`
`Magnetic Resonance Imaging 30 (2012) 1323 – 1341
`
`3D Slicer as an image computing platform for the Quantitative
`Imaging Network
`Andriy Fedorov a,⁎, Reinhard Beichel b, Jayashree Kalpathy-Cramer c, Julien Finet d,
`Jean-Christophe Fillion-Robin d, Sonia Pujol a, Christian Bauer b, Dominique Jennings c,
`Fiona Fennessy a, Milan Sonka b, John Buatti b, Stephen Aylward d,
`James V. Miller e, Steve Pieper f, Ron Kikinis a
`aBrigham and Women's Hospital, Harvard Medical School, Boston, MA 02115, USA
`bUniversity of Iowa, Iowa City, IA 52242, USA
`cMassachusetts General Hospital, Harvard Medical School, Charlestown, MA 02129, USA
`dKitware Inc, Clifton Park, NY 12065, USA
`eGE Research, Niskayuna, NY 12309, USA
`fIsomics, Inc., Cambridge, MA 02138, USA
`Received 1 February 2012; revised 26 April 2012; accepted 29 May 2012
`
`Abstract
`
`Quantitative analysis has tremendous but mostly unrealized potential in healthcare to support objective and accurate interpretation of the
`clinical imaging. In 2008, the National Cancer Institute began building the Quantitative Imaging Network (QIN) initiative with the goal of
`advancing quantitative imaging in the context of personalized therapy and evaluation of treatment response. Computerized analysis is an
`important component contributing to reproducibility and efficiency of the quantitative imaging techniques. The success of quantitative
`imaging is contingent on robust analysis methods and software tools to bring these methods from bench to bedside.
`3D Slicer is a free open-source software application for medical image computing. As a clinical research tool, 3D Slicer is similar to a
`radiology workstation that supports versatile visualizations but also provides advanced functionality such as automated segmentation and
`registration for a variety of application domains. Unlike a typical radiology workstation, 3D Slicer is free and is not tied to specific hardware.
`As a programming platform, 3D Slicer facilitates translation and evaluation of the new quantitative methods by allowing the biomedical
`researcher to focus on the implementation of the algorithm and providing abstractions for the common tasks of data communication,
`visualization and user interface development. Compared to other tools that provide aspects of this functionality, 3D Slicer is fully open source
`and can be readily extended and redistributed. In addition, 3D Slicer is designed to facilitate the development of new functionality in the form
`of 3D Slicer extensions.
`In this paper, we present an overview of 3D Slicer as a platform for prototyping, development and evaluation of image analysis tools for
`clinical research applications. To illustrate the utility of the platform in the scope of QIN, we discuss several use cases of 3D Slicer by the
`existing QIN teams, and we elaborate on the future directions that can further facilitate development and validation of imaging biomarkers
`using 3D Slicer.
`© 2012 Elsevier Inc. All rights reserved.
`
`Keywords: Cancer imaging; Quantitative imaging; Software tools; Medical imaging; Imaging biomarkers; Image analysis; MRI; PET; CT; Brain; Head and
`neck; Prostate; Glioblastima; Cancer treatment response
`
`⁎ Corresponding author.
`E-mail address: fedorov@bwh.harvard.edu (A. Fedorov).
`
`0730-725X/$ – see front matter © 2012 Elsevier Inc. All rights reserved.
