8. Quality Assurance Procedures

8.1 Neuroimaging quality control

      8.1.1 Scanner level QA

            Detailed Descriptions of ACR QA Scans

            Detailed Descriptions of fBIRN QA Scans

      8.1.2 Subject-level QA

      8.1.3 Data Quarantine and Release

      8.1.4 Scan review and reporting of abnormal findings

8.2 Neurobehavioral Data Quality Control

References

Appendix 1A – Image Quality Rating Scale Anatomical MRI

Appendix 1B – Image Quality Rating Scale DTI/HARDI MRI

Appendix 1C – Image Quality Rating Scale Functional MRI

Appendix 2 – A classification system for neuroradiologic findings in imaging research studies with normal health children.

Appendix 3 – Template for reporting neuroradiologic findings in healthy pediatric participants

 

Introduction - This section outlines the methods used for image quality assurance.  Delivery of consistent, high-quality image data to the C-MIND database is imperative for the neuroimaging and brain development communities.  Quality control and quality assurance has been an essential part of the process in the data acquisition, analysis, management and archiving process for the C-MIND database. Quality assurance processes were established for both neuroimaging data and for cognitive and behavioral data.

8.1 Neuroimaging quality control

A 3-phase quality assurance program was implemented to guarantee

  1. optimal scanner performance,
  2. quarantine and quality inspection of all image data from human subjects and
  3. quality rating index assigned to each image uploaded to the database. 

Methods for quality assurance of scanner performance have been adapted from standard procedures published by the American College of Radiology (ACR) and the functional Brain Imaging Research Network (FBIRN).  Specifically, methods have been adopted from the ACR QA manual1 for assuring that each MRI scanner contributing brain image data to the C-MIND database is performing according to its specifications for geometric accuracy and signal to noise of images.  Imaging protocols are carried out on a standard ACR phantom2 for this purpose and image quality checks are performed automatically using a computer program developed in IDL for C-MIND. Performance of the MRI scanner for functional imaging and ASL perfusion imaging is evaluated using the methods developed and promoted for multi-site MR research studies by the FBIRN.3,4 For these tests the FBIRN phantom is used at each site, along with Python scripts that compute the scanner stability and SNR parameters.5 These methods are described in Section 8.1.1 below. Quarantine and quality inspection procedures involve several stages and some human interaction with the imaging data.  These steps are described in Sections 8.1.2 – 8.1.3.  Once completed, a numerical quality factor is assigned to each acquired image and stored in the database. 

Approximately 10% of normal, healthy children who enroll in neuroiomaging studies have abnormal findings upon radiologic review of the anatomical images acquired as part of the C-MIND protocol7. Handling these findings in an ethical and responsible manner is described in Section 8.1.4.

8.1.1. Scanner level QA

Scanner quality assurance testing was performed during each week that a human subject scan was acquired on the scanner. QA scanning was done using two test objects:  The American College of Radiology (ACR) MRI QA phantom2, and the FBIRN phantom5.  A separate QA protocol is saved on the scanner for each phantom and both were run each week during the QA session.  The ACR Quality Assurance Scan Session, listed below in Table 3.1 takes approximately 15 minutes to complete while the FBIRN Quality Assurance Scan Session, listed in Table 8.2 can be completed in less than 10 minutes.  These scan protocols are followed by more detailed descriptions of key scan parameters for each image sequence included in the protocols. 

Positioning of the phantoms for the QA scans is an important part of the QA process and selection of imaging locations requires some skill and training.  Therefore, the QA sessions should be performed by a trained technologist or MR Physicist.  The ACR phantom is marked with cross hairs to indicate its center.  The phantom should be positioned in the head coil on the scanner table such that the laser cross hairs of the scanner align with the cross hairs of the phantom.  This will insure that the phantom is aligned correctly in 2 of 3 planes. To align the 3 planes, it is recommended to use an MR compatible bubble level gauge placed along the small shelf on the front of the ACR phantom.  The FBIRN Phantom should also be landmarked to MR isocenter; though rotation is not critical for this object as it is internally homogeneous without structure.

