DICOM for quantitative imaging biomarker development: a standards based approach to sharing clinical data and structured PET/CT analysis results in head and neck cancer research

Patient cohort selection

The primary data was extracted from HNC patients with squamous cell carcinoma, all treated according to the standard of care at the University of Iowa Hospital and Clinics. Clinical practice was to obtain a FDG PET/CT for staging (prior to treatment) and then a second FDG PET/CT scan for response assessment at approximately three months following the completion of the initial therapy. All patient data was collected in compliance with HIPAA regulations under approval granted by the internal review board of the University of Iowa, approval #200503706. Written consent was obtained from the study participants. The imaging studies were acquired between 2004 and 2013. Patients that had a baseline and at least one post-therapy follow-up PET/CT were included in the research study. Patients were followed clinically and outcomes were available with a minimum of 2 years of follow-up. Patients may have had additional imaging studies following the three month response assessment FDG PET/CT based on clinical judgment and findings.

Clinical metadata for each patient was manually extracted from the electronic health records and included sex, age, smoking, and drinking history as well as pathology, stage, primary site location, and detailed location of involved nodal sites. Treatment details (e.g., radiation dose, technique, surgical intervention and chemotherapy delivery) and disease status and recurrences were recorded. All clinical metadata was de-identified and stored in a Postgres relational database locally. Measurements made on images that were used for clinical purposes and stored in the clinical records were not used during the conversion process, since new measurements were to be made, and homogeneity and accuracy of the clinical measurements could not be easily verified.

A total of 156 patients were identified as eligible for the study, with at least one PET/CT scan and related clinical data available for study (mean 3.05 studies/patient collected during a total of 472 visits). Fifty-nine patients from the cohort were processed using the methodology described in the following text. In one of those 59 patients both pre- and post-treatment imaging studies were processed, while in the rest of the patients only the baseline scan was analyzed.

Image acquisition

Pertinent details related to image acquisition such as reconstruction procedure, image resolution and injected dose were encoded in the DICOM image metadata by the scanner. After initial de-identification, the image data was stored in an eXtensible Neuroimaging Archive Toolkit (XNAT) (Marcus et al., 2007) local research archive at the University of Iowa.

Image processing

SUV is commonly utilized for a simple semi-quantitative analysis of PET images (Lucignani, Paganelli & Bombardieri, 2004). SUV Body Weight (SUVbw) is defined as the ratio of activity in tissue divided by the decay-corrected activity injected to the patient, normalized by body weight: SUVbw = (tissue activity)/(injected activity/weight). Several alternatives to SUWbw approach have been investigated including body surface area corrected (SUVbsa) and lean body mass corrected (SUVlbm) (Graham, Peterson & Hayward, 2000), but SUVbw remains the most commonly utilized quantity.

There are several underlying assumptions made in using FDG SUVs for measuring metabolic activity in lesions, such as accurate measurement of injected dose and accurate decay correction of all measurements (Graham, Peterson & Hayward, 2000). The failure of one or more of these assumptions can introduce variability in calculated SUVs. To mitigate this problem, an SUV ratio (SUVr) can be used, which represents the ratio of the SUV of a lesion to the SUV of a normal tissue Reference Region (RR) defined in the same acquisition.

In the project generating the data presented here, the primary cancer site and all involved lymph nodes were segmented separately to allow quantification of SUV for either the primary cancer site alone, total tumor burden, or on a per-region basis. Segmentation of the primary tumor and lymph nodes was done using two interactive segmentation tools within 3D Slicer (Fedorov et al., 2012). The first tool is a manual contouring tool, requiring the user to draw the boundary of a lesion on every slice using the Editor module of 3D Slicer. The second tool is semi-automated, performing segmentation in 3D using a specialized algorithm for segmenting HNC in FDG PET images, which is described and evaluated in detail in Beichel et al. (2016). This semi-automated segmentation approach treats the segmentation task as a graph-based optimization problem based on the ideas introduced by Yin et al. (2010). Starting with a user-provided approximate lesion center point, a graph structure is constructed in a local neighborhood, and a suitable cost function is derived based on local image statistics. A maximum flow algorithm is used for optimization. The resulting segmentation is converted from a graph-based representation to a labeled volume. To handle frequently occurring situations that are ambiguous, several segmentation modes are introduced to adapt the behavior of the base algorithm accordingly. In addition, “just enough interaction” based approaches are provided to enable the user to efficiently perform local and/or global refinement of initial segmentations. This semi-automated segmentation method is implemented in the PET Tumor Segmentation extension of 3D Slicer (QIICR, 2015a).