`http://dx.doi.org/10.1016/j.mri.2012.05.001
`
`1. Introduction
`
`Cancer is the leading cause of death in the developed
`world and the second leading cause of death in the
`developing countries [1]. With the incidence of cancer
`rapidly increasing, there is an immediate need for better
`understanding of this disease and for the development of the
`
`Medivis Exhibit 1020
`
`1
`
`
`
`1324
`
`A. Fedorov et al. / Magnetic Resonance Imaging 30 (2012) 1323–1341
`
`targeted, personalized treatment approaches. Successful
`translation of such treatments from the lab to the clinic is
`contingent on the availability of biomarkers — objective and
`testable characteristics indicative of normal or pathologic
`processes that ideally should allow for quantitative mea-
`surement of the response to therapy [2,3]. In this regard, in
`vivo imaging biomarkers are particularly promising, as they
`can be highly specific and minimally invasive, providing
`both anatomical and functional understanding of
`the
`response patterns. However, the potential of quantitative
`imaging remains largely underutilized. The Response
`Evaluation Criteria in Solid Tumors — the only imaging-
`based biomarker accepted by the US FDA as a surrogate end
`point for clinical outcome in therapy — rely primarily on the
`anatomical imaging of the lesion measured by its largest
`diameter [4,5]. Continuous advances in multimodality three-
`dimensional (3D) imaging technology and analysis, along
`with improvements in computer science and bioinformatics,
`create an opportunity for a paradigm shift in quantification of
`treatment response. To advance the role of imaging as a
`biomarker of
`treatment,
`the National Cancer
`Institute
`launched the Quantitative Imaging Network (QIN) initiative
`[6]. The goal of QIN is to form a community of
`interdisciplinary teams engaged in the development of
`imaging-based biomarkers and their optimization in the
`context of clinical trials. Research software platforms are
`essential
`in prototyping, development and evaluation of
`novel algorithmic methods as a mechanism for discovering
`image-based surrogate end points. Such platforms should
`also support integration of the algorithmic advances into the
`trial work flows. In this paper, we discuss the
`clinical
`capabilities and the utility of 3D Slicer (Slicer) as an enabling
`research platform for quantitative image computing research.
`3D Slicer is a free open-source extensible software
`application for medical image computing and visualization.
`Slicer emerged as a culmination of several
`independent
`projects that focused separately on image visualization,
`surgical navigation and graphical user interface (GUI).
`David Gering presented the initial prototype of the Slicer
`software in his MIT Master's thesis in 1999 [7] based on the
`earlier experience of the research groups at MIT and Surgical
`Planning Lab (SPL)
`[8]. Subsequently, Steve Pieper
`assumed the role of the Chief Architect, commencing the
`work of transforming 3D Slicer into an industrial-strength
`package. Since 1999, Slicer has been under continuous
`development at the SPL under the leadership of Ron Kikinis.
`Today it is developed mostly by professional engineers in
`close collaboration with algorithm developers and applica-
`tion domain scientists, with the participation of Isomics Inc.,
`Kitware Inc. and GE Global Research and with significant
`contributions from the growing Slicer community. Initially
`envisioned as a neurosurgical guidance, visualization and
`analysis system [7,9], over the last decade, Slicer has
`evolved into an integrated platform that has been applied in a
`variety of clinical and preclinical research applications, as
`well as for the analysis of nonmedical images [10–21].
`
`Improvement and maintenance of the software have been
`possible primarily through the support from the National
`Institutes of Health (NIH). At the same time, its development
`has grown into a community effort, as numerous groups and
`individual users not funded directly to develop 3D Slicer are
`continuously improving it by reporting software problems
`and contributing solutions, suggesting new features and
`developing new tools. As described more fully below, Slicer
`integrates a number of powerful open-source projects into an
`end-user application suitable for clinical researchers.
`The breadth of functionality, extensibility, portability
`across platforms and nonrestrictive software license are
`some of the main features that differentiate Slicer from the
`commercial and open-source software tools and worksta-
`tions that aim to cover similar aspects of functionality.
`Numerous choices of radiology workstations and image
`analysis tools are available from the commercial vendors.
`Some of the popular tools used in the clinic as well as in
`research are AW Workstation (GE Healthcare), syngo.via
`(Siemens), PMOD (PMOD Technologies Ltd., Zurich,
`Switzerland), Definiens (Definiens Inc., Parsippany, NJ,
`USA), MimVista (MIM Software Inc., Cleveland, OH,
`USA). These packages provide users with a set of analysis
`tools (some of which may be specifically approved by the
`FDA for certain clinical
`tasks), compatibility with the
`clinical Picture Archiving and Communication System
`(PACS) systems and customer support. Such clinically
`oriented systems are not always affordable to the academic
`researchers. Commercial solutions are typically not exten-
`sible by the end user, are not oriented towards prototyping
`of the new tools and may require specialized hardware,
`limiting their applicability in projects that involve devel-
`opment of new image analysis methods. In the research
`domain, MATLAB (Mathworks, Natick, MA, USA) has
`traditionally been the “Swiss army knife” of scientific
`computing. Many researchers use MATLAB for initial
`prototyping and experimentation, while some end-user
`tools, such as SPM [22], are built on top of MATLAB.