 

Table 8.1: List of Scans in the ACR Quality Assurance Scan Session – To be performed using the ACR phantom

Table 8.1 List of Scans in the ACR Quality Assurance Scan Session.png

 

Detailed Descriptions of ACR QA Scans

Survey

Fast-field echo sequence – Any fast gradient echo method will do for this purpose.  This scan is not used for any quantitative analysis.

The survey scan is used to position the phantom correctly in the scanner and to readjust if necessary to obtain optimal alignment of the ACR phantom in the isocenter of magnet, at the center of the image field of view (FOV).  As described above, prior to running the Survey scan, the ACR phantom should be carefully positioned in the head coil and landmarked so that the cross hairs on the object are at the isocenter of the imaging gradients.  The survey scans are also used for positioning of the imaging planes for the first QA scan in the QA imaging session. We use a fast gradient-echo method with a total scan time of approximately 30 seconds. Three slices are acquired in each of three orthogonal scan planes so that subsequent scanning prescriptions can be made and confirmed in any of these three planes.

Table 8.1 Resolution and Contrast.png

 

Sagittal Spin Echo

Single slice, proton density weighted spin echo for positioning of QA slices

This is a standard spin echo image with a single slice positioned through the center of the phantom.  If the phantom was positioned properly with the laser cross hair on the cross hair of the phantom, this sagittal scan should land in the correct location.  This single slice image is acquired in the sagittal plane at a spatial resolution of 1 x 1 x 5 mm.  Be sure SENSE is turned off during the entire protocol.  No SENSE reference scan is needed.  Once the initial sagittal scan is complete, this image is used to prescribe the subsequent scans.  All remaining scans are done using the same method and contrast parameters but with 11 slices rather than 1 slice.  Figure 8.1 illustrates the central slice of this 11 slice pack.

Fig_8.1 Sagittal localizer.png

Figure 8.1 Sagittal localizer image showing the 11 required axial slice locations and the paired 45° wedges.  The words “CHIN” and “NOSE” on this image indicate the locations where those same words are etched into the phantom as an aid to orienting it for scanning as if it were a head.

 

KeyImagingGeometryContrast.png

 

Axial Spin Echo

Multi-slice, proton density weighted spin echo for ACR QA imaging

This is a standard multi-slice spin echo image with a 11 slices positioned axially through the phantom such that the bottom slice of the pack is position at the vertex of the 45 degree wedges as shown in the diagram below.  The slices are separated by a 5 mm gap so if the phantom was positioned properly with the laser cross-hair on the cross-hair of the phantom, these axial scans should now prescribe in planes that are parallel with and aligned on top of the grid seen in the phantom, with the top most slice centered on the vertex of the top set of 45 degree wedges. 

Eleven slices will next be acquired in the axial plane at a spatial resolution of 1 x 1 x 5 mm and a gap of 5 mm.  These scans are used to examine spatial resolution and contrast resolution as outlined in the ACR procedure manual.

Be sure to keep SENSE turned off during the entire protocol. All remaining scans are done using the same prescription of 11 axial slices in the same positions but with slice gaps varying from 0 to 5 mm.

AxialSpinEcho.png

 

SliceGap0 (SliceGap0.5, SliceGap1, SliceGap5)

Multi-slice, proton density weighted spin echo for ACR QA imaging

This is another axial multi-slice, spin echo scan with 11 slices as in the axial scan above.  This image and the next three are identical with the exception of the slice gap.  They are used to examine slice cross talk as outlined in the ACR procedure manual.  They are positioned with bottom slice set at the lower edge of the phantom, indicated in Figure 1 by the label CHIN.  The parameters are listed below and are similar to the Axial scan listed immediately above.

SliceGap0.png

 

Table 8.2 List of Scans in the FBIRN Quality Assurance Scan Session.png

 

Detailed Descriptions of fBIRN QA Scans

Survey

Fast-field echo sequence – Any fast gradient echo method will do for this purpose.  This scan is not used for any quantitative analysis.

The survey scan must be repeated after positioning of the FBIRN phantom in the isocenter of magnet for this QA series.  A new Patient File name should be used with the file name of IRC04H_QA_fBIRN_mmddyy.  As described above, prior to running the Survey scan, the FBIRN phantom should be carefully positioned in the head coil and landmarked so that the object are at the isocenter of the imaging gradients.  The survey scans are also used for positioning of the imaging planes for the remaining QA scans in this QA imaging session.  This survey is identical to the Survey scan described above for the ACR phantom so the parameters will not be listed here. Please refer to the paragraph and parameters above for details.

