Dynamic CT Imaging in Acute Stroke Management

SIEMENS 40.00 0.00 White Paper | siemens.com/ct-neurology-engine Dynamic CT Imaging in Acute Stroke Management Francesco Pisana Peter Schramm, MD, PhD International version. Do not distribute in the US. Answers for life. Dynamic CT Imaging in Acute Stroke Management SIEMENS DOC Server, 187,000 H & __ X SIEMENS DOC USERTONE STROKE-20 01 STROKE 20 01 CT Neuro Perfusion t: Motion Correction Segmentation !: Vessel Definition E Normalization Ficase compare the patient's perfusion mopa Findings Navigator 2 Dynamic CT Imaging in Acute Stroke Management Contents Dynamic CT Imaging in Acute Stroke Management Clinical Introduction 4 Patient Preparation 6 Contrast Injection 7 Scan and Reconstruction 8 Advanced Visualization – CT Neuro Perfusion 10 Trouble Shooting 19 Advanced Visualization – CT Dynamic Angio 23 Glossary 27 3 Dynamic CT Imaging in Acute Stroke Management Clinical Introduction [1] Astrup J. et al. Perfusion is the process by which Intracranial hemorrhage can easily be Thresholds in cerebral ischemia oxygen and nutrition are delivered to the ruled out using a non-enhanced head the ischemic penumbra. biological tissues via the vascular system. CT image. If no signs of hemorrhage – Stroke. 1981, 12: 723-725 More specifically the term perfusion are found, further dynamic imaging [2] Tomandl BF. et al. refers to blood flow, which conventionally including CT angiography and CT Comprehensive imaging represents how many milliliters of blood perfusion studies can be run (fig. 1). of ischemic stroke with enter 100 ml of tissue in 1 minute In the presence of stroke, irreversible multisection CT. Radiographics. 2003, [ml/(min*100 ml)]. neuronal damage can occur after just a 23: 565-92 Perfusion CT is now a well-established few hours. In an acute ischemic stroke, technique in stroke imaging. A stroke patient selection for thrombolytic therapy occurs when the oxygen delivery to the is based on neuroradiological findings brain is not sufficient. This can happen to a large extent. The physician has to for two reasons: weigh the risk of intracranial hemorrhage against the potential benefit of • Hemorrhage intravenous or intraarterial thrombolytic • Lack of blood flow ischemic stroke therapy. Decisions in acute stroke treatment can be facilitated by imaging 4 Dynamic CT Imaging in Acute Stroke Management Min Non-enhanced CT 2 Yes End of Exam Hemorrhage No CT Perfusion (CTP) 4 CT Angiography (CTA) CTP Evaluation 7 CTA Reconstruction 9 Possible workflow for a CT perfusion study. A non-enhanced CTA Visualization 11 CT is performed to rule out hemorrhage. Subsequently CT Initial Evaluation for Decision Making perfusion is run. Optionally a 12 Carotid CTA can also be performed. Further Evaluation (Film) Documentation Intra-cerebral angiogram can be derived from perfusion data, without additional acquisitions.[2] Fig. 1 techniques that detect not only the acute characterize functional properties of perfusion deficit including the area of the the brain tissue. For dynamic perfusion so-called “tissue at risk”, but which also imaging, a short and fast bolus of predict final infarct volume. Parameters contrast media is injected intravenously, such as cerebral blood flow and cerebral and images are continuously acquired blood volume are considered key for this before, during, and slightly after the first evaluation[1]. In general, brain tissue passage of contrast media through the presenting both a low blood flow and brain’s vessel architecture. By fitting a low blood volume is considered as the measured data into physical models, non-viable tissue: These neurons are different perfusion parameters are also called “infarct core”. Brain tissue derived. presenting low blood flow but still normal or even increased values of blood volume indicate that intravenous medication can still reach that tissue. These neurons are called penumbra or “tissue at risk”. Perfusion CT aims to 5 Dynamic CT Imaging in Acute Stroke Management Fig. 2A Fig. 2B Fig. 2C Patient Preparation The fundamental idea of Perfusion CT is Using the tiltable head holder will ensure to study the blood dynamic in each tissue greater coverage and less exposure for voxel: The change in concentration of the eye lens (fig. 2B). contrast medium during the time is directly proportional to the changes in Since the number of capillaries in tissue X-Ray attenuation. Therefore, perfusion voxels can be very low (1–10%), the parameters can be calculated for each changes in attenuation might also be voxel starting from its time-attenuation very small (5–10 HU). Noise and artifacts curve. To ensure this, the voxels should can easily mask and corrupt perfusion stay always in the same spatial position measurements. A few solutions can be during the entire acquisition time. implemented to find a balance between The first basic rule for a correct Perfusion acceptable radiation exposure and noise CT study is to instruct the patient not level. One is to position the patient, to move. Additional fixation bands can be in such a way that the head stays in used (fig. 2A). Motion can be corrected the isocenter of the gantry, where the to some extent in postprocessing with optimal SNR is achieved (fig. 2C). registration algorithms. But if too much motion is present, it might be necessary to delete data, and this could affect the results. 