`Being a generic prototyping tool, MATLAB is not
`designed for medical applications and thus lacks support
`for interface and display conventions common in clinical
`environments. As a result, deployment of the developed
`tools for the use by clinical researchers requires translation
`of the code into more generic languages to minimize
`dependencies and simplify integration.
`As opposed to the commercial workstations, 3D Slicer is
`meant to provide a research platform that is freely available
`and does not require specialized equipment. Slicer's use is
`not constrained to a single processing task or research
`application. Its generality and extensibility separate Slicer
`from such task-oriented packages as ITK-Snap (image
`segmentation) [23], DtiStudio (diffusion tensor analysis)
`[24], FreeSurfer, 1 FSL [25] and SPM [22] (neuroimaging
`applications). Several other tools, such as OsiriX [26],
`
`1 FreeSurfer, http://surfer.nmr.mgh.harvard.edu.
`
`Medivis Exhibit 1020
`
`2
`
`
`
`A. Fedorov et al. / Magnetic Resonance Imaging 30 (2012) 1323–1341
`
`1325
`
`BioImage Suite, 2 MIPAV [27] and ImageJ 3 [28], are similar
`to Slicer
`in that
`they provide extensible development
`platforms for biomedical
`imaging applications (for a
`comprehensive comparison of these tools, we refer the
`reader to the earlier surveys [29,30]). ImageJ is an extensible
`Java-based image processing platform that has been applied
`to a variety of applications, including radiological image
`processing [28], with the focus on two-dimensional (2D)
`analysis. MIPAV is a cross-platform Java-based package
`supported by NIH [27]. OsiriX is an open-source PACS
`workstation and DICOM viewer for Mac OS X that provides
`advanced capabilities such as image fusion, volume
`rendering and image annotation, and is extensible via a
`documented plug-in mechanism [26]. ClearCanvas is a
`Windows-based DICOM workstation adopted by the caBIG
`project. 4 The TCGA version of ClearCanvas workstation
`supports AIM model annotation capabilities [31,32] and is
`also extensible. A notable aspect of both OsiriX and
`ClearCanvas is that these systems are made available either
`in a free open-source version or as commercial, FDA-cleared
`products. A practical shortcoming is their dependency on
`specific operating systems (Mac OS X for OsiriX and MS
`Windows for ClearCanvas). Perhaps more importantly, some
`of the aforementioned packages that are similar in their
`intended purpose to 3D Slicer (including BioImage Suite,
`MIPAV, OsiriX and ClearCanvas) are distributed under
`restrictive open-source licenses that
`limit
`the ability of
`outside developers to redistribute parts of those systems, in
`particular, in commercial or other “closed-source” scenarios.
`This can be a practical constraint for QIN investigators
`collaborating with industry partners, as the solutions
`developed on top of these packages cannot be directly
`incorporated into commercial products. Another consider-
`ation is that critical functionality to work with modern
`imaging scenarios may only be available in the commercial
`version of
`the package. OsiriX and ClearCanvas,
`for
`example, do not support 64-bit processing architectures in
`their royalty-free versions, and this limits the maximum size
`of the image data the software can accept.
`3D Slicer does not have any components specific to a
`particular operating system. Binary distributions are avail-
`able for 32- or 64-bit versions of Windows, Mac OS X or
`Linux, and the software can be compiled on other systems,
`such as Oracle's Solaris. It is distributed under a BSD-style
`license agreement
`[33] allowing free distribution of
`derivative software for academic and commercial use.
`Hence, image analysis tools developed within 3D Slicer
`can be adopted directly by the industry collaborators. Since
`new technologies can only become part of routine clinical
`care through their incorporation into FDA-regulated medical
`products, Slicer's permissive software license furthers the
`
`2 BioImage Suite, http://www.bioimagesuite.org/.
`3 ImageJ, http://rsbweb.nih.gov/ij/.
`4 caBIG AIM ClearCanvas Workstation, https://cabig.nci.nih.gov/
`tools/AIM_ClearCanvas.