Ref_Head_32

Proprietary SENSE Reference scan

This scan will differ for each vendor as it is a proprietary series used to calibrate the sensitivity of multi-channel receiver RF coils for SENSE or GRAPPA reconstruction.  No details are provided here except to say that SENSE is used for the FBIRN phantom scans and therefore the reference scan should be acquired at the beginning of the series.

Resting State BOLD

Gradient Echo EPI single shot sequence

This is the same BOLD-weighted EPI gradient echo scan used for the resting state fMRI scan during the Functional Imaging Session. The EPI Scanning runs continuously with a TR=2 seconds and TE=35 msec.  A total of 200 scans are acquired with the key imaging parameters as listed below.  This image time series is analyzed in the QA scripts to monitor temporal variation of SNR and overall signal drift due to heating of the scanner and sample during the fMRI scans.

SV_PRESS_35

Single voxel Point resolved spectroscopy (PRESS) - a multi echo single shot technique

(PRESS) Point resolved spectroscopy is a multi echo single shot technique to obtain spectral data. PRESS is a 90°-180°-180° (slice selective pulses) sequence. Although PRESS is a volume selective spectroscopy method, the voxel should be made large enough to encompass the same field of view used for the Resting State BOLD acquisition above.  The aim is to acquire an FID that can be used to examine the line width of the spectrum obtained from the phantom.  Parameters are not critical other than to insure that an FID is obtained from the entire sample.

Separate filenames are used for the ACR scans and the FBIRN scans using the file naming convention of:  IRC04H_QA_ACR_mmddyy for the ACR phantom scans and IRC04H_QA_fBIRN_mmddyy for the FBIRN phantom scans.

Note that both phantom QA sessions are performed using the same 32 channel SENSE head coil used for the human subjects because checking the RF coil performance is also critical.  The multi-channel SENSE reference scan is omitted from the ACR QA session because SENSE was turned off for this session.  However, the reference scan is acquired as part of the FBIRN QA session.

The QA imaging protocols used for the weekly QA scans permit calculation of several key performance measures that related to scanner gradient, RF and shim gradient hardware as well as receiver characteristics. A complete list of the QA parameters computed from ACR and fBIRN phantom scans is listed in Table 8.3 and Table 8.4.

The QA parameters listed here are computed according to the algorithms outlined in the ARC QC Manual (1) or the fBRIN recommendations (3,5).  The C-MIND database offers access to QA analysis pipelines to regeneration these parameters from the original QA data acquired in closest proximity to any given subject scan or over a time period specified by a users.  These parameters have been computed for all QA data received from Cincinnati and UCLA and the parameters are stored in the C-MIND database.  Any of the QA parameters selected from either site can be plotted for the duration of the study using the dashboards for QA values (https://research.cchmc.org/c-mind-db/dashboards/qa).  An example of the QA dashboard plot of Geometric Accuracy X from ACR data along with Signal Drift % and Signal Fluctuation % from fBIRN data is shown in Figure 8.2.

QA parameters can also be exported in a file that is indexed to the participant scan data so that they can be include as covariates in second level analysis of group data.

Figure 8.2: Quality Assurance dashboard for scanner level performance parameters from C-MIND database. QA data is automatically entered into the database from the QA pipelines that process FBIRN and ACR phantom data. 23 QA parameters computed and stored from each QA scan session and the date of the scan is linked to the nearest human subject scans in the database. Data from 2 scanners and 3 parameters can be displayed simultaneously. Parameters can be queried through the C-MIND database and exported along with imaging and behavioral data, into data analysis pipelines as control variables.

