Perfusion CT in Acute Ischemic Stroke: A Qualitative and Quantitative Comparison of Deconvolution and Maximum Slope Approach

B. Abels; E. Klotz; B.F. Tomandl; S.P. Kloska; M.M. Lell

doi:10.3174/ajnr.A2151

Abstract

BACKGROUND AND PURPOSE: PCT postprocessing commonly uses either the MS or a variant of the DC approach for modeling of voxel-based time-attenuation curves. There is an ongoing discussion about the respective merits and limitations of both methods, frequently on the basis of theoretic reasoning or simulated data. We performed a qualitative and quantitative comparison of DC and MS by using identical source datasets and preprocessing parameters.

MATERIALS AND METHODS: From the PCT data of 50 patients with acute ischemic stroke, color maps of CBF, CBV, and various temporal parameters were calculated with software implementing both DC and MS algorithms. Color maps were qualitatively categorized. Quantitative region-of-interest–based measurements were made in nonischemic GM and WM, suspected penumbra, and suspected infarction core. Qualitative results, quantitative results, and PCT lesion sizes from DC and MS were statistically compared.

RESULTS: CBF and CBV color maps based on DC and MS were of comparably high quality. Quantitative CBF and CBV values calculated by DC and MS were within the same range in nonischemic regions. In suspected penumbra regions, average CBF_DC was lower than CBF_MS. In suspected infarction core regions, average CBV_DC was similar to CBV_MS. Using adapted tissue-at-risk/nonviable-tissue thresholds, we found excellent correlation of DC and MS lesion sizes.

CONCLUSIONS: DC and MS yielded comparable qualitative and quantitative results. Lesion sizes indicated by DC and MS showed excellent agreement when using adapted thresholds. In all cases, the same therapy decision would have been made.

Abbreviations

AIF: arterial input function
ASPECTS: Alberta Stroke Program Early CT Score
CBF: cerebral blood flow
CBV: cerebral blood volume
DC: deconvolution
GM: gray matter
LMS: least mean square
MIP: maximum intensity projection
MS: maximum slope
MTT: mean transit time
NVT: nonviable tissue
PCA: posterior cerebral artery
PCT: perfusion CT
rCBF: relative cerebral blood flow
rCBV: relative cerebral blood volume
rMTT: relative mean transit time
rTTD: relative time to drain
rTTP: relative time to peak
ROI: region of interest
SNR: signal-to-noise ratio
SVD: singular value decomposition
TAC: time attenuation curve
TAR: tissue at risk
TTD: time to drain
TTP: time to peak
TTS: time to start
WM: white matter

PCT is a well-established technique in acute stroke imaging. PCT has been proved superior to noncontrast CT^1–3 and equivalent to MR imaging in terms of infarct and penumbra characterization and selection for reperfusion therapy.^4–6 Quantitative perfusion parameters such as CBF, CBV, TTP, MTT, and TTD can be calculated and displayed as color maps. Indications for reperfusion therapy may be based visually on a mismatch of CBF and CBV and/or quantitatively on thresholds defining TAR (penumbra) and NVT^7–9—possibly even in an extended time window.^10,11

Commercial PCT postprocessing software currently available is based on a variety of implementations of DC or MS models. It has recently been shown that quantitative values as well as color maps can be significantly different among software/algorithms, even when using identical source data.¹² In terms of promoting PCT standardization,^13,14 evaluating quantitative and qualitative differences of PCT algorithms is of great importance. The objective of our study was to determine quantitative as well as qualitative comparability of delay-insensitive DC and MS algorithms by systematically analyzing corresponding color maps as well as region-of interest–based measurements (CBF, CBV, and temporal parameters) of both methods.

Materials and Methods

Datasets

We retrospectively enrolled in our study 50 PCT source datasets of patients with acute onset of neurologic symptoms in ≤9 hours. These datasets were acquired on multisection CT scanners after administration of a compact contrast bolus with an injection rate of at least 5 mL/s via power-injector through an 18-gauge intravenous line, by using the following scanning parameters: section thickness, 10–12 mm; tube voltage, 80 kV(peak); tube current, 180–235 mAs; scanning delay, 4 seconds; scanning time, 40 seconds; and temporal resolution, 1 second. Two sections at the level of the basal ganglia were imaged. In the final evaluation, only the section demonstrating the largest extent of ischemia was selected. Datasets without any hemodynamic disturbance as well as datasets with severe motion artifacts or inadequate contrast bolus were excluded from the analysis.

Postprocessing

Calculation of perfusion parameters and color maps was performed with a commercial perfusion package (VPCT-Neuro; Siemens Healthcare, Forchheim, Germany) containing implementations of a DC model and an MS model as the final step of an otherwise identical preprocessing. Both algorithms are explained in the On-line Appendix. For each dataset, processing was done semiautomatically with default settings as used in routine clinical work. Brain parenchyma was first isolated by skull bone contour findings; CSF and calcifications were removed by thresholding. The AIF was measured in suitable branches of the middle or anterior cerebral artery; the venous outflow (and the maximum enhancement reference) was taken from the superior sagittal sinus, both by semiautomatic vessel detection. Major vessels were removed by relative thresholding applied to the maximum voxel enhancement. Standard relative thresholds were 8% of the maximum enhancement in the superior sagittal sinus. Finally, an adaptive spatial filtering was performed that did not smooth over edges and vessel borders. All settings were completely identical for DC and MS. All quantitative values were determined without including larger vascular structures.

