivity
of the CAD software.
The clinical studies did not attempt to assess the effectiveness of the device on an
asymptomatic lung cancer screening population. The applicability of this device to
such a CT screen population has not been established.
The patient was a 67-year-old male, with a history of bladder cancer. A small 4.4 mm
noncalcified nodule was present in the right lung, in the lower quadrant, marked by
the CAD algorithm. Only eleven of the 15 radiologists rated that quadrant as
containing an actionable nodule in their initial review, whereas three more
radiologists increased their ratings after viewing the CAD marks.
The area under the receiver operator curve (AUC) was computed for each of the 15
radiologists before and after using CAD. The average curve of all 15 radiologists is
shown below for the 90-case study. The AUC increased with the use of CAD. If the
full plot is viewed as a unit square, the area separating the two curves in the Phase II
study is 0.024.
Average ROC Curve Showing Pre-CAD (dashed line) and Post-CAD (solid line)
1. Dorfmann DD, Berbaum KS and Metz, CE. Receiver Operating Characteristic Rating
Analysis. Invest Radiol 1992; 27: 723-731.
2. Rutter, C. Bootstrap Estimation of Diagnostic Accuracy with Patient-Clustered Data. Acad
Radiol 2000; 7: 413-419.
of the validity of the bootstrap mechanism, the analysis was performed first using the
unanimous reference truth. As shown in Table 2, the results of the primary ANOVA-
after-jackknife analysis and the bootstrap analysis using the unanimous reference
truth are very similar.
Table 2: Study analyses of the significance of the improvement in area under ROC curve
when reference truth or statistical methods were varied.
Estimated
Improvement
Analysis Method in AUC p-value 95% CI
Primary: ANOVA-after-jackknife, 0.0240 0.003 (0.0084, 0.0395)
with unanimous reference truth
Secondary: ANOVA-after-jackknife, 0.0213 0.001 (0.0097, 0.0329)
with 2/3 majority reference truth
Secondary: Bootstrap with 0.0246 <0.001 (0.0089, 0.0446)
unanimous reference truth
Secondary: Bootstrap with random 0.0224 <0.001 (0.0076, 0.0403)
3-panel reference truth
Secondary: Bootstrap with random 0.0216 0.002 (0.0077, 0.0387)
2-panel reference truth
Secondary: Bootstrap with random 0.0209 <0.001 (0.0080, 0.0370)
1-panel reference truth
Finally, an analysis at the patient level, rather than at the quadrant level, was
conducted to determine if there was evidence to show a positive effect of the use of
CAD using a methodology closer to the clinical use of the product.
Each case in the study was assigned to one case cohort based on what truth nodules
were present (depending on the definition of the ‘truth’ reference standard). For
example, cases that contained at least one unanimous actionable nodule were placed
in a case cohort that would be considered true nodule-present cases. Each case was
also assigned a rating of ‘actionable’ if at least one quadrant had a rating of 50 or
greater (on our 0–100 scale). A threshold of 50 was chosen based on the way the
readers were instructed to use the scale (a rating of 50 was ‘indeterminate’) and the
actual ratings data showed a bimodal distribution, with most ratings clearly above 50
or below 50. Thus, we can calculate the effect of CAD at the patient level in terms of
reduction in observational oversights (i.e., the number of times a patient had a rating
change from below 50 pre-CAD to above 50 post-CAD in a case that contained a
truth nodule) and change in the false-positive rate (the same ratings change for a
patient with no nodules). This analysis showed that, at the case level, there were 22
corrections of 86 oversights (25.6%) and an increase in the false-positive rate (8
increases in 57 cases; 14.0%).
In summary, the primary analysis of the ImageChecker CT Lung System study shows
a statistically significant improvement in the AUC for an ROC analysis for detecting
solid pulmonary nodules between 4–30 mm in diameter. This result is robust when
different reference truth definitions are used in the analysis.
3. Fraser RS, Muller NL, Colman N, Paré PD, Eds. Fraser and Paré’s Diagnosis of Diseases of
the Chest, Fourth Edition, Volume 1. W.B. Saunders, Philadelphia, 1999: page xxxv. The
definition is: ‘Round opacity, at least moderately well marginated and no greater than 3 cm
in maximum diameter.’
4. ibid, Figure 18–31, page 458.
Table 3: CAD Stand-alone sensitivity numbers for nodules based on classic ratings.
Nodule Classification
Panel Members Rating # of CAD 95% Lower 95% Upper
Nodule as Classic Nodules Sensitivity* Limit Limit
4/5, 5/5 ‘classic’ 64 83% 73% 92%
3/5 13 69% 44% 94%
0/5, 1/5, 2/5 65 32% 21% 44%
*CAD Sensitivity is defined as the number of nodules marked by CAD divided by the total
number of nodules in each category.
A further analysis of the case sensitivity for cases with and without intravenous
contrast media showed consistent results in the cases with contrast compared to the
cases without contrast. The sensitivity for the classic nodules in cases with contrast
was 94% (16/17, 95% CI 83–100%) and for the classic nodules without contrast was
79% (37/47, 95% CI 67–90%). The sensitivity for the solid nodules in cases with
contrast was 55% (34/62, 95% CI 43–67%), and in cases without contrast was 61%
(49/80, 95% CI 51–72%). The false-marker rate measured on the nodule-absent cases
was the same in both groups and the same as the summary numbers above.
These results confirm that the ImageChecker CT Lung System performs at a high
level of sensitivity for classic pulmonary nodules, with low false-marker rates per
normal case.