Guidance for Methods Descriptions Used in Preclinical Imaging Papers

Abstract

Preclinical molecular imaging is a rapidly growing field, where new imaging systems, methods, and biological findings are constantly being developed or discovered. Imaging systems and the associated software usually have multiple options for generating data, which is often overlooked but is essential when reporting the methods used to create and analyze data. Similarly, the ways in which animals are housed, handled, and treated to create physiologically based data must be well described in order that the findings be relevant, useful, and reproducible. There are frequently new developments for metabolic imaging methods. Thus, specific reporting requirements are difficult to establish; however, it remains essential to adequately report how the data have been collected, processed, and analyzed. To assist with future manuscript submissions, this article aims to provide guidelines of what details to report for several of the most common imaging modalities. Examples are provided in an attempt to give comprehensive, succinct descriptions of the essential items to report about the experimental process.

PRECLINICAL MOLECULAR IMAGING is a rapidly growing field where new instrumentation, techniques, and modalities have been developed, predominantly for rat and mouse imaging. Prior to the existence of specialized equipment for small animals, most research made use of well-established clinical imaging systems and procedures. The wide range of preclinical instruments, together with the variety of new procedures and software options for image acquisition and data creation, has enabled a great expansion of imaging research, yet few guidelines exist for the ideal options to choose when designing and conducting research. Discussing these options is beyond the scope of any journal article; however, it is imperative that these choices be adequately described in the methods section of all manuscripts.

Reporting the various elements of an imaging study within a methods section in sufficient detail is an essential part of any publication. ¹ Unfortunately, many published methods lack sufficient detailed information to enable someone else to replicate the work or provide an informed critical review of the techniques employed. This may be because the details might have seemed to be unimportant or stem from a lack of understanding about how the design might affect the experimental findings. As the preclinical imaging field develops, we frequently learn about new ways that animal handling, housing, and other experimental details can determine or alter the metabolic information being measured, ² which means that knowing in detail how published experiments were conducted is imperative. The combination of injection route, anesthesia, timing, and other factors can play an important role in understanding the optimal design and resulting data. ³ This includes appropriate reporting and/or citation for software used (e.g., vendor provided, internally developed, downloaded, or purchased routines).

Page limits and a focus on the experimental results and discussion often mean that the description of methods is inadequate. Frequently, the methods section is overly summarized and has little information on animal housing and handling conditions, specific image acquisition settings, and information about the animal physiology during the experiment. The Institute for Laboratory Animal Research (ILAR) has its own guidelines for publication ⁴ ; however, this level of detail is not possible to include in most journals. We urge scientists and journals to elevate their standards to report more detailed experimental conditions, which, if necessary, can be published as online supplemental content.

To assist with future manuscript submissions, this article aims to provide guidelines of what details to report for several of the most common imaging modalities. Examples are provided in an attempt to give comprehensive, succinct descriptions of the essential items to report about the experimental process. Because these are fairly generic examples, additional information specific to the exact research should also be added as necessary.

Animal Housing, Care, and Physiologic Monitoring

Common to all imaging methods, the housing conditions, anesthesia, heating, physiologic monitoring, and injection information are essential to include in the methods section. Most funding agencies require the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC)-accredited vivarium status. It should be noted if the animals are housed under any particular guidelines and that the Institute Animal Care and Use Committee (IACUC) has approved the research. The animal species, strain, numbers, and any pertinent details about the creation, modification, or intervention to the animals should be described. ⁵ Justification for the number of animals and experimental design along adherence to ethical guidelines needs to be stated.^6,7

Biosafety

Starting with naïve animals, some type of intervention or procedure will be required to create a model of disease for investigation. This may require biosafety controls to be in place for biohazardous or carcinogen agents and for the use of immunocompromised and transgenic animals. A brief description of any pertinent or unusual biosafety requirements should be reported. If using immunocompromised animals, note how the barrier conditions are maintained and if the health status is monitored. Measures taken to protect animals from pathogen spread on a shared instrument should be described or referenced.

Housing Conditions

The type of cages, number of animals per cage, bedding material, food, and light/dark cycle should be specified. The diet has an impact on the physiologic function (i.e., glucose levels and metabolism) and tolerance to anesthesia and therefore has an effect on enzymatic expression and uptake of common radiotracers such as [¹⁸F]fluorodeoxyglucose (FDG). ² Guidelines on animal care and husbandry in the context of preclinical imaging were recently highlighted in a dedicated issue of the ILAR Journal, with specific recommendations by modality, for both practice and reporting. ⁸ Factors such as the fluorophores present in chow with chlorophyll may contribute significantly to background fluorescent signals in optical imaging. ⁹ Bedding materials can alter the microflora present and may alter metabolic measurements. Even the caging type can alter physiologic factors related to glycolysis, blood flow, and animal stress. ¹⁰

Anesthesia

Prescan administrations of anesthetic agents (eg, paralytic, ketamine), dosages (mL/kg body weight), and timing relative to injection of the radiotracer should be noted. Generally, the type of anesthesia (injectable versus inhaled) and route of administration (intraperitoneal [IP], intravenous [IV], intramuscular [IM], or subcutaneous [SC] for the injectable agents) should be noted. For the inhalation anesthesia, the carrier gas, flow rate, amount of anesthetic mixed in the carrier gas (eg, 0.5 L/min of 2% isoflurane in oxygen), and use of humidification should also be provided. This is particularly important for certain imaging studies, such as hypoxia imaging, where the carrier gas used for anesthesia (oxygen or air) can impact radiotracer uptake in hypoxic tumor areas observed with positron emission tomography/single-photon emission computed tomography (PET/SPECT),^11,12 functional magnetic resonance (MR), ¹³ or alter photoacoustic measurements of oxygen saturation signal. ¹⁴ It is important overall to choose an anesthetic approach that is appropriate for the in vivo measure of interest to minimize its impact on specific aspects of animal physiology that can confound functional, pharmacokinetic, and/or more specific protein binding measures. This is especially critical with in vivo imaging, where the benefits afforded by longitudinal imaging could be compromised.^15,16

Heating

Over a decade of preclinical imaging has emphasized the importance of temperature regulation on the results of in vivo imaging studies. During anesthesia, the core temperature of mammals, particularly rodents, decreases very quickly; therefore, animal warming may need to be performed during various phases of the study (prior to and following tracer injection, during uptake periods, and during data acquisition). ¹⁷ Enzymatic processes are sensitive to many parameters, including temperature, time of day, anesthesia, positioning, and other factors. Blood flow in rodent tails is related to core body temperature; thus, tail vein injections require adequate heating. Bioluminescence and the trapping of many radiotracers such as FDG in brown adipose tissue are a result of enzymatic metabolism and are sensitive to these factors. Expression of fluorophores and other biological processes are also affected by these parameters; thus, a stable, reproducible procedure needs to be followed. Specify any warming of animals before, during, and after imaging and while under anesthesia.