6 Dynamic CT Imaging in Acute Stroke Management Injection Volume Flow Contrast medium (350) 35 ml 6 ml/s Saline 40 ml 6 ml/s Table 1 Contrast Injection The contrast injection should best be short and compact. This because Recommendations regarding different models can be used to calculate injection protocol the parameters from the measured • data. All of these models make some Injection flow rate of 5–6 ml/s simplifying assumptions about the or higher outflow through the venous system. • Pre-heat the contrast medium to These assumptions are best fulfilled decrease viscosity and enable if the injection approximates an impulse high flow rates function. To achieve this the Iodine • Delivery Rate (IDR) must be at least 2 g/s. 18 g needle (green), or larger, The minimum flow rate can be calculated into cubital vein as (2000/CM concentration) [ml/s]. • Injection time lower than 8 s For example for a concentration of • 350 mg/ml, the correct flow rate should A contrast medium volume not be at least 6ml/s. Acceptable realistic too small, to ensure detectable values for clinical practice can be found enhancement changes (table 1). • Saline flush at the same flow rate • Short delay between injection and scan (~ 4 s) 7 Dynamic CT Imaging in Acute Stroke Management SOMATOM Definition SOMATOM AS, Edge and Flash Force kV 80 kV 70 kV mAs 200 mAs 200 mAs Acquisition time 45 s 45 s Cycle time 1.5 s 1.5 s Estimated dose 5.3 mSv 4.6 mSv ∼ Scan range ∼10.0 cm 11.4 cm Table 2 Scan and Reconstruction Scan length Scan parameters Traditionally, Perfusion CT scan length Default protocols are tailored to reach was limited by detector coverage. the optimal compromise between Increasing detector coverage can dose and image quality. For the head, introduce artifacts (such as cone beam typically 80 kV with 200 mAs are artifacts) that become not negligible in selected. For accurate quantification of perfusion studies and would, therefore, perfusion metrics, the best compromise need to be corrected in the reconstruc- between dose and model assumptions tion process. Siemens’ unique solution preservation is given by an acquisition “Adaptive 4D Spiral” consists of a periodic time of 45 s, with 30 images acquired and continuous movement of the table every 1.5 s (table 2). However, if in and out of the gantry. Different cycle quantification accuracy is less relevant, times and scan lengths are selectable, for example in the case of 4D CT- so that the optimal compromise Angiography the cycle time can be between coverage and dose, is achieved. decreased to 2.0 s or 2.5 s. This in Generally, in stroke imaging a scan turn increases the scan length with length of approximately 10 cm is optimal the Adaptive 4D Spiral to a maximum to cover the supratentorial brain, while of 80 cm with the SOMATOM Force. minimizing exposure to the eye lenses. Greater scan lengths can be selected if desired. 8 Dynamic CT Imaging in Acute Stroke Management Further dose reduction Reconstruction Further dose reduction is possible with Images can be reconstructed using 70 kV protocols. These only recently different kernels and thickness. By became available and work best in default, thick slices are reconstructed combination with Stellar Detectors. for perfusion evaluation: 5 mm with an Dose reduction is also possible by overlap of 3 mm, with a smooth kernel. manually changing the mAS and cycle Iterative raw data reconstructions are time. To achieve optimal accuracy in the not specifically needed, since the CNR is perfusion metrics, it is recommended to already improved by the 5 mm thickness, have a fast cycle time (1–1.5 s) at least and by further smoothing processes for the first 30–35 s of acquisition, since applied in the perfusion postprocessing. this is when blood exchange between vascular system and tissue capillaries Another reconstruction is pre-set by occurs. A fast temporal sampling will also default for, thin slices Here slices of ensure finer motion correction and a 1.5 mm thickness are reconstructed with more efficient employment of 4D noise an increment of 1 mm, with a medium reduction filters. A higher number of smooth kernel. This second series is measured data will result in a more designed to be used in dynamic angio robust fitting process. It is important evaluation, to visualize the dynamic in any case important, especially when filling of vessels, appreciate delays, using the deconvolution model, to keep collaterals, and real size of occlusions. the total acquisition time shorter then For this application, finer details are 60 s, since after that time backflow preferable to lower noise, since it’s from interstitium into the capillaries mainly a visualization tool and there might occur. are no specific models and results to be calculated. 