`
`overall goal of lowering the barriers for translation of the
`successful research solutions into medical products. On the
`other hand, Slicer is not an FDA-approved device, and its
`license makes no claims about the clinical applicability of the
`software. It is the sole responsibility of the user to comply
`with appropriate safety and ethics guidelines, and any
`products incorporating Slicer technology must be fully tested
`to comply with applicable laws and regulations. Under these
`considerations, Slicer has been applied in a variety of
`projects under appropriate research oversight.
`In this
`manuscript, we aim to introduce the capabilities of 3D
`Slicer as a software platform for clinical imaging research
`and outline its use in the context of biomarker development
`for cancer treatment by several QIN sites.
`In the remainder of this paper, we first present an
`overview of 3D Slicer by discussing its architecture, main
`features and guiding development principles. Next, we focus
`on the capabilities of 3D Slicer viewed from the perspective
`of the clinical researcher. We follow with the overview of the
`3D Slicer platform from the standpoint of a biomedical
`engineer and discuss how Slicer can facilitate development
`of new software tools for clinical research. To demonstrate
`how 3D Slicer is currently used by some of the existing
`teams of QIN, we discuss the clinical research projects
`investigated at Brigham and Women's Hospital (BWH) (PI
`Fiona Fennessy), University of Iowa (PI John Buatti) and
`Massachusetts General Hospital (MGH) (PI Bruce Rosen).
`We conclude with the summary of our findings, discussing
`some of the features and functionalities that would further
`improve applicability of 3D Slicer to biomarker development
`by the QIN investigators.
`
`2. Overview of 3D Slicer
`
`Computerized image analysis plays an increasing role in
`supporting clinical and research needs. Promising method-
`ologies that may lead to new imaging biomarkers often
`involve custom image processing software. The role of
`software evolves over the different stages of the imaging
`biomarker life cycle. In the inception stage, promising
`methodological concepts are identified and translated into
`early prototypes. Such early prototypes are often cobbled
`together from parts of tools designed for other tasks. They
`are typically suitable for use by engineers for small pilot or
`phantom studies. Their purpose is to demonstrate that the
`task can be done. The next step is to demonstrate that the task
`is worth doing. For this to happen, the method has to be
`optimized and thoroughly validated outside the research lab.
`This step requires a software tool that can be used reliably by
`trained clinical researchers and can be applied to a larger
`population of patients, possibly in a multisite clinical trial. At
`this point, the software must be robust and the interface
`intuitive. After establishing the value of a tool, the next step
`is to translate the software and the biomarker into a clinical,
`FDA-approved product.
`
`Medivis Exhibit 1020
`
`3
`
`
`
`1326
`
`A. Fedorov et al. / Magnetic Resonance Imaging 30 (2012) 1323–1341
`
`Fig. 1. 3D Slicer “ecosystem”. Slicer is a cross-platform package, but certain requirements on the graphics system, such as OpenGL drivers, should be met to
`support accelerated rendering. Its dependency libraries are portable across platforms and are distributed under compatible licenses. Those dependencies that are
`part of NA-MIC Kit often share the 3D Slicer developer community and are developed in synergy with the Slicer efforts. Slicer itself consists of the main
`application framework (core) and plug-ins (modules). Custom functionality is introduced by implementing external modules (Slicer extensions).
`
`One of the goals of 3D Slicer is to provide a common set
`of base functionality to facilitate development and validation
`of the medical image computing methods for the “can it be
`done?” and “is it worth doing?” steps of the imaging
`biomarker life cycle. Understanding of the imaging data by
`visualizing its different aspects, applying basic processing
`steps and displaying analysis results in context is critical at
`the initial stages. 3D Slicer is a powerful visualization tool
`that allows exploration of the imaging data sets in two, three
`and four dimensions. Slicer enables fusion of functional and
`anatomical data and provides a variety of generic and
`specialized tools for
`their processing and multimodal
`analysis. To ensure robustness,
`it
`is often required that
`image processing methods are initialized manually. The 3D
`Slicer framework includes the components needed to meet
`such development needs (e.g., “seed” the image segmenta-
`tion algorithm, provide initial pose for image registration or
`“steer” the processing based on the dynamically updated
`computation result).
`The use of an image analysis tool in a clinical research
`environment introduces new requirements. Support of the
`DICOM standard [34] for communicating image data is
`commonly required, as is a graphical user interface that is
`usable by a nontechnical operator. Slicer allows loading
`images in DICOM format from disk or directly from PACS.