Qualitative Analysis

Color maps were visually analyzed and rated by 2 board-certified senior radiologists/neuroradiologists in a consensus reading by using the following scoring system: quality of GM/WM differentiation (for CBF and CBV maps only), grading (for temporal parameter maps only), differentiation of ischemic and normal tissue, homogeneity (contrast, contours, coherency/dissemination of ischemic lesions), and compensation of artifacts were each rated on a 3-point scale (minimum, 0; maximum, 2 points). The scores of the 4 categories were added, resulting in a maximum score of 8. Examinations with a score of >4 points were considered to be of high quality and high diagnostic value (category I); those with scores of ≤4 points and ≥2 points were considered to be of medium quality but still sufficient for diagnosis (category II). Those with a score <2 points were considered poor quality and insufficient for diagnosis (category III).

Quantitative Analysis

To retrieve corresponding quantitative measurements from different healthy and ischemic regions of the brain, 4 different regions of interest were contoured manually (freehand) in each case: In the nonischemic hemisphere, cortical GM and subcortical WM were outlined on the MIP, yielding GM and WM regions of interest, respectively. In the ischemic hemisphere, the GM region with significantly reduced CBF (rCBF, ≤60% of contralateral regions, which corresponded to an absolute threshold of ∼34 mL/100 g/min on the CBF_MS) but without a significant decrease in CBV was outlined as “suspected penumbra”; this rCBF threshold is in accordance with TAR values reported in the literature for MS^7,15,16 and DC.^4,8,17,18 The GM region with significantly reduced CBF and significantly reduced CBV (rCBV, ≤40% of contralateral regions, which corresponded to an absolute threshold of ∼1.2 mL/100 g on the CBV_MS maps) was outlined as “suspected core”; this rCBV threshold is in accordance with NVT values reported in the literature for MS^7,16 and DC.^8,9,19 All regions of interest served for retrieving quantitative measurements for the quantitative comparison of corresponding DC and MS perfusion parameters only; the regions were not meant to delineate final tissue outcome.

PCT Lesion Sizes

Finally, to examine whether there would have been a difference in treatment decisions, we measured PCT lesion sizes for DC versus MS. NVT, TAR, and NVT + TAR lesions were determined by applying adapted absolute CBF/CBV thresholds. Because our results indicated that the absolute CBF_DC threshold for defining TAR is lower than the absolute CBF_MS threshold, 27 mL/100 g/min and 34 mL/100 g/min, respectively, were selected. CBV_DC and CBV_MS thresholds for defining NVT were both 1.2 mL/100 g.

In addition to measuring DC and MS lesion sizes, PCT lesion extents were compared by determining PCT lesion scores in orientation to the ASPECTS used in prior studies for PCT lesion assessment.²⁰ In each case, corresponding NVT lesion scores (indicated by CBV_DC and CBV_MS maps, respectively) and NVT + TAR lesion scores (the whole ischemic area, indicated by CBF_DC and CBF_MS maps, respectively) were compared. To account for only 1 section per examination, we used the following modified 10-point score: M1, M2, M3, insular ribbon, caudate nucleus, lentiform nucleus, internal capsule, anterior cerebral artery, PCA/occipital lobe, and PCA/thalamus were attributed 1 point each if nonaffected and zero if affected.

Statistical Analysis

Statistical analysis was performed with MedCalc 10 (MedCalc Software, Mariakerke, Belgium) and R 2.9.2 (http://www.r-project.org). Qualitative scores were compared by using the Wilcoxon signed rank test. Quantitative region-of-interest values calculated with DC and MS were compared by using descriptive statistics and Bland-Altman plots; median method differences and the relationship between the method difference (DC − MS) and the average size of measurements (DC + MS)/2 were examined separately for all regions of interest with Bland-Altman plots and Wilcoxon signed rank tests. The temporal parameters MTT, TTD, and TTP refer to quantities that are differently defined and thus cannot be compared directly; therefore, only correlation was determined. To discriminate between negligible minor method differences and major method differences that are potentially relevant in terms of affecting lesion extent on PCT color maps diagnosis respectively, limits of agreements were set. Differences exceeding 0 ± 2 SD of (DC + MS)/2 were considered as potentially relevant. NVT, TAR, and NVT + TAR lesion sizes depicted on DC and MS maps were compared by using scatterplots with regression lines. Likewise, PCT lesion scores (modified ASPECTS) of DC and MS were compared.

Results

Of 50 datasets, 8 had to be excluded from the analysis because they did not meet the criteria defined in the “Materials and Methods” section. The mean age of the remaining 42 patients (n = 23 men, n = 19 women) was 63.8 years (range, 44–84 years). The average computational time on a standard workstation needed to generate all color maps of 1 section was ∼2 seconds for both DC and MS.

Qualitative Results

An overview of the qualitative results is given in Table 1 and Fig 1.

View this table:

Table 1:

Results of the qualitative analysis^a

Fig 1.

Overview of the qualitative results of 42 examinations. Detailed information on the quality aspects is provided by the diagrams on the right.

On the MTT maps, ischemic areas appeared less well-defined compared with TTD and TTP. Differences were statistically significant for MTT versus TTD and MTT versus TTP (both P < .0001). The average performances (mean quality score) of TTD and TTP were high, with TTD being slightly superior to TTP (P < .0001).

The average performances of CBF_DC and CBF_MS as well as CBV_DC and CBV_MS maps were high. Differences of CBF_DC and CBF_MS as well as CBV_DC and CBV_MS were not statistically significant (P = .0635 and P = .0968, respectively). In some source datasets (n = 7) with poor SNR and TAC quality, however, CBF and CBV maps calculated with DC were superior to those calculated with MS, due to higher homogeneity and fewer artifacts.

Examples of corresponding maps are shown in Figs 2 and 3.

Fig 2.

Color maps of a 63-year-old man with right middle cerebral artery infarction. CBF_DC and CBV_DC as well as CBF_MS and CBV_MS color maps are of high quality. The ischemic lesion is more clearly delineated on TTD and TTP maps compared with MTT. TTD provides the highest contrast of all temporal parameter maps.

Fig 3.