9 Dynamic CT Imaging in Acute Stroke Management MEMEN'S CT Meuro Perfusion A B C [A] General toolbar [B] t-MIP [C] t-average Fig. 3 Advanced Visualization – CT Neuro Perfusion In syngo.via, the CT neuro perfusion • t-average: In the temporal average workflow guides the user through five image, each voxel will display the easy and semi-automatic steps. Some mean attenuation along the acquisition general tools are always available, like time (i.e. the mean value of the TAC). drawing TAC ROIs and scrolling through Since the main contrast exchange the time points (fig. 3A). happens in the gray matter, and since As soon as the images are loaded, the the averaging process dramatically following volumes are automatically reduces noise, this dataset is very created by default: useful for gray-white matter differen- tiation, and for anatomical detail • t-MIP: In the temporal MIP, each voxel visualization (fig. 3C). will display the maximum attenuation • reached during the entire acquisition Baseline: This dataset is the average of (i.e. the peak of the TAC). All vessels all the time points prior the contrast will be nicely displayed. This dataset arrival. can also be useful to check for motion artifacts (see trouble shooting session) and to correctly visualize the size of the occlusions (see CT Dynamic Angio section) (fig. 3B). 10 Dynamic CT Imaging in Acute Stroke Management SIEMENS DOC Server: 127.0.0.1 Assign 11. Read CT Neuro Perfusion - CT Neuro Perfusion 0.005 C 1: Motion Correction A B [A] Head motion correction command [B] Deletion of time points 2: Segmentation [C] Manual definition of baseline and motion correction base Fig. 4 1. Motion correction In this first step, the user can scroll In this step it is also possible to redefine through the time points to visually check the baseline. The software will auto- for motion artifacts. To run motion matically identify the time point prior to correction, a button with the picture of a the CM arrival and average them. Time head has to be clicked (fig. 4A). Running points used for the baseline are marked motion correction will automatically align in white (fig. 4C), and a baseline series is every dataset with a chosen reference automatically created. one (i.e. the motion correction basis). The blue triangle close to the slider bar identifies the time points currently used as the motion correction basis. In white A correct baseline value for every the time points used to calculate the voxel is crucial for the calculation baseline. Anatomical information is used of HU changes over time. At least for the registration and is (for the head) two time points, prior contrast a rigid transformation in 3D. media arrival in the vessels, should If some time points are still not aligned, be acquired. it is possible to delete them (fig. 4B). 11 Dynamic CT Imaging in Acute Stroke Management EMENS ET Neuro Perfusion 1: Motion Correction 2: Segmentation min 20 max: 100 HU VOO Registered data are segmented. Bone is Findings Medigator subtracted using an 3: Vessel Definition automatic algorithm. Detailed Additionally, other view of structures, such as segmentation CSF and skin, can and noise be subtracted via Findings Navigator reduction thresholds. tools. Fig. 5A Fig. 5B 2. Segmentation Once the images are registered and the baseline is created, the user can In essence 4D Noise Reduction click onto the second step. Here, two is a frequency dependent filter: complementary segmentation algorithms high spatial frequencies which do can be run. A first automatic one not contain the relevant perfusion subtracts the bone using anatomical information are averaged to information from the baseline. Second is improve the Signal to Noise Ratio threshold based segmentation and allows (SNR), while low spatial frequencies the user to specify a lower and higher containing the dynamic perfusion threshold to keep only voxels belonging information are nearly untouched to brain parenchyma. The thresholds to keep the maximal perfusion are applied to baseline HU values, so information. It is important to run that changes due to CM do not affect the 4D noise reduction only if there segmentation. The two algorithms can is no motion left in the dataset, be run independently or together. as residual motion can lead to a Noise can be reduced in the datasets worsening of motion artifacts. using 4D noise reduction filters (fig. 5). 12 Dynamic CT Imaging in Acute Stroke Management TAG ST Neuro Perfusion B [A] Normalized arterial input function in red. Reference vessel TAC in blue. The overlap between the two curves should be minimal. This is ensured through a A B correct injection protocol. C [B] If curves are not plausible, ROIs can be Gurvel Oplimination manually drawn on arteries and veins (see trouble shooting session). [C] A relative threshold has to be manually adjusted to segment all the vessels from 17 38 the dataset (in the picture . ... vessels voxels are defined as the ones enhancing more than 11% of 450 HU, i.e. more than 49.5 HU). Fig. 6A Fig. 6B 3. Vessel segmentation In this step all major vessels (marked in All vessels are defined as those voxels pink), a reference vessel ROI (blue circle, showing a maximum HU increase equal sinus sagittalis superior) and arterial or higher to a certain threshold value. input voxels (marked in red) are This threshold value can be manually automatically detected. The voxels with adjusted as a percentage of the peak the first time of CM arrival are averaged enhancement (Fig. 6CB) to form a smooth Arterial Input Function (AIF). The AIF is normalized to the reference vessel to compensate for partial volume effects in the small Voxels belonging to vessels have cerebral arteries. If needed, ROIs can be to be excluded, since blood flow manually replaced. and blood volume will have a much higher value in respect Voxels belonging to vessels have to be to parenchyma. Without vessel excluded, since Blood Flow and Blood segmentation, not only will ROI Volume will have a much higher value values be biased, but tissue respect to parenchyma. Without vessel voxels close to vessels might also segmentation, not only ROIs values will receive higher values because be biased, but also tissue voxels close to of the smoothing processes. vessels might receive higher values because of the smoothing processes. 