Circadian Effects

Circadian effects are known to be present in animals and may be reversed with nocturnal animals. Rats and mice commonly used for preclinical research are normally asleep during the day and have known circadian patterns. There are chronotolerance and radiotolerance effects that result in different responses to drugs and radiation treatment depending on the time of day. ¹⁸ Luciferase expression is circadian linked, ¹⁹ which can lead to significantly different optical measurements if not considered. This effect has been exploited to examine circadian clocks by Tahara and colleagues. ²⁰ Thus, efforts should be made to make measurements at the same time of day. If this is not done, the variability in measurement times should be reported.

Positioning

The general positioning of the subject (prone or supine), the use of fixation devices (eg, limb ties, tape, stereotactic holders), or the use of an imaging chamber should be noted. Manuscripts should also state the use of surgical procedures such as skull fixation for reproducible repositioning and whether anesthesia was required for such immobilization. If measures were taken to determine the reproducibility of positioning, these should be detailed. Fiducial markers, if used, should be described (ie, liquid substance or concentration of radioactivity). If a “self-made” platform, such as a piece of cardboard, is used, this should also be noted. The positioning of the target region of interest (ROI) within the scanner's field of view (FOV) should also be given as the sensitivity and the spatial resolution may vary within the FOV for certain imaging modalities. This is especially important with the use of animal “hotels” or multiple-animal scanning devices. If the manufacturer's bed was altered in any way, this should be detailed.

Physiologic Monitoring

Physiologic monitoring of temperature, respiration rate, heart rate, oxygen saturation, blood pressure, or other relevant measurements should be noted. Information about the animal monitoring systems should be specified; the manufacturer, model, and usually location of the company are given. Physiologic effects that were corrected through gated image acquisitions should be noted, with the following specifics:

• Whether the study was performed with settings for cardiac, respiratory, or dual gating

• Basic information about the gate bins per cycle used either during the acquisition or in postprocessing

If the animals were cannulated or catheterized using any central or peripheral blood vessels, this information should also be provided as there are subtle influences of cannulation, cannulated injections, and blood sampling that can affect the results of a spectrum of in vivo imaging modalities. Reporting specifics about the nature of the cannulation and its use can provide valuable information, such as the following:

• Location of cannulation (ie, carotid artery, jugular vein) and whether it was used to inject, sample, or both

• Timing and approximate volume of blood samples, including the total volume taken from a single scanning session (ie, continuous sampling for the first 2 minutes of image acquisition, followed by samples every 25 minutes afterward, with 10 200 μL samples would result in 2 mL total blood loss)

• Treatment of indwelling catheters, which should be specified in terms of the solution used to maintain catheter patency and the schedule with which it was given (ie, catheter patency was maintained by flushing with a solution of saline and heparin once every 3 days)

In addition to these general reporting recommendations, modality-specific reporting guidelines for cannulated injections and sampling are described in more detail in respective sections of this guide. For optical and ultrasound experiments, the fur normally must be removed to allow the signal to be visible. The hair/skin and hair/transducer interfaces cause image artifacts due to the change in the acoustic impedance of the different media; therefore, hair must be removed from the areas of interest. The method of hair removal (shaving or depilatory, eg, Nair, Surgi-Cream) should be provided. Cleaning of a depilatory with any substance in addition to water should be provided (ie, a 1% acetic acid solution was used to remove the depilatory and neutralize its high pH).

Injections

Physiologic Reporting Considerations

Due to the limited blood volume of rodents, the volume of any injections and amount of any blood collection is of vital importance. For all administered substances (ie, contrast agents, radiotracer, microbubbles), the concentration of the injected agent given in milligrams per kilogram (mg/kg) of body mass along with the volume of solution given in microliters (μL) or milliliters (mL) injected and the number of moles (nmol) should be stated. Guidelines concerning the administration of substances and removal of blood are available in the Guide for the Care and Use of Laboratory Animals. ⁵ The route of administration (ie, IV, IP, oral) and the location of administration (eg, tail vein) should be stated because uptake to the target organ, tissue, or pathology can vary based on the injection site. The blood sampling site is of interest as in older mice, tail veins are sometimes cross–linked; therefore, the sampling site might be contaminated from the injection site. It is preferable that activity and volume units used be those that are consistent with the type of imaging or field of study. For radionuclides used in small animal imaging, the standard activity concentrations are typically given in either microcuries per milliliter (μCi/mL) or becquerels per milliliter (Bq/mL). The type and duration of administration (bolus, infusion) and the amount of time between injection and image acquisition should also be stated. If larger volumes than given in the guidelines are administered or collected, the motivation behind and possible effects on the experimental results should be discussed.

Reporting Injected Dose Calculations for Contrast Agents or Radiotracers

Often there is residual material at the site of injection, which can be quite substantial and may be outside the imaging FOV. If uptake measurements at sites of interest (tumors, organs) are dependent on the injection contents, some measurement of the residual activity at the site of injection must be made or steps taken to validate the injection method. Without this step, metrics such as percent injected dose and standardized uptake values (SUVs) may be both variable and inaccurate from experiment to experiment. In addition, for ligand-protein and receptor binding studies, the specific activity of the radiotracer (eg, MBq/μmol or MBq/μg) and, therefore, the nonradioactive mass injected should be specified. For a given concentration of available binding sites, this information is needed to estimate the “baseline” level of receptor occupancy that results after tracer injection (eg, if the concentration of binding sites is low, very high specific activities may be required to ensure tracer imaging conditions and sufficient detection sensitivity). For proteins and peptides, the mass may be readily determined; however, for small molecule tracers, this is rarely known. The amount of radioactivity and the radionuclide should also be included so that the reader can then determine the specific activity or activity concentration of the injected dose.