Since both manual and semi-automatic methods for HNC segmentation depend on user input, results are expected to be subject to intra- and inter-operator variation. To allow assessment of the impact of such variation on subsequent processing steps, each data set was reviewed and segmented using both methods by three readers, who were experts in HNC PET/CT image interpretation. Images were presented to the readers in random order. For each combination of the segmentation tool and reader, this process was performed twice, resulting in twelve segmentation sessions per patient. RRs in liver, cerebellum, and aortic arch were segmented automatically using the approach we presented earlier (Bauer et al., 2012).

Given the segmentations of the primary tumor and lymph nodes, a total of 22 quantitative indices were extracted from each of these regions using the PET-IndiC extension of 3D Slicer (QIICR, 2015b). The calculated quantitative indices consist of commonly utilized PET-specific indices such as maximum, mean and peak SUV (Wahl et al., 2009) and Total Lesion Glycolysis (TLG), as well as common summary statistics, which included median, variance and Root Mean Square (RMS) of SUV, and segmentation volume. Mean, maximum, minimum, standard deviation, median, and first and third quartiles were calculated for RRs.

Data modeling and conversion into DICOM representation

The DICOM standard provides a variety of objects that can be used to communicate information derived from the images. Regardless of the specific object type, DICOM requires that all objects contain so-called “composite context.” At the patient level, the composite context includes identifying and descriptive attributes such as patient name, ID, age and sex. The study context includes the date and time that the imaging study started, unique identification of the study and other information common to all series in the study. The composite context enables consistent indexing and cross-referencing of the various objects. In addition to shared composite context, derived DICOM objects typically contain explicit references to the “source” objects from which they were derived, which supports recording of the provenance of the object derivation as well as application functionality such as superimposition during rendering. Various relationships between the objects used in this study are shown in Fig. 2.

In the following sections we discuss the motivation for the choice of specific DICOM objects. We start with the PET and CT objects, since they were produced by the acquisition equipment and underwent only minor editing for de-identification. Next we describe the objects containing patient clinical data (clinical history and outcomes). Then we cover the derived imaging objects from simple to more complex:

• DICOM Real-World Value Mapping (RWVM) objects encode mapping of the image-specific SUV factor that is needed for normalization of the images and subsequent processing;

• DICOM Segmentation (SEG) objects encode labeling of the PET and CT image voxels into anatomical structures, such as primary tumor and liver ROI;

• DICOM Structured Reporting (SR) objects encode various measurements computed from the segmentation-defined regions on the normalized PET image volumes.

We follow the general pattern of discussing the scope and capabilities of the object at a high level, followed by an abbreviated summary of the design decisions made to meet the requirements of our use case. The reader is referred to the preprint version of this article (Fedorov et al., 2015a) for further discussions, which has been omitted for brevity. Corrections to DICOM that resulted directly from our experience are also listed. A separate section covers the implementation details of converting research representations into the DICOM format, and references the tools we developed for this purpose.

PET/CT image data

PET and CT image data were stored in the DICOM Positron Emission Tomography Image (National Electrical Manufacturers Association (NEMA), 2016b) and Computed Tomography Image (National Electrical Manufacturers Association (NEMA), 2016c), respectively. The image data obtained from the scanner was de-identified using a modified version of the Basic Attribute Confidentiality Profile defined by the DICOM in PS3.15 Appendix E.2 (National Electrical Manufacturers Association (NEMA), 2016d). Image de-identification was performed following the standard operational procedures established by The Cancer Imaging Archive (TCIA) (Clark et al., 2013; Moore et al., 2015). Research identifiers of the form QIN-HEADNECK-01-nnnn were assigned in place of the patient names and medical record numbers. Dates were shifted by the same fixed offset across all the datasets to maintain temporal relationships of the datasets. The de-identified images were then used for the remainder of the project (i.e., to make the measurements and convert them into derived DICOM objects), in order to mitigate the risk of leakage of patient identifiers into the publicly accessible analysis results.