Examinations with lower source-data quality. CBF_MS and CBV_MS maps are of reduced quality (category II, see rectangles). Whereas TTP and MTT quality is markedly reduced, TTD quality is good. The case was still considered to be interpretable with both DC and MS.

Quantitative Results

The results of quantitative measurements in all regions of interest are shown in Table 2.

View this table:

Table 2:

Results of the quantitative analysis

In GM/WM regions of interest, we found good agreement of CBF_DC and CBF_MS as well as CBV_DC and CBV_MS. The SDs of CBF_DC and CBV_DC were markedly higher compared with CBF_MS and CBV_MS. Although method differences reached statistical significance (CBF_DC versus CBF_MS: P < .0001 [GM], P = .0074 [WM]; CBV_DC versus CBV_MS: P < .0041 [GM], P = .2143 [WM]), differences were not considered relevant in most cases: CBF and CBV differences were <2 SDs in 95% and 93%, respectively. Regarding the temporal parameters, we found good correlation between MTT and TTD (r = 0.82) as well as fair correlations between MTT and TTP (r = 0.59) and TTD and TTP (r = 0.56) (all P < .0001).

In suspected penumbra regions of interest, mean CBF_DC was significantly lower than CBF_MS (absolute, 29 versus 32 mL/100 g/min, P < .0001; relative, 0.48 versus 0.59, P < .0001). Method differences were >2 SDs in 17%. This corresponded to visual findings of ischemic lesions tending to be larger on CBF_DC maps compared with CBF_MS maps when using identical thresholds for DC and MS. Mean CBV_DC was significantly higher than CBV_MS (absolute, 2.9 versus 2.6 mL/100 g, P < .0001; relative, 0.92 versus 0.80, P = .0025). Method differences were >2 SDs in 26%. The SD of CBV_DC was markedly higher compared with that of CBV_MS. We found good correlation of MTT and TTD (r = 0.91), TTD and TTP (r = 0.89), as well as fair correlation of MTT and TTP (r = 0.76) (all P < .0001).

In suspected infarction core regions of interest, mean CBV_DC was significantly higher than mean CBV_MS (absolute, 1.1 versus 0.9 mL/100 g/min, P = .0011; relative, 0.37 versus 0.30, P = .0001). Method differences were >2 SDs in 12%. This corresponded to visual findings of ischemic lesions tending to be slightly smaller or similar on CBV_DC maps compared with CBV_MS maps when using identical thresholds for DC and MS. Mean CBF_DC was slightly higher than mean CBF_MS (absolute, 12 versus 9 mL/100 g/min, P = .0025; relative, 0.21 versus 0.18, P < .0001). Method differences were >2 SDs in 10%. We found fair correlation for MTT versus TTD (r = 0.65, P < .0001) and TTD versus TTP (r = 0.66, P < .0001), and weak correlation for MTT versus TTP (r = 0.30, P = .05).

Statistical plots of CBF and CBV in suspected penumbra and suspected core regions of interest are shown in Fig 4.

Fig 4.

Box-and-whisker and Bland-Altman plots demonstrate that rCBF_DC is ≤rCBF_MS in suspected penumbra (TAR) regions of interest and rCBV_DC is ≥rCBV_MS in suspected core (NVT) regions of interest. While there is no slope (no significant relationship between method difference and size of measurement) for CBF, for CBV, there is a positive slope in all regions of interest (proportional difference).

PCT Lesion Sizes

Using the CBF/CBV mismatch approach with adapted thresholds, we found excellent agreement of corresponding NVT, TAR, and NVT + TAR lesion sizes for DC versus MS. Correlation was r = 0.999 (P < .0001), r = 0.995 (P < .0001), and r = 0.998 (P < .0001), respectively. NVT/TAR ratio (CBV/CBF mismatch) was <.50 in 38% (16 cases) for both DC and MS. Detailed results are shown in Fig 5.

Fig 5.

Comparison of PCT lesion sizes (DC versus MS). All units are in cubic centimeters. Scatterplots show excellent agreement: NVT_DC = 1.03 × NVT_MS (R² = 0.998), TAR_DC = 1.02 × TAR_MS (R² = 0.989), and (NVT_DC + TAR_DC) = 1.03 × (NVT_MS + TAR_MS) (R² = 0.997). On the lower right, screenshots of 2 cases are shown as examples. Red indicates NVT; yellow, TAR.

Regarding PCT lesion scores (modified ASPECTS), statistical analysis revealed no significant difference for DC versus MS. PCT lesion scores were almost identical for DC and MS. Correlation was r = 0.998 (P < .0001) and r = 0.997 (P < .0001) for NVT and NVT + TAR lesion scores, respectively. NVT and NVT + TAR lesion scores were >7 in 67% (28 cases) and 33% (14 cases), respectively, for both DC and MS. An overview of PCT lesion scores is given in Table 3.

View this table:

Table 3:

Results of PCT lesion scores^a

Discussion

When one compares quantitative perfusion measurements, their accuracy is, in general, limited because PCT values are dependent on a number of factors such as tube settings (current, voltage, thickness, scanning time), contrast delivery (iodine concentration, flow, maximum bolus height, bolus width), and individual patient variables (bolus distribution depending on cardiac output, lung passage, carotid status, patient movement, partial volume effect); and as a result of these factors, PCT values are also dependent on different source data quality (SNR, voxel TAC, motion artifacts). Moreover, data postprocessing may be influenced by different software properties and settings (such as motion correction, reference vessel selection, tissue and vessel segmentation, filtering) and finally different PCT models/algorithms, as recently found by Kudo et al.¹²

Although it has been shown that PCT values in healthy tissue agree well with measurements by xenon-CT, single-photon emission CT, positron-emission tomography, and MR imaging,^21–23 they are reliable/comparable only if source dataset quality is sufficient and postprocessing is done in a highly standardized way.²⁴ This condition explains why relatively heterogeneous PCT values have been reported. Likewise, different thresholds for infarct core and penumbra definition (TAR/NVT differentiation) have been proposed. Although for some postprocessing algorithms TAR/NVT thresholds could be validated in smaller patient collectives,^7–9 none of these has been validated in a large series of patients with a criterion standard as a reference.¹⁴ Thus, no universally accepted TAR/NVT thresholds exist to date.