Image Preprocessing and Presentation

Validation of the study hypotheses often rests on quantitative or qualitative interpretation of image data. Therefore, it is important to ensure that the methods used to generate the images presented in publications are well defined. Such communication helps push the field forward by facilitating the research efforts of other investigators. Many of the same considerations for postacquisition image display and processing apply regardless of the imaging modality. The path from image acquisition to the results section should be described. Details to include are as follows:

• The software package that was used, including the version number, should be reported.

• Any smoothing applied to the data should be specified, including the size and type. As an example, a typical gaussian smoothing operator may allow entry of either the kernel size or the full width at half-maximum (FWHM) voxel size to be entered.

• The scale of any image should be included in the figure or figure legend. Especially with quantitative comparisons, it is important to state the window or level parameters so that the reader can be assured that each data set was thresholded equivalently and can be relatively certain that the images are comparable. This can often be expressed on the image by having scale bars alongside the image that show both the upper and lower values as well as the color gradient.

• If multiple images are overlaid, specify any registration steps required. In many circumstances, it is also important to specify the right/left image orientation given differences in radiologic and neurologic imaging conventions.

When images are coregistered, it should be noted which image was used as the reference (eg, magnetic resonance imaging [MRI], computed tomography [CT], or PET) and if the coregistered image was “resliced.” The original and final image dimensions of the structural and/or functional data sets should be provided. Details of normalization to a template should include the degree of warping and the specifics of the spatial normalization procedure.

Archiving

Many funding organizations now require that data be archived and be made available if requested. Metadata associated with the images should be included in the archival records, which includes the types of information specified above. If a database was used to track imaging sessions, indicate the database program. If the data are available online, include the website location.

Nuclear Imaging: PET and SPECT

Nuclear imaging includes SPECT and PET. There are considerations for both types of imaging that are the same, including the methods used for image reconstruction.

Image Reconstruction

The method of image reconstruction should be noted, for example, whether an algorithm or straight filtered back-projection procedures were used. For algorithmic reconstructions, the type of iterative method should be indicated, for example, whether iterative two-dimensional (2D) or three-dimensional (3D) ordered subset expectation maximization algorithm (OSEM) was used. With iterative reconstruction methods, the number of iterations and subsets used should be indicated, as well as the determination of these values (ie, by convergence or using manufacturer defaults). Changing these values can lead to under- or overiteration and introduction of image artifacts and increased noise. If 3D maximization a posteriori (MAP) reconstruction is employed, then, in addition to the iterations and subsets, it is important to note β values or resolution settings because this parameter will affect the amount of smoothing applied during reconstruction. The reconstructed matrix size of the image should be noted, as well as the voxel size. Voxel dimensions may be needed if the data were reconstructed using nonisotropic voxels. Information provided regarding the image and voxel dimensions helps assess the quality and integrity of the final reconstructed image. Voxel dimensions that are forced to be artificially smaller than the system actually supports may result in unwanted or unexpected image artifacts. If the expected resolution of the scan is ≈ 1 mm, the rule of thumb would be to choose a voxel size of no larger than 0.5 mm. In some cases, the reconstruction software may allow an unwise selection of voxel size, such as 0.01 mm. Although images can be reconstructed using this setting, because it is much smaller than the intrinsic resolution of the system, it will likely result in image artifacts from the interpolation of the small projection data to the significantly larger image dimensions, which could lead to misinterpretation of the data and erroneous conclusions.

Corrections to the PET lines of response, such as attenuation correction and scatter correction, should be reported. Calibration and corrections to voxel values such as normalization, decay correction, and dead time correction should be stated. If attenuation correction is applied to the image data, it should be noted whether the correction is based on transmission sources, the emission data itself, a CT-based attenuation method, or a lookup table method (eg, when using a coregistered MRI data). For systems that may use a transmission measurement to generate the attenuation map, the type of radioactive source, activity, scan duration, and measurement mode (eg, ⁵⁷Co, singles mode) should be reported as a noisy transmission scan can potentially bias the corrected emission data. If the system is not capable of performing attenuation or scatter correction, it should be noted that the data were not corrected for scatter or attenuation effects.

Units of measurement are critically important for proper reporting of scientific work. Some SPECT systems only report values of counts or counts per unit volume, whereas others are capable of calibration to reflect voxel values in terms of activity concentration (eg, Bq/mL). The units for any set of measurements should be reported (or if unitless, this should be specified as well).

Image Display

For image display and analysis, the software package and version that were used should be reported. Including the intensity and color scale in the images will allow the reviewer or reader to compare radiotracer uptake across multiple images. When displaying images from dynamic data sets, it is important to note whether a single frame, a summation image over the whole scan duration, or only part of the scan duration is presented. When images are coregistered, it should be noted which image was used as the reference (eg, MRI, CT, or PET) and if the coregistered image was resliced. The original and final image dimensions of the structural and/or functional data sets should be provided.

MRIs can be coregistered to PET images by rigid transformation and rotation, and as a result, the MRI data are resliced to match the voxel dimensions from the PET data. This might lead to additional smoothing or small variations in quantitation and should therefore be reported. Note that reslicing the higher resolution anatomic data set might be advisable to minimize resolution loss in the PET or SPECT data. Normalization methods as described above should also be reported for PET data that have been warped to fit a standardized template.

Dynamic Imaging

For dynamic images, the number of time frames into which the reconstructed image is divided and the duration of each frame should be reported. The frame duration is generally shorter when more frames are needed to capture rapidly changing radiotracer kinetics. This information could potentially be most easily reported by the simple inclusion of a time-activity curve (TAC) that is generated from regional sampling of the dynamic emission data. TACs can provide insight into the kinetics of the tracer throughout the study timeframe and allow comparison of the author's result to other published works. It is often useful to specify whether the time points of the TAC data correspond to the beginning, middle, or end of the frame duration.

