There is no one roadmap for efficient CT workflow that will succeed for all sites. Each site will need a slightly different solution based on the available equipment and the preferences of the radiologists. In the previous chapters, I have outlined how to use all of the different com- ponents involved with MDCT, from the scan acquisition parameters, to image reconstruction and review, to use of 3D workstations. While each of these components is important individually, in the end it is how they all function together that will determine if workflow is efficient and effective.
One of the most important components to efficient workflow is setting up detailed protocols and pathways for each case. Having a detailed protocol will eliminate the guesswork and minimize errors. It also saves time for both the technologist and the radiologist. If the pro- tocol covers everything that needs to be done from start to finish, the technologist does not need to waste time figuring out what to do. The radiologist will not waste time answering unnecessary questions, and when going to review a case, can be confident that the images needed will be present and in the desired format.
Comprehensive CT Protocol
A different comprehensive protocol should be created for each exami- nation type. Start by making a list of every basic examination type (neu- rology, body, MSK, CT angiography, CT perfusion) that is performed in the imaging center or hospital. Include specialized protocols such as dual phase liver, pancreas, CT urogram, CT colonoscopy etc. Each of these protocols will contain at least five different components (Table 7.1).
The first component of a comprehensive protocol contains all the parameters necessary for data acquisition, including scan mode and slice thickness, pitch, rotation speed, and dose (automatic exposure control when possible). If automatic exposure control is not available then separate protocols need to be created for adults and children of
Chapter 7
Efficient CT Workflow
83
84
AcquistionScan ModekVmAAECRotation TimeFOVSliceThickness Scan#1Helical120400-500Yes0.5 sec500mm1 to 1.5mm AcquistionScan ModekVmAAECRotation TimeFOVSliceThickness Scan #2Helical120400-500Yes0.5 sec500mm.5 to .75mm AcquistionScan ModekVmAAECRotation TimeFOVSliceThickness Scan #3Helical120400-500Yes0.5 sec500mm1 to 1.5mm Injector ValuesContrast NeededContrast VolumeSaline VolumeInjection Rate Yes125cc50cc3cc/sec Contrast TimingBolus DetectionReference LocationTrigger Value (ROI) Delay before Monitoring
Scan Delay After TriggerManualDelay Scan #2Delay Scan #3 OptionalAorta20020 sec15 secYes (O)40 sec60 sec ReconstructionsAlgorithmThin AxialsRecon IntervalThick AxialsSagittalCoronalPACSWorkstation Scan #1StandardNo5mmThick AxialsNo ReconstructionsAlgorithmThin AxialsRecon IntervalThick AxialsSagittalCoronalPACSWorkstation Scan #2Standard.5 to .75mm.5mm2mm2mmThick Axials and MPRThin Axials(O) ReconstructionsAlgorithmThin AxialsRecon IntervalThick AxialsSagittalCoronalPACSWorkstation Scan #3Standard1 to 1.5mm1mm3mm2mmThick Axials and MPRNo
Pancreas Protocol
Scan Parameters Coverage Pancreas - Pre Contrast Coverage Pancreas - Pancreatic Coverage Abdomen - Venous Phase
Chapter 7 Efficient CT Workflow 85
different weights. If a gated study is to be performed, the type of gating and desired heart rate should be included.
The second component involves contrast. When given, the amount and concentration, injection rate, and timing must be specified. When a test bolus method or bolus triggering is used, the specifics of when and how to initiate the scan must be included, as well as the vessel to be monitored. If a split or biphasic contrast injection is to be used, this must be outlined.
The third component includes information about the body part to be scanned, including coverage needed, major landmarks to include/
exclude, and whether or not multiple phases are needed. Any special- ized instructions regarding the acquisition need to be included—
for example, applying abdominal compression for a CT urogram or obtaining prone images for CT colonoscopy after gas insufflation.
The fourth component concerns the image reconstruction. This includes the slice thickness and spacing (overlapping reconstructions if desired), the field of view, the reconstruction filter (will images be reconstructed with more than one filter, i.e., soft tissue and bone?) and reconstruction algorithm (cone beam versus non–cone beam) to be used. If MPR or MIP images are generated directly on the scanner console, the plane and thickness of the reconstructions should be built directly into the protocol.
The fifth and final component addresses what happens to the images once generated. Specifics of which images are sent to PACS and the image archive are very important. If the case is to be sent to one or more workstations, this should also be built directly into the protocol. If filming of images is needed this must also be specified.