13 Dynamic CT Imaging in Acute Stroke Management SIEMENS DOC Server: 127.0.2.1 Raad STRONE-20 01 Female CT Neuro Perfusion 54/028 STROKI 50 01 CT Neuro Perfusion 1.4 + 1: Motion Correction 2: Segmentation #: Vessel Definition 4: Normalization hetgre compare the patient's perfusion maps will Te loons below and mark the icon which Hinoël nopropriate 25 A preview of the results is given. On the right side the user has to select which of the three scenarios is more representative of the current patient. Fig. 7 4. Normalization [1] Astrup J. et al. This step is optional and can be selected Thresholds in cerebral ischemia in the configurations. The software Normalization is one method the ischemic penumbra. presents a preview of the perfusion to improve the reliability of the – Stroke. 1981, 12: 723-725 results and three different scenarios quantitative perfusion results. (left displayed hemisphere defected, Based on histogram analysis of the right displayed hemisphere defected, non-ischemic cerebral hemisphere, no defects) and suggests which one of it is possible to calibrate CBF these is the more representative of the and CBV values to physiological current patient (fig. 7). The user can reference values. This normalization confirm, or select another scenario, method enables the usage of a comparing the icons with the preview standard color representation of the perfusion results. and inter-study and inter-patient Since there are some reference values comparisons of the result images. in the literature for normal brain perfusion (60 ml/min/100 ml), and some well-accepted thresholds for infarct and penumbra[1] in the normalization step the values of the non-defected hemisphere of the patients are re-scaled to match the reference values. Also, the pathological values are re-scaled accordingly, so that comparison with the literature or inter- patient studies is possible. 14 Dynamic CT Imaging in Acute Stroke Management Penumbra Properties Restrict to the stroke hemisphere [A] Penumbra evaluation tool. Restrict to gray matter B The user can select on which parameters the mismatch Define Tissue at Risk Define Non-viable Tissue should be calculated, to identify penumbra and infarct. CBFD < CBVD < [B] It is possible to restrict visually 27 mL/100mL/min A 1.2 mL/100mL the penumbra region only to Yellow Rad the gray matter. This makes more clinical sense, since the white i Please define permanent setings in the task configuration. Apply Default Cancel matter has a normally lower perfusion value. [C] Quantitative information CT Neuro Perfusion are provided, together with relative TACs. Volume of penumbra, infarct, and Perfusion 1: Motion Correction Recuperation Fraction (i.e. the 2 Segmentation ratio between the volume of 3: Vascell Definition Penumbra and the total affected 4: Normalization volume) are automatically calculated. Fig. 8 5. Results In this step, perfusion results are In this step all parameters volume can displayed. It is possible to draw ROIs, be viewed. In fig. 9 are displayed CBF, mirror these ROIs on the contralateral CBV, MTT, TTS and TTD. The software hemisphere, and compare the values labels all the results and the model used for every parameter. A t-MIP is auto- to calculate them (D for Deconvolution, matically displayed in the top row, and M for Maximum Slope). In this step all results are aligned and synchronized quantitative and qualitative evaluation for the navigation. are possible. It is possible to draw ROIs, A penumbra evaluation tool is also display TACs and mirror the ROI to available. The user can select from which compare the values with the contralateral parameters to evaluate the mismatch hemisphere. (to distinguish between penumbra and It is indeed important to look at the infarct). Penumbra visualization can be parameters in a heuristic mode, since restricted only to the gray matter with they give different clinical information: a single click. The volume of infarct, penumbra, and the perfusion recupera- tion ratio are automatically provided by the software (fig. 8). 15 Dynamic CT Imaging in Acute Stroke Management ABIENT [A]: Penumbra evaluation tool. The user can select on which parameters the mismatch should be calculated, to identify penumbra and infarct C [B]: it is possible to restrict visually the penumbra region only to the grey matter, since the white matter has lower perfusion values. [C]: quantitative information is provided, together with relative TACs. Volume of penumbra, infarct, and Perfusion Recuperation Fraction (i.e. the ratio between the volume of Penumbra and the total affected volume) are automatically D calculated. [D]: gallery of result maps. Fig. 9a • CBF (Cerebral Blood Flow) is • MTT (Mean Transit Time) indicates how representing how fast the blood is many seconds the blood take to transit supplying that region of the brain. that specific region of the brain. A decrease