SPECT Imaging

Acquisition

The imaging platform should be described; the author should include the number of detector heads and may also indicate the detector and pixel size or pitch or reference the appropriate article describing the imaging system. The type of collimator that is used determines the sensitivity and resolution of the SPECT acquisition; therefore, the author should indicate whether a collimator is single, multipinhole, parallel hole, slit-slat, etc. The diameter (mm) and number of pinholes along with the distance from the center of rotation (radius of rotation) should also be given. These variables will impact the number of events detected during the acquisition (sensitivity), the resolution, the size of the FOV, and the magnification, which is impacted by the selected radii of rotation. It is important to indicate the number of projections used to acquire each image, the time of data acquisition at each projection, and the degree of rotation. It should be noted whether or not the images were corrected for isotope decay either through postprocessing or during the acquisition. The energy window (keV) that was used to limit the acquisition of events or used in post hoc histogramming or reconstruction should be provided. The energy discrimination settings, chosen as part of the acquisition or in postprocessing, influence the range of energies accepted as valid events in the final reconstructed image. Changes in these settings for a given isotope can result in significant variation in values as more scattered events are accepted with wider window selections. Finally, the total number of events acquired during the scan should be reported. This provides a measure of the statistical integrity of the data set and demonstrates that the noise characteristics of the imaging protocols were sufficient for a given study.

In SPECT systems with helical acquisition capabilities, bed travel can be used to acquire longer axial FOVs. If this is the case, the extent of bed motion (mm) should be provided along with the number of revolutions traversed during the scan. Sampling information could also be given by indicating the number of projections per degree of rotation or by indicating the helical pitch of the acquisition.

An Example

Three mice were injected via the lateral tail vein with 6.5 4mUg of ¹²⁵I-SAP (≈ 5 MBq) and 22 hours thereafter 7.5 μg of ^99mTc-p5 (38 MBq) in 200 μL. After a further 2 hours, themice were given ≈ 500 μL of a 1:1 dilution of iohexol in phosphate-buffered saline intraperitoneally 1 minute before being euthanized by an isoflurane inhalation overdose. Dual energy SPECT images were acquired sequentially using an Inveon trimodality imaging platform (Siemens Medical Solutions USA, Inc., Molecular Imaging, Knoxville, TN). Low- (¹²⁵I; 25–45 keV) and then high- (^99mTc; 126–154 keV) energy gamma photons were acquired at each of 60 projections, 16-second projections with 90 mm of bed travel. A 0.5 mm diameter five-pinhole (mouse whole body) collimator was used at 30 mm from the center of the field of view. Data were reconstructed post hoc onto an 88 Å ≈ 88 Å ≈ 312 matrix with isotropic 0.86 mm voxels using a 3D maximization a posteriori algorithm (zoom = 1; β = 1; 16 iterations; 3 subsets). SPECT data sets were visualized using the Siemens Inveon Research Workplace 3D visualization software package (version 4.0) using gaussian filtering of 3.

PET Imaging

Acquisition

Information about the imaging system with a short description of resolution and performance parameters should be given or referenced appropriately. Parameters such as spatial resolution and sensitivity are of special interest as they have a direct impact on image data quality and possible uncertainties (eg, partial-volume effect). Energy and timing window and the mode of data acquisition should be noted as well. The energy and timing information influence the range of energies accepted as valid events in the final reconstructed image.

For quantitative PET data, a regular phantom-based calibration of the scanner is necessary, and the calibration procedure should therefore be noted. Usually, calibration units are given in becquerels or microcuries per milliliter. When multiple bed positions are used to acquire a whole body image, the overlap between the bed positions should be given. If used, note any markers used to identify the left and right lateral positions (eg, fiducial markers with very low activity for orientation checks).

For all scans, the duration and total injected activity in megabecquerels or microcuries should be specified. Note the time between the injection and the scan start if there is uptake time prior to imaging. If there are multiple imaging sessions per injection over a long time (eg, with ¹²⁴I or ⁸⁹Zr-labeled antibodies or peptides), note the activity at the scan times. This is of importance for longitudinal studies over several days where the kinetics of a certain radiotracer is measured on each day (eg studies with ¹²⁴I-labeled peptides). The remaining activity may also be an important factor for ex vivo gamma counting if samples exceed the counter deadtime capacity.

An Example

PET studies investigating the tumor type–dependent uptake of [¹⁸F]FAZA (hypoxia imaging agent) were conducted with CT26 tumor–bearing BALB/c mice (n = 10 mice). Under 1 to 2% isoflurane anesthesia in oxygen at 0.6 L/min, the lateral tail veins of the 9-week-old mice were injected with 14.2 ± 0.5 MBq [¹⁸F]FDG or 12.3 ± 1.0 MBq [¹⁸F]FAZA in a volume of 100 μL on 2 consecutive days. For [¹⁸F]FDG, a 1-hour uptake time was allowed prior to a 10-minute static scan. After the PET emission scans, a transmission scan was performed for 13 minutes using a rotating ⁵⁷Co source. The body temperature of mice was maintained at ³⁷°C on the scanner bed and in the incubation chambers using heating pads. During the 1-hour [¹⁸F]FAZA uptake time, mice were kept conscious in room air. Static 10-minute duration PET scans were performed at 1, 2, and 3 hours postinjection using an Inveon-dedicated PET scanner (Siemens Medical Solutions USA, Inc., Molecular Imaging, Knoxville, TN) with an axial field of view of 12.7 cm and a spatial resolution of 1.4 mm full width half-maximum.

PET data were reconstructed using Inveon Acquisition Workspace software version 1.4 using an OSEM 2D algorithm into a 128 × 128 image matrix (resulting in final voxel dimensions of 0.79 × 0.79 × 0.8 mm). Dead time, decay correction, attenuation correction, and normalization were applied to all PET data. Data were analyzed by drawing standardized volumes of interest. In the transverse image planes, a circular region of interest of 4.8 mm in diameter was placed over the maximum activity in the tumor. Image analysis was performed with the Siemens ASIPro VM software package version 6.6.

X-Ray CT Imaging

CT imaging parameters can be grouped into hardware settings for the gantry, source, detector, and postprocessing software settings.