This underlines the importance of comparing perfusion measurements from different algorithms in a highly standardized way, by using the same source datasets and processing settings to retrieve information on qualitative and quantitative comparability and ischemic lesion-size comparability. In our study, all processing variables such as AIF and venous vessel selection, vessel/tissue segmentation, and spatial filtering were identical for the 2 algorithms investigated. Thus, the net differences can mainly be attributed to the final calculation algorithm.

Qualitative Analysis

MTT was reported to be the most sensitive parameter for describing tissue at risk of infarction.^9,19,21 Other temporal parameters such as TTP have been discussed less frequently in the literature but have also proved highly sensitive^25–27 and similar to DC-derived parameters in terms of positive predictive value.^11,12 Our results indicate that TTP and TTD, the latter of which has recently been introduced in commercial postprocessing software, are well-suited to delineating the overall extent of the ischemic lesion with high image quality. TTD appears to condense both pathologic changes in contrast bolus delay—time from arterial enhancement to tissue enhancement and tissue transit time—in 1 high-contrast image.

The high quality of CBF and CBV maps found in our study has been reported in numerous publications;^2,3,28,30 our findings additionally show that CBF and CBV maps of DC and MS are very comparable in terms of quality and diagnostic value. In cases with poor source dataset quality, CBF_DC and CBV_DC maps were slightly superior to CBF_MS and CBV_MS maps. However, overall method differences were significant neither for CBF_DC versus CBF_MS nor for CBV_DC versus CBV_MS.

Quantitative Analysis

In GM/WM, mean MTT values of 3.4–4.9 seconds (GM) and 4.0–5.7 seconds (WM) have been described,^17–19,21 as well as TTP values of ∼4–8 seconds.^7,29 To our knowledge, TD values have not been reported yet. In our study, mean MTT, TTD, and TTP values were 3.3, 3.7, and 8.1 seconds (GM) and 4.0, 4.8, and 9.3 seconds (WM), respectively. All temporal parameters showed high SDs, but this was expected because they are known to be rather variable (for they are influenced by individual factors affecting bolus distribution). Physiologic CBF values of 50–60 mL/100 g/min (GM) and 20–30 mL/100 g/min (WM) as well as CBV values of ∼3.5 mL/100 g (GM) and ∼2.0 mL/100 g (WM) have been reported.^{9,15,17,19,21,30} The average CBF_DC/CBF_MS as well as CBV_DC/CBV_MS values found in our study are in good agreement with these reported values.

In suspected penumbra/core regions, temporal parameters are markedly prolonged compared with healthy tissue: The average rMTT in our study was ∼220% in ischemic tissue (absolute MTT in suspected penumbra and suspected core was 7.7 seconds and 6.9 seconds, respectively). This compares well with an rMTT of 145% (or an absolute MTT of 5–7 seconds) described in literature for defining penumbra extent.^9,17,18 The average rTTD in our study was ∼300% in ischemic tissue (absolute TTD in suspected penumbra and suspected core was 10.6 seconds and 13.0 seconds, respectively). rTTP was ∼170% (absolute TTP in suspected penumbra and suspected core was 13.5 seconds and 14.2 seconds, respectively). This is a TTP delay of ∼4–6 seconds, which is in perfect accordance with the average TTP delay reported in the literature.^7,11,26,31 That rMTT in our study tended to be somewhat higher than in other studies may be explained by the fact that we measured mean region-of-interest values, not lower threshold values, and also by high temporal parameter variability.

An important finding of our study was that in penumbral regions, CBF_DC was lower than CBF_MS. This is probably because the no-venous-outflow assumption of the MS model is violated earlier for nonischemic tissue with shorter transit times than for ischemic tissue. Our quantitative results and visual findings on color maps indicate that identical thresholds are not applicable for DC and MS; rather, adapted thresholds should be used for either model. When compared with DC, MS tends to overestimate CBF in penumbral regions; therefore, the absolute CBF_DC threshold for TAR definition must be lower than the CBF_MS threshold. These findings are in agreement with rCBF_DC ≤ .55–.65 (absolute, ≤25 mL/100 g/min)^4,8,17,18 and rCBF_MS ≤ ∼.60 (absolute, ≤35 mL/100 g/min)^7,15,16 for penumbra (TAR) definition in the literature.

Another important finding was that the average value as well as the SD of CBV_DC was markedly higher than those of CBV_MS in suspected penumbra regions of interest. This can be explained by the fact that CBV_MS is flow-weighted and does not fully account for possible vasodilation in penumbral areas.²³ However, because autoregulation is not maintained in nonviable tissue, CBV_DC was only slightly higher than CBV_MS in suspected core regions of interest (relative, 0.37 versus 0.30 mL/100 g; absolute, 0.12 versus 0.9). Our quantitative results and visual findings on color maps indicate that the absolute CBV_DC threshold for NVT definition must be slightly higher than or similar to the CBV_MS threshold. Values for defining the infarction core (NVT) in the literature are rCBV_DC ≤ ∼.60 (absolute, ∼1.0–2.0 mL/100 g)^8,9,19 and rCBV_MS ≤ ∼.40 (absolute, ∼1.2–1.6 mL/100 g).^7,16 The rCBF_DC and rCBF_MS values of ∼.20 in suspected infarction core regions of interest found in our study are in perfect agreement with rCBF values of ∼.20 (absolute, ∼10–15 mL/100 g/min) reported.^7,15,25

PCT Lesion Sizes

The central task of a modern stroke-imaging technique is to detect and delineate an ischemic lesion and to provide reliable information on its severity. In our study, corresponding region-of-interest–based measurements from DC and MS were within the same range, but our results suggest that identical CBF/CBV thresholds for TAR/NVT definition are not applicable for DC and MS. Nevertheless, corresponding NVT, TAR, and NVT + TAR lesion sizes for DC versus MS showed excellent agreement when using adapted CBF/CBV thresholds. Because both methods yielded essentially the same information in terms of tissue classification by using PCT lesion scores (modified ASPECTS), it can be concluded that DC and MS fulfilled this task equivalently in our study. No different therapy decisions would have been made when DC and MS were interchanged.