 

 

8.1.2. Subject level QA

Criteria were established for data certification and quality rating prior to entry the into database, using published scales 6. Quantitative ratings indicating data quality (0 to 3 scale) for each image were assigned via an online interface and stored in the database. These scales are based on visual review of each image by a trained reader using the C-MIND web interface. The Quality Review feature of the database provides image display as well as a dropdown menu to select the appropriate value for each scale. In this interface, the quality scales are at the fingertips of the reviewer to help insure consistent application of the criteria to all images. These values are then written to the database and become searchable parameters for defining the quality of images included in subsequent analyses. Image Quality Factors (QF) can be exported from a database search and saved for used as co-variates in-group analysis or other operations. Subject level Quality Scales are included in Appendix 1 for anatomical, diffusion and functional file types (includes ASL and BOLD) and the end of this chapter. Figure 8.3 shows a C-MIND database dashboard display of children in each age bracket that completed the Story Listening Task to yield images with quality ratings of 0 to 4.


Figure 8.3: Story Task image by age and quality factor.

 

 8.1.3. Data Quarantine and Release

As image data were acquired, every image was initially quarantined before being added to the release database.  Prior to release each image was reviewed and assigned an image quality factor according to a carefully designed scale that was tested against published standards. 6 Once the image quality assessment was made, image data was passed into the database along with an appropriate quality factor. Images and associated variables are now stored in the C-MIND database, and the C-MIND query interface allows users to search for images by type and quality as well as associated metadata. Figure 8.4 shows the results of a search of the database for all images from the Story Listening task with a graph indicating how many images for each age bracket and quality factor as an example of the sort of search that can be conducted.

The query interface for the C-MIND database offers countless options for searching the database for imaging, behavioral and metadata for subsequent analysis. Once completed, a search can be saved and used again later or by other users. A notable and unique feature of the C-MIND database is that it also contains all of the imaging data from the cohort and allows image data to be downloaded by a user to a local store or submitted to a selected pipeline for online analysis of the selected subset of the data. An example of a simple C-MIND query for participants with Story Listening ASL/BOLD fMRI data with quality factors of <2 is shown in Figure 8.4.  The dialog window superimposed on the lower right corner of the search window shows the “Save Query” function of the database.


Figure 8.4: C-MIND database query interface showing an example of a search for functional imaging data with quality factor < 2.  The pop-up dialog box demonstrates the ability to save a search for later use by clicking the disk icon circled in red.

Newer, more advanced analysis tools will offer further opportunities to combine data across imaging modalities and across age.  One unique advantage of the data housed in the C-MIND database is that it can serve as a control population for the study of childhood neurological disease/injury; specifically, disease/injury where the mechanism of neurological damage may be related to changes in cerebral blood flow in the developing brain.

 

8.1.4. Scan review and reporting of abnormal findings

To standardize the radiologic review of the anatomical images we developed a study classification system.  The classification system is designed to: 1. Identify pathology that could interfere with analyses that require strictly normative data, and 2. Provide a clear framework for identifying those subjects that may have medically important abnormalities.  We defined three image classifiers: A. Image classification – a parameter that classifies potential abnormalities on a scale of 0-4 described below; B. Anatomical Distortion - a binary (0/1) scale that indicates whether the pathology detected in the image causes any significant distortion of the geometry within the brain (e.g. displacement by a mass, or enlarged ventricles); and C. Follow Up - a binary (0/1) scale that indicates whether any clinical follow-up is needed (0 – no, 1 – yes).  Detailed descriptions for the numerical scales are as follows, with more detail provided in Appendix 2.

A. Image classification (0-4):

0 - Normal:  No abnormalities or anatomic variations detected

1 - Normal anatomic variants:  Anatomic variations that have no clinical significance. 

2 - Potentially significant abnormality: Imaging findings out of the range of normal or normal anatomic variation, that require correlation with clinical findings to determine their true significance for the health of the subject (if any). 

3 - Likely clinically significant abnormality:  Imaging findings out of the range of normal or normal anatomic variation that have a high likelihood of clinical significance requiring clinical and / or imaging follow-up.

4 - Imaging markedly degraded by artifact:  No interpretation possible. Imaging is so limited that no confident evaluation as to clinically significant abnormalities is possible.

B. Anatomical Distortion (0/1):

0 - No:  No significant distortion of normal anatomy on the T1 weighted images.

1 - Yes:  Possibly significant anatomic distortion on the T1 weighted images. 

C. Follow Up (0/1):

0 - No: Imaging findings do not warrant follow up with family or primary care provider.