Gantry Parameters

Gantry parameters control the way in which the gantry moves about the object during the CT study. This information gives insight into the quality of the scan based on the physical acquisition of the data, including the angular sampling, angular coverage of the source, and detector, as well as communicating an overall picture of how the data were collected. CT systems can vary widely in their acquisition modes even though the concepts for each remain identical. The total amount of gantry rotation should be reported along with the number of projections.

CT systems vary in the methods used for acquisition. In vivo imaging platforms typically keep the object stationary while the source and detector rotate about the object. Systems designed for ex vivo imaging often have a stationary source and detector and rotate the object in the FOV. Some systems have helical acquisition modes to extend the axial FOV; thus, the bed travel and overlap settings should be noted in the acquisition parameters.

Most current preclinical CT systems have automated gantry hardware that enables easy movement of the x-ray source and detector components to points within the gantry. Although the physical location of the source and detector need not be known, it is important for authors to understand and mention the magnification or position setting used during a study to give information regarding the setup of the gantry. The magnification settings can be critical for readers to make determinations about the setup of a given study.

X-Ray Tube and Detector Parameters

The primary x-ray tube settings to report are the beam energy (kVp), current (mA, μA), and exposure time (ms). For preclinical CT systems that contain two sources or that were acquired using multiple energies, all energies should be reported. Beam energy should be noted for all scans as this particular tube setting is the determining factor for the overall radiographic contrast of the image. The beam energy also has an effect on the number of photons recorded at the detector (radiographic density) because a higher beam energy setting will result in more photons that are capable of penetrating the object. Current (μA) should also be noted for all studies as this value is the primary determination of the radiographic density. The energy and current settings indicate the total beam power used for the study and are an important factor in being able to obtain an estimate of the dose delivered to the object during a particular study. For any CT study, the exposure time per projection or the current should be recorded as this directly affects the scan time and total x-ray radiation dose.

X-ray sources create a range of photon energies, with the maximum (or peak) usually being specified as the performance metric (ie, 80 kVp). The low-energy photons are useless for imaging because they never reach the detector and only contribute to radiation dose; therefore, they are filtered out before reaching the animal. The type of filter and thickness should be noted.

Some systems have the ability to acquire data in a fluoroscopy or semifluoroscopy mode. Report the total time that images were acquired and the frequency with which images were taken.

Image Reconstruction

Several key parameters should be reported to give information on how the CT data were reconstructed. For most CT data, the algorithm used will be a modified Feldkamp reconstruction; however, iterative reconstruction techniques are becoming more popular as workstation processor power has improved. The algorithm type used should be reported as the representation of the data between analytical and iterative reconstruction techniques is markedly different. If iterative algorithms are used, then the number of subsets and iterations should be reported, as well as any additional filters that may be applied to the data.

Use of reconstruction filters should be noted as this can have a significant effect on final voxel values and the overall representation of the data. Typical reconstruction filters for CT include Shepp-Logan and Hamming filters, but many other specialized filters exist. Additionally, different manufacturers often use proprietary names for a given type of filter. The author should attempt to name the type of filter and give a one-line description about its effect on the data, although for well-known filters, the description should not be necessary. For example, “a bone filter was used that provides greater image contrast for high-density materials.”

A major issue in CT reconstruction is how to handle the effects of beam hardening and metal artifacts within CT images. Calibration of CT data with and without these corrections can have a significant effect on the voxel values calculated during reconstruction. Lack of knowledge as to whether these corrections were used can lead to inaccuracies in those wishing to validate results or build on a given body of work. Authors should report whether these corrections were used to create the final reconstructed CT data.

The standard metric used for CT image data are Hounsfield units. The system needs to be calibrated, typically with air (–1,000) and water (0). Routine white and dark field calibrations are required to correct for detector and readout noise.

The x-ray detector settings also should be reported as they provide information regarding the collection of the CT data. Publication submissions should include the total imaging FOV in physical dimensions and pixels. Most preclinical CT platforms allow the user to control cropping of the detector readout; therefore, the cropped FOV dimensions should be given rather than the FOV with which the system is capable of imaging. A rule of thumb is that the reconstructed pixel size should be approximately half of the desired resolution. Resolution or reconstructed pixel sizes are acceptable ways to report the minimum detail available in an image; however, resolution should be stated as a FWHM or full width tenth-maximum measurement based on the protocol studied. Reconstructed pixel size is usually an operator-chosen parameter that is affected not only by the acquisition settings but also by downsampling, or binning, the acquired data within the reconstruction software. Miscommunication of these values can lead to impossible expectations from a given publication with a given piece of imaging hardware. Ideally, both parameters are known and can be reported in a single line, such as “The data were reconstructed with a pixel size of 10 microns, resulting in an image with a FWHM resolution of approximately 20 microns.” This provides the greatest range of information and assists readers in determining what imaging capabilities are needed to perform the published study.

Contrast

For CT, it is common that contrast agents are used when imaging small animals because of the low soft tissue contrast that is inherent in these systems and the small subjects being imaged. Typically, contrast agents are either injected via the tail vein, injected intraperitoneally, or given orally. The type of contrast used should be reported as well as the delivery method and the concentration that was delivered.

Image Display

CT data are most often presented with the assistance of software that is specifically designed for advanced 3D rendering of the image volume. Any software used to visualize the data should be reported, as well as primary image display parameters such as window or level, image filters, and any processing of the data that may have been performed on the image data as part of the visualization and analysis workflow.

An Example

Mice were imaged in a MicroCAT II small animal CT system (Siemens Medical Solutions USA, Inc., Molecular Imaging, Knoxville, TN). Exposure settings were 70 kVp, 500 mAs, 500 ms exposure time, and 360° rotation in 1° steps with 2.0 mm aluminum filtration. Images were reconstructed using a modified Feldkamp reconstruction process to a cubic voxel size of 0.20 mm in a 256 × 256 × 496 matrix. Using the vendor software, CT values were converted into Hounsfield units (HU) using the formula HU = (μt – μw)/(μw – μa)*1,000), where μw and μa are the linear attenuation coefficients of water and air, respectively, and μt is the linear attenuation coefficient of tissue.