`The Qt toolkit 5 provides a cross-platform GUI front end to
`Slicer that allows each processing module to easily define
`custom user interfaces. Image analysis tools implemented as
`Slicer processing modules can be developed in such a way
`that both interactive and batch execution is possible. This is
`particularly valuable in situations when experiments have to
`be performed on a large number of cases or when certain
`parameters of the algorithm have to be optimized.
`it can
`Once a method is optimized and validated,
`potentially be developed into an FDA-cleared clinical
`device. At
`this stage,
`it can be important
`that
`the
`implementation that was developed in the earlier stages
`
`5 Qt cross-platform application and UI framework, http://qt.nokia.com/.
`
`the key
`the need to reimplement
`can be used without
`components. Slicer adopts a licensing model that does not
`place any restrictions on the use of its source code in the
`derivative works. This aims to broaden the user community
`to include both academic and industry partners and to
`simplify the transitioning of
`the research tool
`into a
`commercial product.
`The architecture of 3D Slicer follows a modular and
`layered approach, as shown in Fig. 1. At the lower level of
`the architecture are the fundamental libraries provided by
`the operating system that are not packaged with Slicer, such
`as OpenGL and hardware drivers that allow efficient usage
`of
`the windowing and graphics resources of
`the host
`system. At the level above, there are languages (primarily
`C++ and Python, but increasingly JavaScript) and libraries
`that provide higher level functionality and abstractions.
`Some of the libraries used by the application are Qt (cross-
`platform GUI framework), the DICOM Toolkit (DCMTK) 6
`(implements parts of DICOM standard and is used to
`interact with DICOM data and DICOM services) and
`jqPlot 7 (provides charting capabilities). All of the external
`dependencies of 3D Slicer are cross-platform portable and
`are distributed under licenses fully compatible with Slicer,
`which do not restrict
`their use in either commercial or
`open-source products.
`Some of the libraries contributing to the foundation of
`3D Slicer are designed in close collaboration and often
`share the same developer community. These libraries are
`distributed as part of the National Alliance for Medical
`Image Computing (NA-MIC) Kit [35], a collection of the
`software tools supported in part by the NA-MIC project.
`The tools and interfaces provided by the NA-MIC Kit
`components are largely focused on the needs of
`the
`developers of medical
`image computing applications.
`CMake 8 enables cross-platform build system configuration,
`
`6 DCMTK — DICOM Toolkit, http://dicom.offis.de/dcmtk.
`7 jqPlot, http://www.jqplot.com.
`8 CMake: cross-platform build system, http://cmake.org.
`
`Medivis Exhibit 1020
`
`4
`
`
`
`A. Fedorov et al. / Magnetic Resonance Imaging 30 (2012) 1323–1341
`
`1327
`
`packaging and testing of 3D Slicer and NA-MIC Kit
`libraries. CDash 9 is a Web-based server that organizes the
`results of software testing. The Visualization Toolkit
`(VTK) 10 provides the key building blocks for 3D computer
`graphics and visualization. The Insight Toolkit (ITK) 11 [36]
`is a library developed specifically for the tasks related to
`medical
`image registration and segmentation and for
`implementing new image analysis algorithms. The Common
`Toolkit (CTK) 12 is a biomedical image computing library
`with a focus on application-level DICOM support, plug-in
`framework and specialized GUI widgets.
`3D Slicer itself consists of the lean application core, Slicer
`modules and Slicer extensions. The core implements the
`Slicer user interface and provides support for data input/
`output (IO), visualization and developer interfaces that
`support extension of the application with new plug-ins.