Clinical information: DICOM SR

Relevant clinical information available for the subjects enrolled in the study included clinical history (such as the diagnosis and pathology, surgery and radiotherapy administration, and demographics) and outcomes (follow-up date and status, and the date of death, when applicable). This information is important for the interpretation and secondary reuse of the image and quantitative data set, since it contains the clinically relevant end-points for the evaluation of the biomarker performance, and it provides non-imaging predictors that can be used for machine learning. The clinical information was extracted from the operational Postgres research database, and retrospectively encoded in DICOM SR, one SR object per patient. The choice of DICOM SR for encoding clinical information is explained in Appendix S1.

DICOM SR objects (sometimes referred to as SR “documents”) contain information organized as a hierarchical content “tree” consisting of content items (tree nodes) (Clunie, 2000). These content items include containers, textual information, codes describing concept names (we will use “term” and “concept” interchangeably in this document) and values (where appropriate), references to images, and numeric values (National Electrical Manufacturers Association (NEMA), 2016e). DICOM SR templates define a pattern of content items and their relationships, constraining the general infrastructure for specific use cases (National Electrical Manufacturers Association (NEMA), 2016f). Each SR template is assigned a Template Identifier (TID). Templates may define the entire content of an object (i.e., be a “root” template) or may be a reusable common pattern of nested content to be included by higher level templates (i.e., be a “subordinate” template).

Each content item, except for those that are containers, can be thought of as a “name-value pair” (or alternatively, as a “question” and an “answer”). Containers can be considered “section headings,” and are often explicitly used as such when rendered in human-readable form. The top level (root node) of the content tree is always a container, and its name (concept) is often referred to as the “document title.” The concept name of a container or name-value pair (mandatory in most cases) is always coded using a code from a controlled terminology. The value may or may not be coded depending on the value type.

The use of controlled terminology is fundamental to DICOM SR. DICOM SR codes are defined as triplets of code value, coding scheme designator and code meaning (e.g., (F-02573, SRT, “Alcohol consumption”), where “SRT” is the DICOM designation for the SNOMED coding scheme). While DICOM allows for reuse of the codes defined in other terminologies, such as SNOMED, as well as those defined in the DICOM standard itself, so called “private” codes can also be defined by the creator of the object, when no standard codes are available. Such private codes are distinguished by a coding scheme designator that starts with a “99” prefix. The use of predefined codes not only provides semantic information, but also simplifies validation of the resulting objects. The codes that are allowed are constrained by the template. The constraints for values may be defined in the template itself, or in a “value set,” which in DICOM is called a Context Group (and labeled with a Context Group ID (CID)).

Though DICOM contains templates for clinical data for a few specific applications (e.g., cardiovascular (National Electrical Manufacturers Association (NEMA), 2016g) and breast (National Electrical Manufacturers Association (NEMA), 2016h)), it does not define a template to represent all the clinical data items of interest in our HNC QI research use case. Given the lack of a suitable standard template to represent this data, we developed our own set of custom templates for communicating the clinical information. In DICOM, such custom templates are referred to as “private templates,” even though they may be publicly shared and are required to be documented in the DICOM conformance statement of the product. These templates included information about biopsy, treatment and other relevant data. The relationships between the private templates are shown in Fig. 3, with a detailed description provided in Appendix S2. These templates follow the patterns of existing DICOM templates, with the intent that they might form the basis for future enhancements of the standard.

No structured terminology was used at the time of initial clinical data collection, so terms with codes were selected retrospectively at the time of conversion of the data to DICOM SR. Our approach for selecting codes leveraged SNOMED (Cornet & De Keizer, 2008) and UMLS (Bodenreider, 2004) terminology as much as possible. The few concepts that could not be located in the SNOMED, UMLS or DICOM terminologies were added to a private coding scheme. All of the codes that are of relevance to this project are listed in Appendix S2.

Standardized uptake value: DICOM RWVM

The DICOM Real World Value Mapping (RWVM) object provides a mechanism to describe the calculation that was used (and can be reused) to create “real world values” (such as SUV) from stored pixel data values. A RWVM can be embedded within another DICOM object (such as an acquired or derived image), or it can be encoded as a standalone object (National Electrical Manufacturers Association (NEMA), 2016i), which in turn can either be referenced from other objects, such as SRs, or recognized as being relevant from the commonality of patient and study identifiers.

We chose to create a standalone RWVM object to encode SUVbw factor and leave the original (de-identified) activity-concentration images unchanged. The RWVM object encodes the scale factor, the range of stored pixel values to which it applies, and standard codes that specify the quantity that the scaled (real world) value represents (in this case, the SUV), the measurement method (the SUV body weight calculation method) and the measurement units (g/ml{SUVbw}). The DCM coding scheme is used for the quantity and the measurement method, and, as is the case throughout DICOM, the Unified Code for Units of Measure (UCUM) system (Schadow et al., 1999) is used for the units. The RWVM object also includes references to all of the PET image objects to which it applies.