1 - Yes: Clinical correlation with symptoms and/or potential imaging follow-up is appropriate given the possible significance of the initial findings on the research scans. 

Using these scales, all imaging findings were classified by the study neuroradiologist using the template in Table 8.5.  Note that Table 8.6 provides a summary of the frequency and types of findings that were present in the data collected for C-MIND.  Only images that have Image Classification scores of 0 or 1 AND Anatomical Distortion score of 0, are included in the release database for public access.  Images with Classification or Distortion scores greater than these values were rejected during the quarantine and evaluation process and were not included in the database for public release.

 

 

Appendix 2 includes the definitions of the classification variables in a stand-alone document that can be exported and used for other studies.  Adhering to these written guideline and definitions will facilitate future multi-site neuroimaging studies in human subjects.

Image classification variables are also stored in the online database for query and export, though in most cases the values will be 0 with an occasional 1 for Image Classification if no Anatomical Distortion was present.

In addition to the standardized image review process for research brain images, a process was established for reporting the findings to the participants through their primary healthcare provider and confirming the follow-up with research participants.  Key steps in this process are as follows:

  1. Radiologist reviews all anatomic brain imaging scans in a timely fashion after notification by study coordinator.  All studies (including planned repeat and longitudinal studies) are reviewed and reported.
  2. If a potentially clinically significant abnormality is detected (class 2 or 3), the radiologist notifies the principle investigator (PI) and/or research coordinators of the findings with a written summary and recommendations. If an urgent finding is identified, additional direct contact of the study coordinator / PI is made.
  3. PI or delegate contacts family primary healthcare provider (PCP) and explains subject’s participation and findings.
  4. The radiologist discusses interpretation with the PCP physician if desired. 
  5. PCP will follow up with family regarding findings and any referrals needed in follow up.
  6. PI will follow up with family and PCP in writing, 1-2 weeks later to insure that the above chain did not break at any point.

 

8.2. Neurobehavioral data quality control

Clinical research coordinators performing assessments are trained by a board certified pediatric neuropsychologist. This training included observation of all tests performed by the neuropsychologist or another experienced examiner on at least two occasions. After observation, each trainee performed two mock assessments with volunteers while being observed by the neuropsychologist. Additional practice assessments were arranged as necessary. To maintain their skills, each examiner was observed by the neuropsychologists for a full assessment at least once per year. Examiners scored tests on the day of administration and completed a preliminary data capture form. All protocols were reviewed by neuropsychologist on an item-by-item basis for both administration and scoring accuracy in order to ensure adherence to standardized processes and consistency across assessments. Data capture forms were be finalized after this quality control review was completed. The data was then entered into the database. Finally, a research coordinator who did not administer the tests entered the data into the database a second time (double data entry). The database certifies the neuropsych data entry as complete after the second data entry has been completed and matches the initial entry.  In the case of discrepancy between the first and second data entry, conflicts are resolved by the database manager and the coordinators with reference to the primary data sources.  These procedures have been used successfully for neurobehavioral data collection and entry into the C-MIND database.

References:

1.         ACR. MRI Quality Control Manual (2004). Vol P-MRQCM04. Reston, VA: American College of Radiology; 2004.

2.         ACR. Phantom Guidance Test. Reston, VA: American College of Radiology; 2005.

3.         Function-BIRN. Supplemental Material For JMRI's FBIRN Recommendations for Prospective Multi-Center Functional Neuroimaging Studies. Appendix IV. FBIRN Qualtiy Assurance Program. Irvine, CA: FBIRN; 2011.

4.         Friedman L, Glover GH. Reducing interscanner variability of activation in a multicenter fMRI study: controlling for signal-to-fluctuation-noise-ratio (SFNR) differences. Neuroimage. Nov 1 2006;33(2):471-481.

5.         Glover GH. FBIRN Stability phantom QA procedures:. 2008; http://www.fbirntutorials.org/, 2008.

6.         **Yuan W, Altaye M, Ret J, et al. Quantification of head motion in children during various fMRI language tasks. Hum Brain Mapp. May 2009;30(5):1481-1489.