Limitations

A limitation of our study could be that the accuracy of lesion sizes indicated by PCT could not be verified by follow-up imaging. However, the objective of our study was not the verification of PCT lesions but rather a systematic analysis of qualitative and quantitative method comparability. Nevertheless, in comparing TAR/NVT lesion sizes delineated by DC and MS, we found excellent agreement and no clinically significant difference in terms of therapeutic decision making between both approaches. Another limitation of this study is that perfusion measurements and lesion sizes were retrieved from 1 axial section only. Furthermore, the inhomogeneity of source data (nonuniform protocols) can be regarded as a limitation, but this was intended to test different datasets with different image quality with both algorithms. Finally, our trial was conducted retrospectively on a limited number of cases. Therefore, larger randomized prospective trials should follow to further investigate advantages and disadvantages of both techniques, validate TAR/NVT threshold values, and examine the equivalence of DC and MS in terms of patient outcome.

Conclusions

By directly comparing DC and MS, we found that both approaches are—despite some model immanent advantages and disadvantages—very similar with regard to color map quality and quantitative values. For exact TAR/NVT differentiation, different CBF and CBV thresholds must be applied for DC and MS. Lesion sizes indicated by DC and MS showed excellent agreement when using adapted thresholds. In all cases, the same therapy decision would have been made.

Footnotes

Indicates article with supplemental on-line appendix.