The following corrections to the standard were proposed to remedy the errors or limitations of the standard identified while developing DICOM representation of the SUVbw factors for this project:

1. CP 1387⁴ :addition of quantity descriptors to Real World Value Maps (applies to the 2014b version of the standard). The original definition of the RWVM in DICOM only defined the encoding of measurement units. We proposed an improvement to the standard to include the definition of quantity in the RWVM encoding.

2. CP 1392: addition of quantity descriptors and measurements for PET (applies to 2014b). This CP added new concepts related to encoding of the PET measurements that were missing in the standard, but were required by our use case.

Image segmentation: DICOM SEG

The imaging time point was defined as an ordinal number corresponding to the imaging study performed for the patient in the course of management of the specific condition, with time point 1 corresponding to the baseline/staging study. For each such time point we encoded segmentations prepared using image processing steps discussed earlier.

DICOM provides different mechanisms for encoding ROIs obtained by segmentation, as discussed in Fedorov et al. (2015a). The choice of the most suitable mechanism depends on the use case. Since the native representation of the segmentation results were labelled individual voxels, rather than a surface mesh or isocontours, we selected the DICOM Segmentation image (SEG) object as the most appropriate for encoding the ROIs.

The SEG objects were organized as follows, to be consistent with the pattern that would likely be used by tools that created them prospectively rather than retrospectively:

• Each of the RRs is stored as a separate object, since each of the RRs was segmented using a distinct automatic method, using data from different modalities (the aortic arch was segmented on the CT images, and the cerebellum and liver ROI were segmented on the PET images).

• The primary tumor and involved lymph nodes segmented for each combination of operator/segmentation method/session were stored together as different segments in a single object, since both the tumor and nodes were segmented during the same session, with the segmentation of one structure being identified while considering the neighboring structures.

• The identifier of the operator (reader) for the manual and semi-automated segmentation results was stored in the ContentCreatorName⁵ attribute.

• The identifier of the imaging time point was encoded as a positive integer, stored in the ClinicalTrialTimePointID attribute.

• The identifier of the segmentation session for primary tumor and lymph nodes was encoded in the ClinicalTrialSeriesID attribute.

• The type of algorithm used was encoded in the SegmentAlgorithmType attribute as MANUAL, SEMIAUTOMATIC or AUTOMATIC, as appropriate.

• The suggested color for each of the segmented structures was encoded in the RecommendedDisplayCIELabValue attribute.

The semantics of the segments were communicated using the standard AnatomicRegion (and its modifier in AnatomicRegionModifier sequence, when necessary), SegmentedPropertyType and SegmentedPropertyCategory sequences. For example, the semantics of a primary tumor was encoded as follows:

 Segmented Property Category =
   (M-01000, SRT, "Morphologically Altered Structure")
 Segmented Property Type = (M-80003, SRT, "Neoplasm, Primary")
 Anatomic Region = (T-53131, SRT, "base of tongue")

DICOM defines a relatively small set of segmentation property categories, listed in CID 7150 (National Electrical Manufacturers Association (NEMA), 2016j), and a considerably larger set of segmentation property types in CID 7151 (National Electrical Manufacturers Association (NEMA), 2016k). There is no direct relationship specified in the standard between category and type, and the choice of an appropriate category is left to the discretion of the implementer (arguably the standard could be improved by grouping the types and assigning them to, and requiring them for, specific categories).

Sometimes segmentations are performed for purely anatomical reasons (e.g., for anatomical atlases), in which case there is no meaningful additional property type to record. In such cases, the anatomy is encoded directly in SegmentedPropertyType, without the need for a separate AnatomicRegionSequence. In other cases, segmentations are performed that apply to anatomical structures, but which segment them into different types of tissue. In these cases, the SegmentedPropertyType is used to encode the type of tissue (e.g., primary tumor, secondary tumor, necrosis) and the AnatomicRegionSequence can be used to encode the anatomic location (e.g., which organ, group of lymph nodes, etc.). Sometimes the anatomy is irrelevant and not encoded at all, and the SegmentedPropertyType just encodes the type of material segmented. This distinction was clarified by the authors in an earlier DICOM correction proposal CP 1258. In this project we are encoding both the nature (category and type) of the segmented area and its anatomic location.