7.         Kaiser, D, Leach, JL, Vannest, J, Schapiro, M, Holland, SK. Unanticipated Findings in Pediatric Neuroimaging Research:  Prevalence of Abnormalities and Process for Reporting and Clinical Follow-up. Brain Imaging & Behavior. 2015.

 

Appendix 1A – Image Quality Rating Scale Anatomical MRI

Quality Rating Scale – Anatomic  Data

Applies to T1, T2, and T1 Estimation images*

 

Instructions: Each dataset should be rated using the 0-3 scale below: 0=perfect and 3=poor

 

0. Perfect data, no discernable artifacts, homogeneous signal intensity across field of view and clear contrast between tissue types.

 

1. Good image quality and contrast.  Some artifacts where all slices contain a variation in intensity including variation in homogeneity or discrete artifacts such as zippers, ringing, wrap-around, etc. 

 

2. Serious artifacts caused by motion, B0 or B1 non-uniformity.  All slices contain a severe intensity variation across FOV.  Poor contrast between tissue types but discernable tissue boundaries.

 

3. Excessive image artifacts due to motion or other factors affecting all slices.  No visible tissue contrast or discernable tissue boundaries. Data should clearly be discarded.

 

*Note that for the T1 estimation images as the TI approaches the T1 values of GM and WM (800-1500 msec) the image intensity will drop sharply.  This not an artifact and images should be rated consistently with the other T1 estimation images unless severe motion is detected.

 

*These criteria are based on the image quality alone and make no reference for the intended use for normalization or VBM segmentation, which may be subjective criteria.

Appendix 1B – Image Quality Rating Scale DTI/HARDI MRI

 

Quality Rating Scale – DTI/HARDI Data

Applies to DTI and HARDI images

 

Instructions:   Each dataset should be rated using the 0-3 scale below.  DTI data is rated by viewing a movie loop of the sequential b=0 images that are spaced at regular intervals in the DTI data set.  With 68 direction DTI and HARDI data, the b=0 images occur at intervals of 11 images.

 

0. Perfect data, no discernable motion or slice drop out

 

1. Some minimal motion where some slices contain a change in intensity and/or position.

* Degree of motion does not pose a threat to the integrity of the data.

 

2. Continuous, moderate motion where all slices contain a change in intensity and position.  This might result in many or all slices dropping out entirely for an interval. Degree of motion poses a threat to the integrity of the study’s data

 

3. Continuous, excessive motion where all slices contain a change in intensity and position; data should clearly be discarded.

 

* In DTI data transient head motion causes individual slice dropout.

 

  • A quantitative approach for assessing the quality of DTI and HARDI data has also been developedThe full 61 direction DTI/HARDI data is run through DTIPrep, (http://www.nitrc.org/projects/dtiprep) to detect slice dropout artifacts, slice interlace artifacts, and/or excessive motionIf a given diffusion direction fails to meet the specified quality thresholds it will be discarded. The QA’d dataset and direction information are stored in the database in addition to the original raw HARDI/DTI data. The number of discarded directions is a searchable variable within the database, so a threshold on the number of “good” directions can be chosen before proceeding further with a group analysis.

 

  • The existing literature indicates that at least 45 well-distributed directions are needed to perform HARDI processing and at least 30 directions are optimal for standard DTI tractography.
     

Appendix 1C – Image Quality Rating Scale Functional MRI

 

Quality Rating Scale – Functional Data

Applies to ASL/BOLD, Alpha, Resting State BOLD, Baseline CBF

 

Instructions:   Each dataset should be rated using the 0-3 scale below. 

 

0. Excellent data, no discernable motion. Intensity modulation may be present in a SOME slices.

 

1. Some minimal motion where ALL slices contain a change in intensity and/or position.  Degree of motion does not pose a threat to the integrity of the data.

 

2. Continuous, moderate motion where all slices contain a change in intensity and position.  Degree of motion may pose a threat to the integrity of the data.

 

3. Continuous, excessive motion where all slices contain a change in intensity and position; data should clearly be discarded.

 

 

Appendix 2 - A classification system for neuroradiologic findings in imaging research studies with normal health children.