References

1.↵
1. Wintermark M,
2. Fischbein NJ,
3. Smith WS,
4. et al
. Accuracy of dynamic perfusion CT with deconvolution in detecting acute hemispheric stroke. AJNR Am J Neuroradiol 2005; 26: 104– 12
Abstract/FREE Full Text
2.↵
1. Kloska SP,
2. Nabavi DG,
3. Gaus C,
4. et al
. Acute stroke assessment with CT: do we need multimodal evaluation? Radiology 2004; 233: 79– 86
CrossRef PubMed
3.↵
1. Lin K,
2. Do KG,
3. Ong P,
4. et al
. Perfusion CT improves diagnostic accuracy for hyperacute ischemic stroke in the 3-hour window: study of 100 patients with diffusion MRI confirmation. Cerebrovasc Dis 2009; 28: 72– 79
CrossRef PubMed
4.↵
1. Wintermark M,
2. Reichhart M,
3. Cuisenaire O,
4. et al
. Comparison of admission perfusion computed tomography and qualitative diffusion- and perfusion-weighted magnetic resonance imaging in acute stroke patients. Stroke 2002; 33: 2025– 31
Abstract/FREE Full Text
5.↵
1. Wintermark M,
2. Meuli R,
3. Browaeys P,
4. et al
. Comparison of CT perfusion and angiography and MRI in selecting stroke patients for acute treatment. Neurology 2007; 68: 694– 97
Abstract/FREE Full Text
6.↵
1. Eastwood JD,
2. Lev MH,
3. Wintermark M,
4. et al
. Correlation of early dynamic CT perfusion imaging with whole-brain MR diffusion and perfusion imaging in acute hemispheric stroke. AJNR Am J Neuroradiol 2003; 24: 1869– 75
Abstract/FREE Full Text
7.↵
1. König M,
2. Kraus M,
3. Theek C,
4. et al
. Quantitative assessment of the ischemic brain by means of perfusion-related parameters derived from perfusion CT. Stroke 2001; 32: 431– 37
Abstract/FREE Full Text
8.↵
1. Murphy BD,
2. Fox AJ,
3. Lee DH,
4. et al
. Identification of penumbra and infarct in acute ischemic stroke using computed tomography perfusion-derived blood flow and blood volume measurements. Stroke 2006; 37: 1771– 77
Abstract/FREE Full Text
9.↵
1. Wintermark M,
2. Flanders AE,
3. Velthuis B,
4. et al
. Perfusion-CT assessment of infarct core and penumbra: receiver operating characteristic curve analysis in 130 patients suspected of acute hemispheric stroke. Stroke 2006; 37: 979– 85
Abstract/FREE Full Text
10.↵
1. Donnan GA,
2. Baron JC,
3. Ma H,
4. et al
. Penumbral selection of patients for trials of acute stroke therapy. Lancet Neurol 2009; 8: 261– 69
CrossRef PubMed
11.↵
1. Konstas AA,
2. Lev MH.
. CT perfusion imaging of acute stroke: the need for arrival time, delay insensitive, and standardized postprocessing algorithms? Radiology 2010; 254: 22– 25
CrossRef PubMed
12.↵
1. Kudo K,
2. Sasaki M,
3. Yamada K,
4. et al
. Differences in CT perfusion maps generated by different commercial software: quantitative analysis by using identical source data of acute stroke patients. Radiology 2010; 254: 200– 09
CrossRef PubMed
13.↵
1. Wintermark M,
2. Albers GW,
3. Alexandrov AV,
4. et al
. Acute stroke imaging research roadmap. AJNR Am J Neuroradiol 2008; 29: e23– 30
Abstract/FREE Full Text
14.↵
1. Wintermark M,
2. Rowley HA,
3. Lev MH.
. Acute stroke triage to intravenous thrombolysis and other therapies with advanced CT or MR imaging: pro CT. Radiology 2009; 251: 619– 26
CrossRef PubMed
15.↵
1. Klotz E,
2. Konig M.
. Perfusion measurements of the brain: using dynamic CT for the quantitative assessment of cerebral ischemia in acute stroke. Eur J Radiol 1999; 30: 170– 84
CrossRef PubMed
16.↵
1. Tomandl BF,
2. Klotz E,
3. Stemper B,
4. et al
. MSCT combines three stroke studies in one. Diagnostic Imaging Europe. 2002; 18: 41– 44
17.↵
1. Sparacia G,
2. Iaia A,
3. Assadi B,
4. et al
. Perfusion CT in acute stroke: predictive value of perfusion parameters in assessing tissue viability versus infarction. Radiol Med 2007; 112: 113– 22. Epub 2007 Feb 22
CrossRef PubMed
18.↵
1. Arakawa S,
2. Wright PM,
3. Koga M,
4. et al
. Ischemic thresholds for gray and white matter: a diffusion and perfusion magnetic resonance study. Stroke 2006; 37: 1211– 16
Abstract/FREE Full Text
19.↵
1. Eastwood JD,
2. Lev MH,
3. Azhari T,
4. et al
. CT perfusion scanning with deconvolution analysis: pilot study in patients with acute middle cerebral artery stroke. Radiology 2002; 222: 227– 36
PubMed
20.↵
1. Pexman JH,
2. Barber PA,
3. Hill MD,
4. et al
. Use of the Alberta Stroke Program Early CT Score (ASPECTS) for assessing CT scans in patients with acute stroke. AJNR Am J Neuroradiol 2001; 22: 1534– 42
Abstract/FREE Full Text
21.↵
1. Miles KA,
2. Eastwood JD,
3. Koenig M
. Multidetector Computed Tomography in Cerebrovascular Disease: CT Perfusion Imaging. Abingdon, Oxon, UK: Informa Healthcare Ltd; 2007
22.↵
1. Wintermark M,
2. Thiran JP,
3. Maeder P,
4. et al
. Simultaneous measurement of regional cerebral blood flow by perfusion CT and stable xenon CT: a validation study. AJNR Am J Neuroradiol 2001; 22: 905– 14
Abstract/FREE Full Text
23.↵
1. König M,
2. Klotz E,
3. Luka B,
4. et al
. Perfusion CT of the brain: diagnostic approach for early detection of ischemic stroke. Radiology 1998; 209: 85– 93
PubMed
24.↵
1. Soares BP,
2. Dankbaar JW,
3. Bredno J,
4. et al
. Automated versus manual post-processing of perfusion-CT data in patients with acute cerebral ischemia: influence on interobserver variability. Neuroradiology 2009; 51: 445– 51
CrossRef PubMed
25.↵
1. Mayer TE,
2. Hamann GF,
3. Baranczyk J,
4. et al
. Dynamic CT perfusion imaging of acute stroke. AJNR Am J Neuroradiol 2000; 21: 1441– 49
Abstract/FREE Full Text
26.↵
1. Reichenbach JR,
2. Rother J,
3. Jonetz-Mentzel L,
4. et al
. Acute stroke evaluated by time-to-peak mapping during initial and early follow-up perfusion CT studies. AJNR Am J Neuroradiol 1999; 20: 1842– 50
Abstract/FREE Full Text
27.↵
1. Dittrich R,
2. Kloska SP,
3. Fischer T,
4. et al
. Accuracy of perfusion-CT in predicting malignant middle cerebral artery brain infarction. J Neurol 2008; 255: 896– 902
CrossRef PubMed
28.↵
1. König M,
2. Klotz E,
3. Luka B,
4. et al
. Perfusion CT of the brain: diagnostic approach for early detection of ischemic stroke. Radiology 1998; 209: 85– 93
PubMed
29.↵
1. Kloska SP,
2. Fischer T,
3. Sauerland C,
4. et al
. Increasing sampling interval in cerebral perfusion CT limitation for the maximum slope model. Acad Radiol 2010; 17: 61– 66. Epub 2009 Sep 5
CrossRef PubMed
30.↵
1. Tomandl BF,
2. Klotz E,
3. Handschu R,
4. et al
. Comprehensive imaging of ischemic stroke with multisection CT. Radiographics 2003; 23: 565– 92
PubMed
31.↵
1. Olivot JM,
2. Mlynash M,
3. Thijs VN,
4. et al
. Optimal Tmax threshold for predicting penumbral tissue in acute stroke. Stroke 2009; 40: 469– 75. Epub 2008 Dec 24
Abstract/FREE Full Text

Received January 27, 2010.
Accepted after revision April 5, 2010.

View Abstract

In this issue

Download PDF

Email Article

Citation Tools

Purchase

Cited By...

More in this TOC Section

Show more BRAIN

[1] 1.↵
Wintermark M,
Fischbein NJ,
Smith WS,
et al
. Accuracy of dynamic perfusion CT with deconvolution in detecting acute hemispheric stroke. AJNR Am J Neuroradiol 2005; 26: 104– 12
Abstract/FREE Full Text

[2] Wintermark M,

[3] Fischbein NJ,

[4] Smith WS,

[5] et al

[6] 2.↵
Kloska SP,
Nabavi DG,
Gaus C,
et al
. Acute stroke assessment with CT: do we need multimodal evaluation? Radiology 2004; 233: 79– 86
CrossRef PubMed

[7] Kloska SP,

[8] Nabavi DG,

[9] Gaus C,

[10] et al

[11] 3.↵
Lin K,
Do KG,
Ong P,
et al
. Perfusion CT improves diagnostic accuracy for hyperacute ischemic stroke in the 3-hour window: study of 100 patients with diffusion MRI confirmation. Cerebrovasc Dis 2009; 28: 72– 79
CrossRef PubMed