Lymph nodes are encoded similarly, but with only the general region (head and neck) recorded rather than a precise code for the lymph node group, because of the lack of the detailed information about the specific lymph node name in the original dataset due to practical difficulties in assigning such a precise name when segmentation was performed:

 Segmented Property Category =
   (M-01000, SRT, "Morphologically Altered Structure")
 Segmented Property Type = (M-80006, SRT, "Neoplasm, Secondary")
 Anatomic Region = (T-C4004, SRT, "lymph node of head and neck")

Semantics of the RR segmentations are communicated using the “spatial relationship concept” category:

 Segmented Property Category =
   (R-42018, SRT, "Spatial and Relational Concept")
 Segmented Property Type = (C94970, NCIt, "Reference Region")
 Anatomic Region = (T-62000, SRT, "Liver")

Binary segmentations are encoded in the PixelData attribute of the SEG object, and are represented as a contiguous array of bits, with one bit per voxel for each frame. There are separate frames for each slice of the volume, though all are encoded in a single multi-frame object. When multiple segments (i.e., primary tumor and lymph nodes) are produced by the operator during a single session using a single segmentation tool, they are stored in a single SEG object, with each segment for each slice stored in a separate frame. Empty frames that do not contain any voxels of the segmentation are elided, to reduce the size of the encoded objects. The matrix size (rows and columns) is not abbreviated to a rectangular bounding box enclosing the region of interest, which would be a further possible object size optimization (i.e., each frame has the dimensions of the original image). Similar to the RWVM objects, SEG objects include references to the SOP Instance UIDs of the images (slices) that were segmented.

The process of encoding the segmented ROIs in DICOM led to the development of the following correction proposals:

1. CP 1406: add codes for tumor sites (applies to 2014c). The uncoded (plain text) labels of all the tumor regions used in this project were analyzed to identify common terms that were then mapped to SNOMED concepts. The resulting terms were introduced into the DICOM standard in the form of new context groups for lymph nodes (CID 7600) and HNC anatomic sites (CID 7601). A distinction between concepts for primary and secondary neoplasms was introduced in the same proposal.

2. CP 1426: correct condition in pixel measures, plane position and orientation functional groups for segmentation (applies to 2015a). Prior to this correction, the presence of the essential attributes that are needed for volumetric reconstruction of the segmentation image volumes was conditioned on attributes that were optional or not defined in segmentation objects.

3. CP 1464: add reference region segmentation property type (applies to 2015c). This correction added the codes needed to describe RRs, using the NCI Thesaurus terminology.

4. CP 1496: add tracking identifier and UID to segmentation instances (applies to 2015c). Use of a common Tracking UID allows to establish correspondence between segments encoded in various segmentation objects that represent the same region being segmented (i.e., across different time points, modalities, operators). Tracking UIDs were already present in the SR measurements objects, which can reference segmentation objects, but were not encoded directly in the segmentation objects themselves.

Quantitative measurements: DICOM SR

To encode the PET SUV ROI measurements in DICOM, we specified the terminology that defines the measurement quantities, modifiers and units for each measurement of interest needed. The vocabulary required was not specified in any single standard context group. Concepts from various standard context groups were therefore leveraged as appropriate. The strategy to find a suitable term was to first consult those already in DICOM, then search for related concepts in UMLS, SNOMED, and the NCI Thesaurus. If no existing concept could be found, we introduced a new code and definition in a private 99PMP coding scheme, while referencing a relevant publication, if available. All of the terms used are described in Appendix S3. The reader is referred to Fedorov et al., (2015a) for additional discussion of the selection of the quantity codes.

The measurements were encoded as DICOM SR objects using the standard root template TID 1500 defined in PS3.16 (National Electrical Manufacturers Association (NEMA), 2016l), which makes use of the sub-ordinate templates shown in Fig. 4. TID 1500 contains a preamble that describes general characteristics relevant to the measurement, such as an Image Library container (National Electrical Manufacturers Association (NEMA), 2016m), which lists the UIDs of the images in the original image series, radiopharmaceutical agent, and other items related to the acquisition protocol that may be relevant during interpretation. The Imaging Measurements container (section heading) includes the following attributes, which have special meaning in the context of our use case:

• Activity session: a positive integer that encoded the segmentation session by the operator.