Magnetic Resonance Imaging

The image intensity in MRI is influenced by a large number of factors. These factors can be exploited to change image contrast to highlight desired features. The way that MRI data are collected for a particular experiment is described by what is called a pulse sequence. This is a representation of the timing in which radiofrequency (RF) and gradient pulse are executed, along with the timing of the data acquisition. The pulse sequence is typically graphically depicted. There are a large number of pulse sequences that are routinely used. In addition to the RF and gradient pulses and timing of data acquisition, the type and magnitude of the pulses and the presence of any externally administered contrast agents directly change the image contrast. The hardware used can also affect the signal to noise (SNR) and contrast to noise (CNR) ratios.

Hardware

It is important to specify what hardware is used to clarify system performance in terms of several factors, including but not limited to the magnitude of the signal, the relaxation times, and the spatial and temporal resolution that are achievable. The manufacturer of the hardware, along with the location, should be reported. The static magnetic field strength should be provided, along with the maximum gradient strength achievable in units of Tesla.meter⁻¹ (T.m⁻¹) and the gradient slew rate in T.m⁻¹ s⁻¹. Also, it should be noted whether the gradients are shielded as this reduces eddy current effects. The size and type of RF coil should be noted. Some common types of coils include surface, birdcage, saddle, and Helmholtz coils, and these can be further classified as linear, quadrature, and phased array. For phased array imaging, the number of elements should be noted. The manufacturer and model number of any ancillary equipment, such as that used for physiologic monitoring and gating, should also be noted.

Acquisition

MRI can be used with nuclei that have a net spin angular momentum and a resulting magnetic moment. These include nuclei such as ¹H, ³¹P, ¹⁹F, ¹³C, ²³Na, ³He, and ¹²⁹Xe. The nucleus being imaged should be stated. These nuclei are magnetically active and can interact with one another. This interaction can be minimized using spin decoupling. It should be noted whether a multinuclear RF coil was used and whether spin-decoupling was applied.

The type of pulse sequence used should be noted. It is outside the scope of this article to describe each of these; there are a number of good textbooks of MRI pulse sequences.^21,22 Types of pulse sequences include but are not limited to spin echo, fast spin echo, gradient echo, diffusion weighted, diffusion tensor, echo-planar, spiral, perfusion weighted, arterial spin labeling, susceptibility weighted, MR angiography (further divided into time of flight and phase contrast), functional MRI, and steady-state imaging sequences such as FLASH/SPGR, FISP/GRASS, and PSIF/SSFP. It should be noted if any specialized RF pulse was used, for example, a spectrally and spatially selective pulse, a binomial pulse, or an Ernst angle.

These sequences have acquisition parameters that can be adjusted to accommodate differences in subjects and in the imaging target of interest. Common parameters that influence the image SNR and CNR include the repetition time between RF excitation pulses (TR), the echo time (TE, which is usually the time between the excitation pulse and the center of the readout time), the flip angle of the excitation pulse (α), and the readout bandwidth. Other common adjustable parameters include the in-plane size of the FOV, the slice thickness, and the number of points acquired in both the readout and phase-encode directions. The spatial resolution is the FOV divided by the number of points in the readout and phase-encode directions. The SNR will increase linearly with the square root of the number of signal averages. The number of averages should be reported.

Any physiologic gating or triggering should be noted. It should be stated what was triggered (eg, cardiac, respiratory, or both) and whether prospective triggering or retrospective reordering was used. The placement of the triggering with respect to the acquisition of data and the MR pulse sequence should be described.

If magnetization preparation pulses are used, they should be described, along with their timing in the pulse sequence. These pulses could include an inversion pulse for short tau inversion recovery (STIR) or fluid-attenuated inversion recovery (FLAIR) imaging, fat or water saturation pulses, spatially selective saturation pulses, magnetization transfer pulses, spatial modulation of magnetization (SPAMM) pulses for myocardial strain imaging, etc.

It is also possible to acquire MR spectroscopy data. For these studies, the nucleus under study should be stated and the type of pulse sequence should be noted, as well as whether the acquisition was accomplished using a single-voxel technique such as stimulated echo acquisition mode (STEAM) or point resolved spectroscopy sequence (PRESS) or whether a mutlivoxel technique such as chemical shift imaging (CSI) was used. Because the signal of interest in spectroscopy is largely from metabolites that are at low concentration levels, efforts must be made to suppress signal coming from sources that are highly concentrated, such as water and fat. The type of suppression should be described. The suppression techniques commonly used include chemical shift selective pulses and spatial presaturation pulses. The number of points acquired during the readout should be noted along with the readout bandwidth.

Contrast Agents

If a contrast agent is used, the chemical composition of the agent should be identified, including the active metal of the agent (eg, Gd, Fe), the chelating group, and the MR property that it affects (eg, T₁, T₂*, chemical shift). The concentration of the injected agent, along with the amount of solution injected, should be stated, along with the route of administration (eg, IV, IP, intratracheal). The amount of time between injection and image acquisition should also be stated.

Image Reconstruction

The method of image reconstruction should be described. If this is not reported, it should be assumed that a standard Fourier transformation (FT) was used. The apparent spatial resolution can be increased by zero-filling the raw data before an FT; if this is done, it should be reported. Any filtering of the data before or after FT should be noted. If the data were not acquired rectilinearly, then the means of regridding should be described. If a non-FT means of converting the spatially frequency data to spatial space is used, such as projection reconstruction, this should be noted and described in detail. The final reconstructed resolution should be reported.

An Example

Images were acquired on a 7 T ClinScan MR system (Bruker, Ettlingen, Germany) using a gradient insert capable of 6.5 × 10⁻¹ T/mmaximum strength and 6.67 × 10⁻³ Ts/m maximum slew rate. The radiofrequency coil was a Bruker quadrature birdcage transceiver. Cardiac triggering pulses were generated using an SAII monitoring and gating system (SA Instruments, Inc., Stony Brook, NY). For both bright and black blood imaging, a gradient echo pulse sequence was used to acquire data, with an echo time of 3.7 ms, a matrix of 128 × 128 zero-filled to 256 × 256, a field of view of 2.56 × 2.56 cm, a slice thickness of 1mm, a flip angle of 20°, and four signal averages. Six to seven short-axis slices (average 6.4 ± 0.5 slices) were needed to image the entire murine left ventricle. Repetition time was determined by the heart rate and the number of phases (12–14) acquired across the heart cycle. For black blood imaging, a nonselective/selective combination of 180° pulses (double inversion recovery [IR]) was used. A 500 μs square pulse was used for the nonselective inversion and a 2 ms sinc pulse was used for selective second inversion. The selective pulse was centered on the imaging slice and had a width equal to three times that of the imaging slice. The selective pulse was applied immediately after the nonselective pulse. The double IR pulses were played out immediately after the detection of the cardiac R wave. On the detection of the next R wave, a multiphase acquisition was employed whereby 12 to 14 lines with the same phase-encoding value were acquired. This process was repeated, incrementing the phase-encode gradient until all 128 phase-encode values were obtained. Image data for these sequences were acquired on alternating heartbeats to ensure that the beginning of the cardiac cycle was captured.