in the CBF (< 20–30 ml/ • (min*100 ml)) is thus a symptom TTD (Time To Drain) indicates how of tissue at risk (or Penumbra). many second the blood needs to arrive and transit through that specific • CBV (Cerebral Blood Volume) is region of the brain. It is defined as indicating if the blood is still reaching TTD=TTS+MTT and gives a combined that region of the brain. A small value overview of perfusion defects (low of CBV (< 1–2 ml/100 ml) indicates MTT), and a delay defect (for example that capillaries in that region are not normal MTT but high TTS). For this reached by blood. In this case the reason there is an increasing interest in tissue is considered to be no longer this time parameter. Additionally the saveble. TTD maps look usually smoother, since • TTS (Time To Start) indicates how many they combine information from two second the blood takes to reach that parameters. region of the brain. Perfusion defects For all the other parameters description reflect normally in an increase in time refer to the Glossary section. maps, and a decrease in CBF and/or CBV. TTS reflects the if there is a delay in the contrast arrival in those specific voxels. 16 Dynamic CT Imaging in Acute Stroke Management GT Neuro Perfusion t-Avg t-Avg t-Avg All parameters volumes are shown in the results Perturion Evaluation step. ROIs can be drawn t-MIP CBF CBV and mean values and st. dev can be displayed for Findings Navigator each ROI in each volume. It is also possible to mirror the ROI, to compare the D values between the affected hemisphere and MTT TTD TTS the healthy one. Fig. 9b 6. Saving results 7. Perfusion evaluation Available results are visualized in the Once the study is completed, the results image gallery and check marks indicate are automatically sent to the PACS and the maps that will be automatically saved stored on the syngo.via server. The and sent to the PACS after the workflow results can be reevaluated in the task is completed (fig. 9D). It is possible to Perfusion Evaluation. This allows a simple save the results in different formats and efficient comparison with prior and (CT Greyscale, Enhanced CT, Color RGB). follow-up results. Also the t-MIP, Baseline, 4D data (once they have been registered/segmented/ filtered) can be saved as results. 17 Dynamic CT Imaging in Acute Stroke Management 8. Configuration and underlying models [1] Abels B. et al. Two different calcultion models can be accurate calculation of the perfusion Perfusion CT in acute ischemic selected in the configuration panel. It is metrics. The most widely used acquisition stroke: a qualitative and recommended to select only parameters protocol consist of acquiring 30 images quantitative comparison of from one model.[1] every 1.5 s over 45 s. deconvolution and maximum slope approach. Deconvolution model Maximum Slope model AJNR. 2010, 31: 1690-8 The Deconvolution model is the most The Maximum Slope model can be advanced and therefore recommended employed when few data are available. for default use. The completeness of For example, in case of late data deleted parameters, together with the short due to the presence of motion or acquisition time, and the robustness other artifacts, or, in case of very short of the assumptions make this model acquisition times (~ 30 s). In such cases preferable over the Maximum Slope this model might be able to salvage a model in most cases. All perfusion dataset where the Deconvolution model parameters can be obtained: CBF, CBV, would produce sub-optimal results. The MTT, Tmax and FE. For best results a Maximum Slope model is very robust fast injection is recommended (flow rate with respect to noise and artifacts, higher than 5 ml/s and injection time since only some data are used for the lower than 8 s). The scan time doesn’t parameters calculation. It is nevertheless have to be longer than 45–60 s. important to utilize data up to the tissue A fast cycle time (smaller than 2 s) is peak, since those are the data required recommended. This ensures enough by this model. Its limitations include that time points to fit into the model for MTT and FE cannot be calculated, and the 18 Dynamic CT Imaging in Acute Stroke Management MIMENT DOS CT Meuro Pertusion - VOO Findinga Newsgator Too few vessels segmented. The relative threshold is too high. Some arteries might be segmented out, so the red curve becomes closer to the blue one. Fig. 10A Trouble Shooting 1. Motion correction assumptions of this model require even • It is highly recommended to always run more strictly a fast injection (flow rate motion correction, even if there is only not less than 6 ml/s). For this reason little motion in the dataset. the Deconvolution model is preferable • in cases with a suboptimal injection A quick and fast way to check for protocol. motion is to look at the t-MIP and see if there are double contours. • If after motion correction there are still time points misaligned, it is possible to delete them manually. Care should be taken when deleting data before the peak of the tissue. It is possible also to delete all the late data (if considerable motion occurs at the end of the examination, and it cannot be corrected): Be careful when deleting data before tissue peak because quantitative numbers may be corrupted. • Also the baseline could be redefined in this step. One way to do that is