[12] Lin K,

[13] Do KG,

[14] Ong P,

[15] et al

[16] 4.↵
Wintermark M,
Reichhart M,
Cuisenaire O,
et al
. Comparison of admission perfusion computed tomography and qualitative diffusion- and perfusion-weighted magnetic resonance imaging in acute stroke patients. Stroke 2002; 33: 2025– 31
Abstract/FREE Full Text

[17] Wintermark M,

[18] Reichhart M,

[19] Cuisenaire O,

[20] et al

[21] 5.↵
Wintermark M,
Meuli R,
Browaeys P,
et al
. Comparison of CT perfusion and angiography and MRI in selecting stroke patients for acute treatment. Neurology 2007; 68: 694– 97
Abstract/FREE Full Text

[22] Wintermark M,

[23] Meuli R,

[24] Browaeys P,

[25] et al

[26] 6.↵
Eastwood JD,
Lev MH,
Wintermark M,
et al
. Correlation of early dynamic CT perfusion imaging with whole-brain MR diffusion and perfusion imaging in acute hemispheric stroke. AJNR Am J Neuroradiol 2003; 24: 1869– 75
Abstract/FREE Full Text

[27] Eastwood JD,

[28] Lev MH,

[29] Wintermark M,

[30] et al

[31] 7.↵
König M,
Kraus M,
Theek C,
et al
. Quantitative assessment of the ischemic brain by means of perfusion-related parameters derived from perfusion CT. Stroke 2001; 32: 431– 37
Abstract/FREE Full Text

[32] König M,

[33] Kraus M,

[34] Theek C,

[35] et al

[36] 8.↵
Murphy BD,
Fox AJ,
Lee DH,
et al
. Identification of penumbra and infarct in acute ischemic stroke using computed tomography perfusion-derived blood flow and blood volume measurements. Stroke 2006; 37: 1771– 77
Abstract/FREE Full Text

[37] Murphy BD,

[38] Fox AJ,

[39] Lee DH,

[40] et al

[41] 9.↵
Wintermark M,
Flanders AE,
Velthuis B,
et al
. Perfusion-CT assessment of infarct core and penumbra: receiver operating characteristic curve analysis in 130 patients suspected of acute hemispheric stroke. Stroke 2006; 37: 979– 85
Abstract/FREE Full Text

[42] Wintermark M,

[43] Flanders AE,

[44] Velthuis B,

[45] et al

[46] 10.↵
Donnan GA,
Baron JC,
Ma H,
et al
. Penumbral selection of patients for trials of acute stroke therapy. Lancet Neurol 2009; 8: 261– 69
CrossRef PubMed

[47] Donnan GA,

[48] Baron JC,

[49] Ma H,

[50] et al

[51] 11.↵
Konstas AA,
Lev MH.
. CT perfusion imaging of acute stroke: the need for arrival time, delay insensitive, and standardized postprocessing algorithms? Radiology 2010; 254: 22– 25
CrossRef PubMed

[52] Konstas AA,

[53] Lev MH.

[54] 12.↵
Kudo K,
Sasaki M,
Yamada K,
et al
. Differences in CT perfusion maps generated by different commercial software: quantitative analysis by using identical source data of acute stroke patients. Radiology 2010; 254: 200– 09
CrossRef PubMed

[55] Kudo K,

[56] Sasaki M,

[57] Yamada K,

[58] et al

[59] 13.↵
Wintermark M,
Albers GW,
Alexandrov AV,
et al
. Acute stroke imaging research roadmap. AJNR Am J Neuroradiol 2008; 29: e23– 30
Abstract/FREE Full Text

[60] Wintermark M,

[61] Albers GW,

[62] Alexandrov AV,

[63] et al

[64] 14.↵
Wintermark M,
Rowley HA,
Lev MH.
. Acute stroke triage to intravenous thrombolysis and other therapies with advanced CT or MR imaging: pro CT. Radiology 2009; 251: 619– 26
CrossRef PubMed

[65] Wintermark M,

[66] Rowley HA,

[67] Lev MH.

[68] 15.↵
Klotz E,
Konig M.
. Perfusion measurements of the brain: using dynamic CT for the quantitative assessment of cerebral ischemia in acute stroke. Eur J Radiol 1999; 30: 170– 84
CrossRef PubMed

[69] Klotz E,

[70] Konig M.

[71] 16.↵
Tomandl BF,
Klotz E,
Stemper B,
et al
. MSCT combines three stroke studies in one. Diagnostic Imaging Europe. 2002; 18: 41– 44

[72] Tomandl BF,

[73] Klotz E,

[74] Stemper B,

[75] et al

[76] 17.↵
Sparacia G,
Iaia A,
Assadi B,
et al
. Perfusion CT in acute stroke: predictive value of perfusion parameters in assessing tissue viability versus infarction. Radiol Med 2007; 112: 113– 22. Epub 2007 Feb 22
CrossRef PubMed

[77] Sparacia G,

[78] Iaia A,

[79] Assadi B,

[80] et al

[81] 18.↵
Arakawa S,
Wright PM,
Koga M,
et al
. Ischemic thresholds for gray and white matter: a diffusion and perfusion magnetic resonance study. Stroke 2006; 37: 1211– 16
Abstract/FREE Full Text

[82] Arakawa S,

[83] Wright PM,

[84] Koga M,

[85] et al

[86] 19.↵
Eastwood JD,
Lev MH,
Azhari T,
et al
. CT perfusion scanning with deconvolution analysis: pilot study in patients with acute middle cerebral artery stroke. Radiology 2002; 222: 227– 36
PubMed