 

This document describes a process for classifying MRI brain studies from normal healthy control subjects in order to: 1. Identify pathology that could interfere with analyses that require strictly normative data, and 2. Provide a clear framework for identifying those subjects that may have medically important abnormalities.  There are three image classifiers: A.  Image classification – a parameter that classifies potential abnormalities on a scale of 0-4 described below;  B. Anatomical Distortion - a binary 0/1 scale that indicates whether the pathology detected in the image causes any significant distortion of the geometry within the brain (e.g. displacement by a mass, or enlarged ventricles); and C.  Follow Up - a  binary 0/1 scale that indicates whether any clinical follow-up is needed (0 – no, 1 – yes).  Detailed descriptions for the numerical scales are as follows:

 

Imaging classification:

0 - Normal:  No abnormalities or anatomic variations detected

1 - Normal anatomic variants:  Anatomic variations that have no clinical significance.  Examples would include: prominent cisterna magna, cavum septum pellucidum, cerebellar tonsillar ectopia <5mm without CSF space effacement, slight ventricular asymmetry, vascular variations. Although unlikely, some of these variants may lead to anatomic distortion.

2 - Potentially significant abnormality: Imaging findings out of the range of normal or normal anatomic variation, that require correlation with clinical findings to determine their true significance for the health of the subject (if any).  Examples would include:  Small regions of non-specific white matter signal, cerebellar tonsillar ectopia > 5mm or with CSF effacement,  callosal anomalies,  significant paranasal sinus opacification.

3 - Likely clinically significant abnormality:  Imaging findings out of the range of normal or normal anatomic variation that have a high likelihood of clinical significance requiring clinical and / or imaging follow-up.  Examples include:  Mass lesions, areas of encephalomalacia, multifocal or confluent white matter signal abnormalities, hydrocephalus, aneurysm.

4 - Imaging markedly degraded by artifact:  No interpretation possible. Imaging is so limited that no confident evaluation as to clinically significant abnormalities is possible.

 

Anatomic distortion:

0 - No:  No significant distortion of normal anatomy on the T1 weighted images.  Include: small areas of FLAIR signal abnormality not producing anatomic distortion or clear abnormality on review of T1-weighted images, normal variants in posterior fossa CSF spaces.

1 - Yes:  Possibly significant anatomic distortion on the T1 weighted images.  Examples include:  Callosal hypogenesis, visible malformation of cortical development, heterotopic gray matter, significant or localized ventricular enlargement or asymmetry, significant localized volume loss, encephalomalacia, significant white matter signal changes visible on T1 or with associated ventricular or gyral changes.

 

Follow up:

0 - No: Imaging findings do not warrant follow up with family or primary care provider.

1 - Yes: Clinical correlation with symptoms and/or potential imaging follow-up is appropriate given the possible significance of the initial findings on the research scans.  PI will follow the specified referral pathway to begin the follow up process. By definition, a study with an imaging classification of 2 or 3, requires PI notification.
 

Appendix 3 - A classification system for neuroradiologic findings in imaging research studies with normal health children.

 

Template for Radiologic  Review of Research Generated Images

 

Exam Date:   

 

Subject ID#: 

 

Imaging Classification:

           

____ 0 - Normal                                              

____ 1 - Normal anatomic variant      

____ 2 - Potentially clinically significant abnormality  

____ 3 - Likely clinically significant abnormality         

 

Findings:

 

Notes:

 

Anatomic distortion: __ Y (1)   __N (0)

 

Notes:

Follow-up:  __ Y(1)   __ N (0)

Evaluation recommendation:

 

Reported to: Holland, Scott; Mark Schapiro; Julie Franks-Henry

Date/Time Reported:

Signature:

 

Imaging classification:

0 - Normal:  No abnormalities or anatomic variations detected

1 - Normal anatomic variants:  Anatomic variations that have no clinical significance. 

2 - Potentially significant abnormality: Imaging findings out of the range of normal or normal anatomic variation, that require correlation with clinical findings to determine their true significance for the health of the subject (if any). 

3 - Likely clinically significant abnormality:  Imaging findings out of the range of normal or normal anatomic variation that have a high likelihood of clinical significance requiring clinical and / or imaging follow-up. 

Anatomic distortion:

0 - No:  No significant distortion of normal anatomy on the T1 weighted images. 

1 - Yes:  Possibly significant anatomic distortion on the T1 w

 

Back to the Index Page.