to draw an ROI in an artery, and define the baseline as the time points prior to the contrast arrival time. 19 Dynamic CT Imaging in Acute Stroke Management CT Meuro Perfusion The optimal relative threshold is chosen. All vessels are segmented in purple, but tissue is preserved. The red and the blue curves are more separated. Fig. 10B 2. Segmentation 3. Vessel segmentation • 4D noise reduction has to be avoided • One of the most important factors in if considerable motion remains after perfusion software is the AIF (Arterial motion correction. Input Function), which is the red- To have the greatest benefit from colored TAC. It should have a smooth • 4D noise reduction filters, cycle time bell shape, high and narrow. The should not be too long (usually not overlap between the red and the blue higher than 1.5 s). curve should be minimal. However, if the curves don’t look plausible or • If automated bone removal results are acceptable, they can be manually not satisfactory, the user might choose redefined. to refine manually. This by using HU- • based segmentation and adapting the The enhancement of the automatically thresholds. found reference ROI should be sufficient (as a rule of thumb at least above 200 HU, better above 300 HU). If this is not the case (and if the reason is not a wrong injection protocol), the reference vessel ROI can be manually placed somewhere else to reach a sufficient enhancement; in most cases a placement within the sinus sagittalis gives best enhancement (thick vessel eliminating partial volume effects). Check that the reference ROI does not cover bone tissue. 20 Dynamic CT Imaging in Acute Stroke Management THE Too many vessels are .... segmented. The relative threshold is too low. Tissue voxels are excluded because of oversegmented vessels. The red and blue curves visually overlap too much. Fig. 10C • The peak of the red arterial curve the more consistent sampling time is should lie in front of the peak of the reached. To minimize the effects of blue venous curve. The red voxels partial volumes, ROI should be drawn which are used to define the AIF should in a slice where the vessel path is be correctly placed: they should not lie perpendicular to the scan direction (the within the bone and should not be vessel lumen should look like a circle). placed at the edge of segmented • volume. If this is not the case, one Once the reference vessel is correctly reason could be a wrong reference identified, and the relative threshold vessel segmentation, or a wrong is correctly set, all vessels are threshold value. If too many voxels are segmented in purple. Among these, excluded (i.e. colored in pink), then the the software will highlight in red the relative threshold should be increased. ones presenting the earliest time of Inversely, to exclude more voxels, the arrival of contrast media, and average relative threshold should be lowered them together to recalculate an (fig. 10). optimal arterial input function. • To place an ROI in an artery, it can It as well possible to manually select • help to use the t-MIP. Care should be a slice where the arteries have to be taken to minimize the thickness of the identified. With one button (labeled t-MIP, so to avoid partial volumes. “auto” button) the software will check The optimal place to put ROIs is in the all the voxel with earliest enhancement middle level of the scan range, where in the selected slice. 21 Dynamic CT Imaging in Acute Stroke Management • If still needed, an artery can be • Having the correct vessels exclusion is manually segmented to re-trace the also helping to improve the AIF. If too AIF. To do this is preferable to use many or too few vessels are segmen- the t-MIP with the lowest possible ted, the AIF might move closer to the thickness, and to identify an artery reference vessel TAC (blue curve). perpendicular to the axial plane and in The optimal relative thresholds can the middle level of acquisition range. be reached when the main vessels are colored in purple on the t-MIP, the • The reference vessel TAC is needed to re-scale the AIF. If the peak is too low, tissue voxels are preserved, and the both the AIF and the vessels segmen- red and blue TACs are more separated tation will be affected. It is worthy in (fig. 10). this case to re-draw also the reference • If the injection was not fast and sharp vessel ROI. The optimal level to place enough, considerable overlapping it is in the sagittal sinus, preferably in between the red and blue curve might the middle level of the scan range. occur. In this case it’s preferable to use the deconvolution model for post- processing. • If the injection protocol is correct and late data present motion artifacts, it’s preferable to use the maximum slope model. 