`Internally, Slicer uses a scene data structure to organize
`images and annotations, coordinate layouts and maintain the
`application state. The XML-based Medical Reality Markup
`Language (MRML) is used to serialize the content of the
`scene. Slicer modules are the plug-ins that depend on the
`Slicer core to implement new functionality. Individual
`modules can be independent or can rely on other modules
`(for example, a module that provides lesion segmentation
`functionality can depend on the volume rendering module to
`enable 3D visualization of the anatomy and segmented
`structure). Unlike Slicer modules that are packaged as part of
`Slicer distribution, Slicer extensions are external plug-ins
`installed “on demand” by the user, akin to the Web browser
`extensions. The extensions mechanism enables sharing of
`the Slicer-based tools that cannot be included into the
`package due to incompatible licenses, development timelines
`or other constraints. It also provides a “pathway” for
`integrating new functionality since any extension that
`satisfies specific requirements is a candidate to be included
`into the Slicer distribution. The requirements for integration
`to the distribution include the following: a designated
`maintainer to serve as a point of contact, a nonrestrictive
`Slicer-compliant
`license, adherence to the Slicer coding
`standards, availability of software tests and appropriate user-
`level documentation. We emphasize that tools that do not
`satisfy these requirements can still be available to Slicer
`users, clearly marked as Slicer extensions and accompanied
`by an appropriate disclaimer.
`Since its inception in late 1990s, 3D Slicer has been
`evolving, with major architectural, functional and GUI
`redesigns occurring every 4–5 years. The current (fourth)
`generation of Slicer was released in November 2011. The
`most notable improvements of the software as compared to
`the previous (third) version are improved visualization
`performance, reengineered DICOM support, completely
`
`9 CDash: software Web-based testing server, http://cdash.org.
`10 VTK: Visualization Toolkit, http://vtk.org.
`11 ITK: Insight Toolkit, http://itk.org.
`12 CTK: The Common Toolkit, http://commontk.org.
`
`redesigned GUI and availability of the Python development
`interfaces. With each major redesign, several policies are
`followed to ease the transitioning of users and developers to
`the updated platform. The base and core modules are
`released to the broader development community first,
`followed by the migration of the specialized modules. For
`example,
`the migration strategy for the modules in the
`image-guided therapy (IGT) category is the most conserva-
`tive due to the mission-critical nature of these applications.
`As such, the modules in this category are usually migrated
`after the base of the new release is stable and thoroughly
`tested. Another policy has been to provide backward
`compatibility to support reading of data produced using the
`earlier version.
`Within each generation, new and updated releases of
`Slicer are prepared every 2 to 6 months. These releases
`include performance improvements, bug fixes and new
`functionality, but no major changes to the base architecture
`or GUI. A release includes a tagged version of the source
`code in the source code repository as well as binary
`installation packages for the supported platforms (Windows,
`Mac OS X, Linux; 32-bit, 64-bit). Daily binary installation
`packages are also prepared for the major platforms to track
`the current version of the source code. These daily packages
`allow Slicer users to access the new functionality under
`development. The Slicer software development process is
`collaborative and geographically distributed, with over 80
`authorized developers contributing source code through
`SVN and Git revision control systems. To ensure the stability
`of the software, the source code is compiled and tested on a
`daily basis on a variety of platform configurations. The
`testing results are summarized and reported using a Web-
`based centralized CDash dashboard. Users and developers of
`3D Slicer can also report issues on the open mailing lists or
`using Web-based bug tracking system.
`Documentation,
`training and user support are high
`priorities for
`the Slicer community. Hands-on training
`sessions are organized regularly as part of ongoing
`outreach initiatives at major conferences, such as the
`annual Radiological Society of North America (RSNA),
`Medical
`Image Computing and Computer Assisted In-
`terventions (MICCAI) and International Society for Optics
`and Photonics (SPIE) meetings, or in response to request
`by host
`institutions at both national and international
`venues. Semiannual hands-on week-long project weeks are
`open for participation to anyone interested in developing
`or using Slicer tools, and are an ideal place to exchange
`ideas and experience. Remote learning of 3D Slicer is
`supported by the online resources and community mailing
`lists. Focused training materials that include sample data
`sets and step-by-step instructions are available for basic
`Slicer operation as well as advanced workflows. Wiki-
`based Web pages accompany every module of the 3D
`Slicer, providing reference documentation of functionality
`and usage examples. User and developer community
`mailing lists have been active for the last 10 years and
`
`Medivis Exhibit 1020
`
`5
`
`
`
`1328
`
`A. Fedorov et al. / Magnetic Resonance Imaging 30 (2012) 1323–1341
`
`currently include over 770 and 440 subscribers, respec-
`tively, providing a source of Slicer expertise extending
`well beyond the community of the core Slicer developers.
`Around 2800 messages were exchanged on these Slicer
`mailing lists in 2011.