Ultrasound Imaging

Ultrasound imaging is used for the assessment of anatomic features (tumor volumes), ²³ blood flow in large vessels, ²⁴ and measurement of cardiac function. ²⁵ Ultrasound imaging uses a transducer to produce pulsed sound waves that propagate into tissue and that reflect from interfaces between tissue components of different density and compressibility. The detection of the reflected sound waves by the transducer is processed to form an image, based on the intensity of the reflected sound waves and the time delay, which corresponds to the depth from which the echoes originated. Therefore, to fully describe the ultrasound process, it is necessary to provide information on the transducer, pulsed sequence, and focus depth.

Hardware

The imaging platform should be described, as well as the positioning of the mouse on the imaging table (eg, prone, supine, left or right-lateral, taping of limbs) and the transducer (manufacturer, model, center frequency, geometric focus, image depth, power rating, and frame rate), and the author should reference the appropriate publication describing the imaging system.

Acquisition

In addition to hardware parameters and animal handling techniques such as the application of a depilatory and anesthesia, as described above, information should be provided for specific acquisitions, that is, gating such as cardiac or pulmonary to reduce image motion artifacts. A heated gel is applied to maintain matching acoustic medium between the animal and the transducer to eliminate image artifacts.

The primary modes for ultrasound imaging are B mode (bright mode), used for anatomic imaging, ²⁶ and M mode (motion mode), which is used for analysis of cardiac tissue motion (ie, ejection fraction). ²⁷ The position of the transducer with respect to the mouse and organ or tumor, the image mode (B or M modes), slice thickness, and a description of the cine loop (total number of frames and frame rate) should be provided. Given that ultrasound transducers produce 2D images, it is necessary to specify the step size (slice thickness) for acquiring 3D images. If a vendor-specific acquisition package is used, the author should provide all details or a reference.

For dynamic acquisitions and contrast-mode imaging, the number of frames for each phase, including the frame rate (injection, wash-in, and predestruction phase, time of microbubble burst, reference, and postdestruction phase), should be provided. This can be provided as a figure for easy reference for the reader.

Doppler Ultrasonography

Doppler ultrasonography is based on the difference in the transmitted frequency and the return frequency due to the relative motion of the blood and is displayed as color Doppler ultrasonography. ²⁸ The relative velocity between the red blood cells and the transducer depends on the cosine of the angle between the sound traveling from the transducer and the direction of motion of the red blood cells. When acquiring color Doppler ultrasonography, the angle between the vertical line of the ultrasound pulse to the direction of the vascular flow should be stated. Doppler angles between 30° and 70° result in minimal error in the velocity, whereas high angles result in incorrect determination of flow rates. Power Doppler ultrasonography ²⁶ has been implemented for imaging low-flow states and is independent of transducer angle.

Image Reconstruction and Display

The image matrix size should be noted as well as the voxel size. Also include any filters used to process the images and the type of ROI (manual or automatic). The software package that was used, including the version, should be reported. If the program was developed by the institution, then an overview and reference should be provided.

Contrast Agents: Microbubbles and Nanoparticles

Ultrasound contrast agents include microbubbles, ²⁹ echogenic liposomes, ³⁰ and perfluorocarbon droplets, ³¹ which have a density and compressibility significantly different from those of blood and tissue, resulting in an enhanced echo (signal). A description of the imaging agent(s) used in the experiment should be provided. It is important to indicate the route of administration (IV, IP, SC, IM, intratumoral, retro-orbital), needle gauge, and nanoparticle and/or microbubble construct because the uptake to the target organ will vary based on the injection site and construct of the imaging agent. The volume of the injected imaging agent and the solution in which it is suspended should also be reported. In addition, the type of injection (bolus or constant infusion), including the infusion rate (mL/min), should be reported.

An Example

All animal experiments were conducted in accordance with the animal and care use procedures of the Frederick National Laboratory for Cancer Research (Frederick, MD). Subdermal tumors were induced in 10 athymic nude mice with 5 × 10⁵ BT-474 breast cancer cells into the lateral aspect of the hind leg. Tumors were imaged at 4 weeks postinoculation. Prior to imaging, the mice were prepared by shaving the area around the tumor using hair clippers and applying depilatory cream (SurgiCream, American International Industries, Los Angeles, CA) to remove fine hair, followed by 1% acetic acid to remove the depilatory and neutralize its high pH. The cage of mice was placed on a heated blanket maintained at 36°C for 30 minutes prior to initiating the imaging study. A tail vein catheter with a 27-gauge needle attached to a 1 mL syringe containing 100 μL of the microbubble contrast agent was inserted. The syringe was mounted on an infusion pump (40 μL/min) located on a rocker system to keep the microbubbles in suspension.

The untagged microbubble contrast agents (MicroMarker Contrast agents, Bracco Research SpA, Milan, Italy) are composed of a perfluorocarbon gas encapsulated by a lipid shell, average diameter of 2 to 3 μm. A high-frequency (40 MHz) ultrasound imaging system (Vevo 2100, VisualSonics, Toronto, ON) was used in these studies. B-mode images were acquired at the center frequency of 40 MHz (transducer MS-250) with an axial and lateral spatial resolution of 75 and 165 μm and a transmit power of 4%. A heated gel (Aqua-Gel, Parker Laboratory, Inc., Fairfield, NJ) was applied to maintain matching acoustic medium between the animal and the transducer. Cine loops of approximately 500 frames were collected at 45 Hz with all acquisition and transducer parameters specified by the manufacturer's protocol. The acoustic focus was placed at the center of the tumor. Throughout the imaging session, mice were kept anesthetized with 2% isoflurane in medical air at 1 L/min on a heated stage according to the manufacturer's protocol. Respiratory gating was used to synchronize data acquisition with the mouse respiratory cycle to reduce motion artifact.