22 Dynamic CT Imaging in Acute Stroke Management CT dynamic angiography Conventional CT angiography (single phase) The size of a vessel occlusion can be more reliably measured The size of an occlusion highly depends on the timing of the on a t-MIP (available only with dynamic acquisition), since the CT angiography acquisition. The previous vessel below the previous part of the vessel both before and after the occlusion occlusion is filled with some delay and it is usually not or little are highlighted. enhanced in a conventional CT angiography. This leads to an overestimation of occlusion size. With dynamic acquisition it is possible to visually Delays are not helpful in a conventional CT angiography. appreciate delays in the circulation of the scanned district. Retrograde vessel filling can be visualized. Retrograde filling can hardly be visualized. CT dynamic angiography permits visualization Collateral fillings can hardly be visualized in conventional of delayed collateral fillings in case of arterial occlusion. CT angiography. Venous circulation (e.g. venous thrombus) can In conventional CT angiography, acquisition timing is such be evaluated with CT dynamic angiography. that only arteries are enhanced. Veins are bigger and with a very high enhancement and this can sometimes make it more difficult to read the arteries. Veins and arteries can both be enhanced in a conventional CT angiography at the expense of using more contrast media. Incomplete occlusions can be more easily identified, since it is In a conventional CTA it is difficult (and strictly phase- possible to visualize if and how fast the distal portion of the dependent) to classify complete and incomplete occlusions occluded vessel is enhanced. Table 3 Advanced Visualization – CT Dynamic Angio CT Perfusion quantitatively evaluates The CT Dynamic Angio workflow is perfusion parameters based on a dynamic independent from CT Neuroperfusion. scan. Once a dynamic acquisition is For this reason, images can be opened obtained, it is possible to use the images without any specific pre-processing. not only for quantification, but also for The same registration, segmentation visualization of dynamic filling of vessels. and noise reduction tool as in CT Thin slices images can be reconstructed Neuroperfusion are available. Similarly from the same perfusion acquisition to CT Neuroperfusion, t-Average. t-MIP without any additional radiation, and and Baseline volumes are automatically can be loaded into a specific syngo.via created. workflow called “CT Dynamic Angio”. This workflow allows for dynamic visualization, ROI measurements, definition of contrast phases, and video recording. For example movies can be created that visualize arterior and venous flow. Thin slices (1.5 mm) are preferred, to see also small vessels without a spatial blurring. 23 Dynamic CT Imaging in Acute Stroke Management CT Dynamic Angio - CT Dynamic Angio 0.00% 1: Motion Correction 2: Segmentation MM Reading Fig. 11 1. Motion correction 2. Segmentation Three different motion correction The same algorithms as in CT neuro algorithms are available: perfusion are available. One algorithm • A rigid one for head studies removes the bone using bone mask subtraction. The second one is removing • A deformable one for the body all voxels with a baseline attenuation • A deformable one optimized for heart value between user-defined thresholds. In this step, it is also possible to define the reference time point to use for One can immediately toggle between the motion correction, to define the results of the different algorithms. This baseline, and to delete time points makes it very quick and easy for the user (fig. 11). to visually decide if to run only one or both of them. Similar to CT neuro perfusion, a 4D noise reduction filter can be run (see CT neuro perfusion section). 24 Dynamic CT Imaging in Acute Stroke Management CT Dynamic Angio ¥ CT Dynamic Angio A B C + [A] Movie functionality 0.005 [B] TAC ROI creation [C] Volume creation Fig. 12 Fig. 13 Movie functionality A benefit of CT Dynamic Angio is the Moreover, it is always possible during possibility to play a movie of the blood the movie to interactively change the circulation in the scanned district orientation. The VRT preset is also (fig. 12A). Movies can be paused, and editable: Advanced tools are provided played at different speeds (fig. 13). to create new VRT layouts (for example The image type of the movie can be one if the user also wishes to visualize of the following: also the bone structure with some transparency) (fig. 14). • Thin MPR, thickness is definable • Thin MIP, thickness is definable Different cropping tools are available MIP to hide parts of the volume. • • Thin VRT • VRT 25 Dynamic CT Imaging in Acute Stroke Management R. F MIF @ VAerag Ö IMP | Create 11.77% 20.01¢ ...... Fig. 14 Fig. 15 Volume creation CT Dynamic Angio also offers the The result will be a t-MIP of the arterial possibility to create temporal MIP, a phase. temporal average volume using defined intervals of time (fig. 12C). T-average images can be useful to reduce the noise, better gray-white matter Optimized t-MIPs can be helpful to depict differentiation. For t-average images, vessels enhancing in a specific time time points from the arteries’ peak interval. For example, an arterial phase until the end of the examination are (similar to a conventional CT Angio MIP) normally used. can be created in this way (fig. 15): • Drawing a TAC ROI in an artery • Drawing a TAC ROI in a vein • Creating a t-MIP using the time point where the venous enhancement is still low 26 Dynamic CT Imaging in Acute Stroke Management Glossary AC-PC Line Anterior commissure (AC) – posterior MTT Mean Transit Time: time needed by a commissure (PC) line has been adopted as a convenient standard by the neuroimaging [S] theoretical unit of contrast media to transit through one specific voxel community NVT Non