• Tracking identifier: a human-readable identifier of the finding, which is not required to be unique. In our project, RRs had tracking identifiers coded as “referenceRegionName reference region,” where referenceRegionName was one of “liver,” “cerebellum” or “aortic arch.” The primary tumor identifier was always set to “primary tumor,” individual lymph nodes were identified as “lymph node nodeID,” where nodeID is a positive integer. As mentioned earlier, lymph nodes were not tracked (i.e., their nodeID did not identify the specific lymph node across time points or reading sessions).

• Tracking unique identifier: a DICOM standard UUID-derived (random) identifier with a “2.25.” prefix (National Electrical Manufacturers Association (NEMA), 2016n): a primary lesion unique identifier that was used to track the lesion and reference regions across time points.

• Time point: a positive integer that encoded the temporal order of the imaging study within the course of management of the given patient.

• Referenced segment and source series for image segmentation: the identifiers of the segment and the segmentation object representing the ROI used in the measurement group, and the identifier of the series that was segmented.

• Finding site: the coded anatomical location of the finding.

Related groups of measurements were encoded as a list, preceded by the codes of one or more findings, following the structure defined by TID 1411 Volumetric ROI Measurements (National Electrical Manufacturers Association (NEMA), 2016o), which in turn invokes TID 1419 ROI Measurements (National Electrical Manufacturers Association (NEMA), 2016p), as summarized in Fig. 4. Each group of measurements was derived from the ROIs that applied to the voxels of a single reconstruction of a PET acquisition (image series). One SR measurement object was created for each SEG object. Voxels in the ROI used for the derivation of the measurements were encoded as one segment of a SEG object. Both the SEG image objects and the segment number used by the derivation were referenced for each measurement group in the SR. An example of the structure of the Imaging Measurements is presented in Appendix S4.

The following DICOM standard corrections were contributed while developing the conversion methodology:

1. CP 1366: correction of relationships in planar and volumetric ROI templates (applies to 2014b). In the process of data encoding, we identified errors in the definition of the relationships in some templates.

2. CP 1386: addition of measurement report root template for planar and volumetric ROIs (applies to 2014b). Before the introduction of this root template, measurement templates could only be used to construct subordinate objects included in other templates, but not to encode standalone measurement objects. This CP also added some of the codes needed for this project, and allowed common content items to be factored out of individual measurements to the group level.

3. CP 1388: add real world value map reference to measurements (applies to 2014b). This CP added an explicit reference to the RWVM instance that was used to calculate the measurements to the measurements SR object template.

4. CP 1389: factor common descriptions out of image library entries (applies to 2014b). We introduced simplifications to the structure of the measurements SR object by allowing a group of images to share common image library attributes, greatly reducing the size and improving the readability of the object in cases when measurements were derived from many single frame images.

5. CP 1465: add type of finding to measurement SR templates (applies to 2015c). The measurement template was amended to include the type of finding, which is distinct from its anatomical location.

6. CP 1466: add session to measurements group (applies to 2015c). An extra item was added to the measurement template to enable encoding of the session identifier to support experiments where the measurement of the same finding is performed several times in order to evaluate its repeatability.

Clinical information: DICOM SR

Clinical data was exported from the internal SQL database as a tab-delimited text file. An XSLT script was used to convert the tab-delimited representation into XML form, followed by another XSLT transformation that produced an XML representation of an object that follows DICOM SR template TID QIICR_2000 documented in Appendix S2. Finally, the resulting XML representation was converted into DICOM format using existing functionality of the PixelMed toolkit (Clunie, 2015a). The conversion scripts are available in a public source code repository (QIICR, 2015d). The DICOM series containing the clinical data DICOM SR were assigned to a study separate from the one for the imaging and derived data, with both the StudyDescription and SeriesDescription attribute set to “Clinical Data.”

Standardized uptake value: DICOM RWVM

RWVM objects were generated in batch mode using the SUV calculation plugin of 3D Slicer (QIICR, 2015e). The plugin operated on the list of files corresponding to the PET series DICOM objects, calculated SUVbw factor and produced a single RWVM object. Injected dose, patient weight, radionuclide half-life and injection time were obtained from the DICOM PET image header.