`There are two major communities of Slicer users. First,
`there are medically trained researchers that apply the tools
`available in the 3D Slicer to novel clinical applications.
`Second,
`there are biomedical engineers and computer
`scientists that use 3D Slicer as a platform for development
`and delivery of their algorithmic work. In the next two
`sections, we discuss the main capabilities provided by 3D
`Slicer from the standpoint of scientists who are members of
`those two communities.
`
`3. Clinical research platform
`
`From the perspective of a clinical researcher, Slicer is an
`advanced image visualization and analysis workstation, and
`it shares a lot of generic functionality with the commercially
`available packages. Unlike most of
`the commercially
`available workstations, 3D Slicer is not an FDA-cleared
`product, and its intended use is clinical research applications.
`In this sense, Slicer is very different from its commercial
`counterparts since it incorporates experimental tools that
`cannot be packaged within the workstations used clinically.
`3D Slicer is available in the form of binary packages that
`have platform-specific installers for all major operating
`systems. These packages are self-contained, which means
`they include all the dependency libraries and toolkits needed
`to use Slicer on a given platform. This also makes it possible
`to run two or more versions of Slicer at the same time since
`the packages do not share any components. Slicer includes
`core functionality that covers common needs for a variety of
`applications as well as task-specific modules. The main
`application GUI provides a consistent
`interface for the
`software, simplifying the learning process since all modules
`follow the same conventions and share the same set of GUI
`controls. The two major components of Slicer GUI are the
`layout-controlled data viewers and the module GUI panel, as
`shown in Fig. 2.
`
`3.1. Visualization elements and capabilities
`
`3D Slicer visualization capabilities support various
`imaging modalities [e.g., computed tomography (CT),
`positron emission tomography (PET), magnetic resonance
`imaging (MRI) and ultrasound] and can be used for
`visualization of 2D, 3D and four-dimensional (4D) data
`sets. Support of 3D data sets (MRI, CT, PET) has enjoyed
`most attention based on the substantial number of use cases
`resulting in a large number of tools developed specifically
`for this kind of data. Although support of the newer and less
`commonly applied 2D and 4D modalities is possible, it is a
`fertile ground for the future development. Data visualization
`in Slicer is enabled by the 2D and 3D viewers. The 2D
`
`viewers allow visualization of cross-sections from 3D or 4D
`image volumes and support the associated image operations
`such as basic image manipulations (zoom, window/level,
`pan), multiplanar reformat, crosshairs and synchronous pan/
`scroll for the linked viewers. Advanced capabilities of 2D
`viewers enable display of various types of glyphs for
`visualization of tensors and vector fields. Lookup tables
`(LUTs) support the mapping between the image values and
`the colors displayed in 2D viewers (e.g., grayscale LUT is
`typical for CT images, but color can be preferred for PET).
`Each 2D viewer supports three image “layers”: the user can
`select “background” and “foreground” image volumes as
`well as a “label” image volume (which can represent
`segmentations) and fuse the three layers by adjusting
`transparency. 2D viewers also support a “Lightbox” mode,
`where multiple slices from a volume are tiled within the
`viewer. The 3D viewers enable visualization of volume data,
`such as triangulated surface models, fiber tracks, glyphs and
`volume renderings. Augmented reality applications can
`utilize stereoscopic capabilities of the 3D viewers. The 2D
`and 3D elements can be combined in the 3D viewers to
`provide an integrated visualization of the various data to the
`end user, as shown in various figures throughout
`this
`manuscript. Coherent
`integration and visualization of
`multiple images and data types in 2D and 3D viewers are
`achieved via a common, scanner-independent, patient-
`centered physical space coordinate frame reference system.
`Spatial alignment of individual volumes can be achieved by
`introducing hierarchies of rigid and deformable spatial
`transformations. These transformations can be defined
`manually by the operator or using the automated registration
`tools. Finally, chart viewers, a relatively new feature under
`development, are available for displaying plots.
`A variety of image markup and annotations are supported.
`Following the convention used by Rubin et al. [31], we use
`the term image markup to refer to the graphical elements
`overlay, and image annotation for the text-based informa-
`tion, both of which describe a certain finding in a given
`image. The markup elements provided by the 3D Slicer (as
`of writing this) are fiducials (points), unidimensional
`measuremen