Viable Tissue: Tissue presenting CBF Cerebral Blood Flow: conventionally measured in [ml/(min*100 ml)] represents [cm3] CBF and CBV values lower than define [ml (min*100 ml)] thresholds. It is considered dead infarcted the percentage of tissue voxel filled by tissue, that can no longer be saved arterial blood in 1 minute. If the value is higher than 100 ml/(min*100 ml) it means Perm Permeability: equivalent to Ktrans and to that, in that voxel, a complete filling of FE. It is measured in [ml/(min*100 ml)], and tissue capillaries takes less than 1 minute represents the transfer rate of blood from the capillaries to the interstitum CBV Cerebral Blood Volume: conventionally PPR Perfusion Recuperation Ratio: percentage [ml*100 ml] measured in [ml/(min*100 ml)] represents the percentage of tissue voxel occupied by [-] of how much tissue can still be saved in the blood entire ischemic volume. It is defined as PPR = TAR/(TAR+NVT) CNR Contrast to Noise Ratio TACs Time Attenuation Curves: Graphs showing [-] [HU] the ROIs changes in mean HU value over time DC De-convolution: mathematical process by [ml (min*100 ml)] which the voxel Impulse Response Function TAR Tissue at Risk or penumbra: volume of is calculated. The IRF represents the [cm3] tissue that potentially can be saved. Tissue distribution over time of a theoretical presenting low values of CBF, but still impulsive bolus of contrast in one specific acceptable values of CBV. voxel. IRF parameters fully characterize that voxel TAvg Temporal Average Series: Average over all [HU] time points E Extraction Fraction: fraction of a theoretical unit of contrast media that remains trapped in the interstitum before flowing out Tmax Time needed by a theoretical unit of contrast media to reach the maximum concentration through the venous system [S] in one specific voxel. Mathematically defined EEspace Extravascular Extracellular Space or as TMax = TTS + MTT/2. It reflects the transit interstitum time to the center of the IRF (ideal impulsive bolus response function) at the voxel FE Flow Extraction Product: Product between location. TMax is the sum of the arteries’ [ml/(ml*100 ml)] Blood Flow and Extraction Fraction. It is bolus delay and the tissue transit time to the equivalent to capillaries, permeability. center of the voxel. It is measured in [ml/(min*100 ml)] and represents the transfer rate of blood from TMIP Temporal MIP Series: Maximum the capillaries to the interstitum [HU] enhancement over all time points IRF Impulse Response Function: distribution Time to Drain: time of earliest washout of over time of a theoretical impulsive bolus of TTD [-] contrast in one specific voxel IDR [g/s] Iodine [S] the contrast medium Delivery rate. For CT perfusion imaging a IDR of higher than 2 g/s is required TTP Time to Peak: time measured from the start [S] of enhancement in the artery to the local Ktrans Permeability: equivalent to Perm and to FE. perfusion peak It is measured in [ml/(min*100 ml)] and represents the transfer rate of blood from TTS Time to Start: time measured from the start the capillaries to the interstitum [S] of the enhancement in the artery to the local perfusion onset MS Maximum Slope: represents the fastest increase of one specific voxel CT value due to contrast arrival. In the MS model, CBF is calculated as the ratio between the fastest increase in HU value of one voxel and the artery peak enhancement 27 In the event that upgrades require FDA The information in this document clearance, Siemens cannot predict contains general technical descriptions whether or when the FDA will issue of specifications and options as well as its clearance. Therefore, if regulatory standard and optional features which clearance is obtained and is applicable do not always have to be present in to this package, it will be made available individual cases. according to the terms of this offer. Siemens reserves the right to modify the On account of certain regional limitations design, packaging, specifications, and of sales rights and service availability, options described herein without prior we cannot guarantee that all products notice. Please contact your local Siemens included in this brochure are available sales representative for the most current through the Siemens sales organization information. worldwide. Availability and packaging may vary by country and are subject Note: Any technical data contained in to change without prior notice. Some/All this document may vary within defined of the features and products described tolerances. Original images always herein may not be available in the lose a certain amount of detail when United States. reproduced. Please find fitting accessories: www.siemens.com/medical-accessories Global Business Unit Siemens AG Medical Solutions Computed Tomography & Radiation Oncology Siemensstr. 1 DE-91301 Forchheim Germany Phone: +49 9191 18 0 Fax: +49 9191 18 9998 Global Siemens Headquarters Global Siemens Legal Manufacturer Siemens AG Healthcare Headquarters Siemens AG Wittelsbacherplatz 2 Siemens AG Wittelsbacherplatz 2 80333 Muenchen Healthcare Sector DE-80333 Muenchen Germany Henkestraße 127 Germany 91052 Erlangen Germany Phone: +49 9131 84-0 www.siemens.com/healthcare Order No. A91CT-00272-02C1-7600 | Printed in Germany | CC CR 2143 08141. | © 08.2014, Siemens AG www.siemens.com/healthcare