Image segmentation: DICOM SEG

The process of converting segmentation results into DICOM representation was facilitated by the FOSS DICOM software library implementation available in DCMTK (DICOM Toolkit) and maintained by OFFIS in Germany (Eichelberg et al., 2004). To simplify the task of creating SEG objects for this project and other similar efforts, we extended DCMTK with three new libraries, which are now included in the official distribution of DCMTK: dcmiod, dcmfg and dcmseg (Fedorov et al., 2015a). The conversion was performed using batch mode tools SEG2NRRD (conversion from DICOM SEG to NRRD research format) and EncodeSEG (conversion from research segmentation format to DICOM SEG). These tools are included in the Iowa2DICOM repository referenced above.

Volumetric measurements: DICOM SR

The process of calculation and encoding of the ROI measurements was implemented in 2 steps. First, measurements of interest were calculated in batch mode using the QuantitativeIndicesCLI tool available within PET-IndiC extension of 3D Slicer (QIICR, 2015b). The tool accepted the SUV-normalized image volume and the segmentation label saved using a domain-specific format, such as NRRD or NIfTI, and produced a text file encoding the measurements as key-value pairs. The keys of the output correspond to the research labels assigned to the measurement classes. Not all of the measurements were generated for each of the ROIs. Specifically, calculation of a meaningful value for SUV peak (Wahl et al., 2009) was not possible when the ROI was too small. In the cases when the measurement was not generated by the tool, it was omitted from the DICOM SR measurements object.

Next, we used EncodeMeasurementsSR converter available within the Iowa2DICOM repository (QIICR, 2015c) to generate DICOM SR objects containing the calculated measurements. This converter accepted as input the list of DICOM PET object file names, the SEG object file name, and the text measurements, and produced the DICOM SR object according to TID 1500. The conversion utilized the dcmsr library of DCMTK, which provided interfaces to create and iterate through a tree of DICOM SR object content.

Validation of DICOM encoded objects

The dciodvfy tool (Clunie, 2015b) was used to ensure that an object complied with the basic DICOM encoding rules and contained the appropriate required attributes for the images, SEG, RWVM, and SR objects. This tool did not validate compliance with specific SR templates, only that valid combinations of content items and relationships were present.

The dcentvfy tool was used to validate that a set of DICOM objects contained the correct values for all attributes for the same entity level in the DICOM Information Model (i.e., that all patient attributes were the same for the objects with the same PatientID value, that all study attributes were the same for objects with the same StudyInstanceUID value, etc.). This tool was particularly helpful when objects were created along different paths or by using different tools than the original images, and/or uploaded to the distribution archive on separate occasions.

The com.pixelmed.validate.DicomSRValidator tool (Clunie, 2015c) was applied to validate compliance with the subset of SR templates that were supported by the tool, which included the TID 1500 root template and the subordinate templates used in this project. The validation consisted of checking that the required content items were present at the correct level in the content tree, that conditional content items were present when specified conditions were satisfied, that correct concepts and required values from specified context groups were used, and that concepts were encoded with the expected code meanings. Warnings were triggered when unrecognized content items were detected (which often signaled that a content item had been misplaced in the tree).

Code availability

All of the code, with the exception of that for the automatic segmentation of PET RRs, is available as FOSS without any restrictions on its use. Specifically, we share the following FOSS tools used for PET/CT data analysis:

1. 3D Slicer (Fedorov et al., 2012) was used as the platform for implementation of all the processing tools. Home page: http://slicer.org. Source code: http://github.org/Slicer/Slicer.

2. PET SUV conversion: 3D Slicer PETDICOMExtension extension. Home page: http://wiki.slicer.org/slicerWiki/index.php/Documentation/Nightly/Modules/DICOMPETSUVPlugin. Source code: https://github.com/QIICR/Slicer-PETDICOMExtension/tree/master/DICOMPETSUVPlugin.

3. Manual PET segmentation: Editor module of 3D Slicer (documentation and source code URLs are as above for 3D Slicer).

4. Semi-automated PET segmentation: PETTumorSegmentation extension. Home page: http://wiki.slicer.org/slicerWiki/index.php/Documentation/Nightly/Extensions/PETTumorSegmentation. Source code: https://github.com/QIICR/PETTumorSegmentation.

5. PET quantitative index calculation: 3D Slicer PET-IndiC extension. Home page: http://wiki.slicer.org/slicerWiki/index.php/Documentation/Nightly/Extensions/PET-IndiC. Source code: https://github.com/QIICR/PET-IndiC.

In addition to the image processing tools listed above, we provide source code of the FOSS tools used to create DICOM representations of the analysis results in the Iowa2DICOM repository: https://github.com/QIICR/Iowa2DICOM.