[87] Eastwood JD,

[88] Lev MH,

[89] Azhari T,

[90] et al

[91] 20.↵
Pexman JH,
Barber PA,
Hill MD,
et al
. Use of the Alberta Stroke Program Early CT Score (ASPECTS) for assessing CT scans in patients with acute stroke. AJNR Am J Neuroradiol 2001; 22: 1534– 42
Abstract/FREE Full Text

[92] Pexman JH,

[93] Barber PA,

[94] Hill MD,

[95] et al

[96] 21.↵
Miles KA,
Eastwood JD,
Koenig M
. Multidetector Computed Tomography in Cerebrovascular Disease: CT Perfusion Imaging. Abingdon, Oxon, UK: Informa Healthcare Ltd; 2007

[97] Miles KA,

[98] Eastwood JD,

[99] Koenig M

[100] 22.↵
Wintermark M,
Thiran JP,
Maeder P,
et al
. Simultaneous measurement of regional cerebral blood flow by perfusion CT and stable xenon CT: a validation study. AJNR Am J Neuroradiol 2001; 22: 905– 14
Abstract/FREE Full Text

[101] Wintermark M,

[102] Thiran JP,

[103] Maeder P,

[104] et al

[105] 23.↵
König M,
Klotz E,
Luka B,
et al
. Perfusion CT of the brain: diagnostic approach for early detection of ischemic stroke. Radiology 1998; 209: 85– 93
PubMed

[106] König M,

[107] Klotz E,

[108] Luka B,

[109] et al

[110] 24.↵
Soares BP,
Dankbaar JW,
Bredno J,
et al
. Automated versus manual post-processing of perfusion-CT data in patients with acute cerebral ischemia: influence on interobserver variability. Neuroradiology 2009; 51: 445– 51
CrossRef PubMed

[111] Soares BP,

[112] Dankbaar JW,

[113] Bredno J,

[114] et al

[115] 25.↵
Mayer TE,
Hamann GF,
Baranczyk J,
et al
. Dynamic CT perfusion imaging of acute stroke. AJNR Am J Neuroradiol 2000; 21: 1441– 49
Abstract/FREE Full Text

[116] Mayer TE,

[117] Hamann GF,

[118] Baranczyk J,

[119] et al

[120] 26.↵
Reichenbach JR,
Rother J,
Jonetz-Mentzel L,
et al
. Acute stroke evaluated by time-to-peak mapping during initial and early follow-up perfusion CT studies. AJNR Am J Neuroradiol 1999; 20: 1842– 50
Abstract/FREE Full Text

[121] Reichenbach JR,

[122] Rother J,

[123] Jonetz-Mentzel L,

[124] et al

[125] 27.↵
Dittrich R,
Kloska SP,
Fischer T,
et al
. Accuracy of perfusion-CT in predicting malignant middle cerebral artery brain infarction. J Neurol 2008; 255: 896– 902
CrossRef PubMed

[126] Dittrich R,

[127] Kloska SP,

[128] Fischer T,

[129] et al

[130] 28.↵
König M,
Klotz E,
Luka B,
et al
. Perfusion CT of the brain: diagnostic approach for early detection of ischemic stroke. Radiology 1998; 209: 85– 93
PubMed

[131] König M,

[132] Klotz E,

[133] Luka B,

[134] et al

[135] 29.↵
Kloska SP,
Fischer T,
Sauerland C,
et al
. Increasing sampling interval in cerebral perfusion CT limitation for the maximum slope model. Acad Radiol 2010; 17: 61– 66. Epub 2009 Sep 5
CrossRef PubMed

[136] Kloska SP,

[137] Fischer T,

[138] Sauerland C,

[139] et al

[140] 30.↵
Tomandl BF,
Klotz E,
Handschu R,
et al
. Comprehensive imaging of ischemic stroke with multisection CT. Radiographics 2003; 23: 565– 92
PubMed

[141] Tomandl BF,

[142] Klotz E,

[143] Handschu R,

[144] et al

[145] 31.↵
Olivot JM,
Mlynash M,
Thijs VN,
et al
. Optimal Tmax threshold for predicting penumbral tissue in acute stroke. Stroke 2009; 40: 469– 75. Epub 2008 Dec 24
Abstract/FREE Full Text

[146] Olivot JM,

[147] Mlynash M,

[148] Thijs VN,

[149] et al

Main menu

User menu

Search

American Journal of Neuroradiology

Perfusion CT in Acute Ischemic Stroke: A Qualitative and Quantitative Comparison of Deconvolution and Maximum Slope Approach

Abstract

Abbreviations

Materials and Methods

Datasets

Postprocessing

Qualitative Analysis

Quantitative Analysis

PCT Lesion Sizes

Statistical Analysis

Results

Qualitative Results

Quantitative Results

PCT Lesion Sizes

Discussion

Qualitative Analysis

Quantitative Analysis

PCT Lesion Sizes

Limitations

Conclusions

Footnotes

References

In this issue

Citation Manager Formats

Related Articles

Cited By...

More in this TOC Section

Similar Articles

News and Updates

Resources

Opportunities

American Society of Neuroradiology

Main menu

User menu

Search

Perfusion CT in Acute Ischemic Stroke: A Qualitative and Quantitative Comparison of Deconvolution and Maximum Slope Approach

Abstract

Abbreviations

Materials and Methods

Datasets

Postprocessing

Qualitative Analysis

Quantitative Analysis

PCT Lesion Sizes

Statistical Analysis

Results

Qualitative Results

Quantitative Results

PCT Lesion Sizes

Discussion

Qualitative Analysis

Quantitative Analysis

PCT Lesion Sizes

Limitations

Conclusions

Footnotes

References

In this issue

Citation Manager Formats

Jump to section

Related Articles

Cited By...

More in this TOC Section

Similar Articles

News and Updates

Resources

Opportunities

American Society of Neuroradiology