After a 2-minute infusion resulting in a stable flow of contrast within the area of interest, the image sequence was initiated, and a destruct pulse occurred at 10% of the loop, which destroyed the microbubbles within the imaging plane. The image sequence was completed after the predetermined number of frames was acquired. Tumor perfusion parameters were analyzed using the analysis software VEVOCQ (VisualSonics). Prior to image analysis, motion correction techniques were implemented according to the manufacturer's software parameters.

Optical Imaging

State whether bioluminescent or fluorescent imaging was employed. Specify the animal type and species (such as nude mouse) and any steps taken to remove hair to improve light transmission.

Bioluminescence Imaging

For bioluminescent imaging, describe any unique details about the luciferase enzyme or luciferin substrate (if any). Describe how the insertion of the gene into animals was accomplished (eg, viral vectors, cell transfusion) and if there were any specific promoter sequences or additional genes present as dual or trifusion constructs. If cells were manipulated and inserted, describe the process and cell type. Substrate injection route should be noted (IP, IV, SC), along with the volume and amount.

Fluorescence Imaging

Specify the fluorophore, excitation and emission wavelengths, and wavelengths used for measurements. Report the exposure time because long exposures could lead to photobleaching and a loss of signal. Note whether the excitation light source was on the same side of the animal as the detector (epi-illumination) or on the opposite side going through the animal (transillumination). Describe how the fluorophore was placed in the animal, specifying the method of insertion (eg, injection, volume). If the fluorophore was genetically engineered to be expressed by cells within the animal, describe the process, including the promoter and any additional genes included with the fluorophore (ie, dual or trifusion constructions for multiple imaging or gene expression options). Describe any steps taken to reduce endogenous fluorophores or autofluorescence (eg, changes in diet, shaving the hair). If multiple fluorophores are present, describe how the emission wavelengths were separated.

Equipment

Specify the equipment used and any pertinent settings required to replicate conditions with the imaging system software or unique requirements. State whether the optical data is from planar, tomographic, or transmission images.

Acquisition Parameters

Authors are encouraged to note the time between substrate injection and measurements, along with the duration of measurement in seconds and how many data points were acquired to arrive at a measurement value. If data were collected by number of counts or photons/s, or another metric other than time, note the method and reason for choosing a particular value as a standard procedure. Note if there were changes in imaging duration over the course of an experiment and ensure that comparisons of different data sets use rates of emission (acquisition time independent) rather than a total number of detected events (time dependent).

Image Generation

Note the software used and version number, along with any settings used to create the images. If there are background subtractions, such as dark current and cosmic rays, note those along with any other corrections. When displaying multiple images, if possible, use the same scale (minimum and maximum values) for all images; otherwise, make sure to specify the scale for each image. Describe any processing steps necessary to create images and data from the images.

Image Analysis

Given that the light that makes it out of the body to the camera is scattered and much of it is absorbed, care must be taken with interpreting the results from optical methods. Light transmission is wavelength dependent, and each medium the light passes through has different scattering and optical properties, along with changes due to the index of refraction at tissue/tissue or air/tissue interfaces. When specifying a number with respect to the light measured, keep in mind that much of the light may have been absorbed from deeper tissues. Given that some tissues are more optically transparent at certain wavelengths, the true location of the signal may not be exactly where the image displays the most light.

For quantitative analysis of optical data, specify the method used to determine the measurement. This includes the way in which the region is defined and the units of measure. In some cases where multiple detector elements are summed together for each image pixel (binning), a maximum pixel value may be most representative of the light output because this would occur from the brightest location closest to the surface. In other cases, the total light or average light measured within a region may be most suitable. Determining a region for measurement can be very subjective; thus, reporting on the method for selecting a specified region is highly encouraged.

An Example

Nude mice (nu/nu) with subcutaneously implanted tumors on the left shoulder transfected with dual fusion herpes simplex virus thymidine kinase/luciferase were imaged on days 7, 10, and 13 using luciferin bioluminescence. Prior to imaging, mouse cages were warmed on a 36°C heating plate for 20 minutes prior to anesthetizing the mice using 2% isoflurane gas anesthesia in oxygen at 2 L/min, followed by intraperitoneal injection of 100 μL luciferin stock solution in phosphate-buffered saline (30 mg/mL). Up to five mice were placed prone with the tumor facing upward in a preheated imaging box with anesthesia connections and then placed inside the IVIS Lumina II bioluminescence/florescence scanner (PerkinElmer, Alameda, CA) for imaging using Living Image software version 4.0 (PerkinElmer). Multiple 30-second images were acquired beginning at 10 minutes postinjection until three sequential images had approximately the same light emission from a region drawn around the primary signal, and the values reported are the average of three measurements. Optical measurements were acquired at a 25 cm field of view (FOV) using the extended FOV lens as photons/s, using binning of 4, an F stop of 1.2, corrected using default settings for cosmic rays, flat field correction, and background subtraction. Optical data were overlaid with a photograph acquired with a binning of 2, an F stop of 8, a 0.2-second exposure, and the same corrections applied as for optical data. Optical data are reported as the maximum pixel values for each region with units of photons/s/cm²/steridian.

Summary

Detailed reporting of methods is necessary for both reviewers and people interested in the science being presented. The details are extremely important when it comes to replication of and building on previous work. The information necessary to report only increases with time as more methods, more instruments, and more detailed examinations are made using preclinical imaging models. Online publications now offer a great opportunity to provide additional detailed information through copublication of supplemental materials. The work by Tahara and colleagues is a good example of how supplemental materials can offer both more information and validation of the choices made for an experiment. ²⁰

Authors are encouraged to follow the recommendations outlined herein as these practices will likely improve and speed up manuscript reviews and result in higher quality publications that are more likely to be referenced in the future.