Although it sounds complicated, much of the above information can be built into the protocol library of most current MDCT scanners. This will help automate the process. The more information that can be built directly into the scanner protocol library the better. The information should also be kept in a protocol book next to the scanner for easy reference by the technologists. This will eliminate guesswork and minimize errors and interpretation delays caused using the wrong parameters or reconstructing the wrong data.
Efficient Image Review
Efficient workflow also requires optimizing how the images are viewed on the equipment available. This requires some knowledge of the capa- bilities of the network and computers that are used. Current MDCT data sets can easily reach 1000 to 2000 images and will get even bigger in the future. Networks and computers vary greatly in their ability to handle this volume of data.
For those sites that are fortunate enough to have a highly robust data network and powerful computers or workstations for image review, it is preferable to reconstruct the thinnest slice data available and send this entire data set to the review station. This gives the radiologist maximal flexibility. The data can be instantly manipulated in real time
to view either thin or thicker slices. Multiplanar reconstructions can be rapidly generated on the computer in any plane and thickness, and advanced tools such as MIP or volume rendering may be instantly available. As long as the workstation is able to load and scroll though the data in a short period of time, this is an ideal solution allowing the radiologist maximal flexibility without sacrificing efficiency. This model is clearly the future for MDCT.
Unfortunately, at this moment the vast majority of sites do not have the network and computer resources available to make submillimeter image review a viable option for routine cases. For slower networks it may take 5 minutes to 10 minutes (or even more) to transfer these large data sets from the scanner to the PACS. Many PACS workstations will take 2 minutes to 5 minutes just to fully load a large CT case into memory. Once the case is loaded, the operation of the computer may be quite slow and it could take several minutes just to scroll from one end of the data set to the other. For these sites it is crucial to decrease the number of images being sent and reviewed. If this is not done, widespread frustration with inefficiency and data overload will ensue (Figure 7.1).
By reconstructing thicker slice axial and MPR images directly on the scanner console, data sets can be reduced substantially. Depending on the specific parameters chosen, data sets for body cases and CTA can be limited to between 100 images and 300 images without significantly reducing diagnostic image quality. This is a much more manageable size for most existing computers and networks. High-quality, high-
Figure 7.1. CT workflow. Flow of image data from the scanner to the PACS system and workstation and subsequently to the radiologist and referring doctors.
Chapter 7 Efficient CT Workflow 87
resolution 2-mm to 5-mm axial images with multiplanar reconstruc- tions can be reviewed on PACS rapidly and accurately. Diagnostic accuracy will remain high without sacrificing efficiency. This approach does require frequent utilization of the 3D workstation. Cases needing volume rendering or surface rendering or MIPs should have the thin section data sent straight to the workstation for reconstruction and review. Oblique MPRs can be created either on the workstation or the scanner console by the technologist. Processed images from the workstation can then be sent back to PACS for additional review and archival. The thin submillimeter source images are not routinely reviewed but remain easily accessible when needed.
True efficiency also requires a frame shift in the minds of interpret- ing radiologists. For years the CT paradigm has dictated careful review of axial images with multiplanar correlation in select cases. This para- digm is obsolete in the multidetector era with volumetric data sets.
Radiologists familiar with MRI can easily understand this. No experi- enced MRI reader would ever feel content with axial images alone. For years MRI readers have been able to review images in whichever plane best demonstrates the anatomy and pathology, and to correlate the findings with other planes. Computed tomography should now be treated exactly the same way. Many if not most examinations are better suited to primary image review in sagittal, coronal, or even oblique planes. This is fairly obvious for examinations like the spine, sinuses and hips, but can even be true for body cases. The body is much longer than it is wide and therefore 50 2-mm coronal images can be reviewed much more quickly than 200 3-mm axial images.
Making the transition from axial image–based review to a true multiplanar approach does not occur overnight. It is a gradual process that takes weeks to months to fully integrate into daily practice. The rewards are substantial, however: more efficient image review coupled with the potential for improved diagnostic accuracy.
Selected Readings
1. Jhaveri KS, Saini S, Levine LA, et al. Effect of multislice CT technology on scanner productivity. Am J Roentgenol 2001 Oct;177(4):769–772.
2. Roos JE, Desbiolles LM, Willmann JK, Weishaupt D, Marincek B, Hilfiker PR. Multidetector-row helical CT: analysis of time management and work- flow. Eur Radiol 2002 Mar;12(3):680–685.
3. Rubin GD. Data explosion: the challenge of multidetector-row CT. Eur J Radiol 2000 Nov;36(2):74–80.
