Quantitative evaluation of C-arm CT cerebral blood volume in a canine model of ischemic stroke

Abstract

Background and purpose: Previous studies have shown the feasibility of assessing qualitative CBV measurements in the angiography suite by using FPD-CBCT systems. We have investigated the correlation of FPD-CBCT CBV lesion volumes to the infarct volume.

Materials and methods: Unilateral strokes were created in 7 adult dogs. MR imaging and FPD-CBCT data were obtained after MCA occlusion. FPD-CBCT CBV and ADC maps were generated for all subjects. The animals were sacrificed immediately following the last imaging study to measure infarct volume on histology. The reliability of FPD-CBCT-based lesion volume measurements was compared with those measured histologically by using regression and Bland-Altman analysis.

Results: The best correlation (R(2) = 0.72) between lesion volumes assessed with FPD-CBCT and histology was established with a threshold of mean healthy CBV - 2.5 × SD. These results were inferior to the correlation of lesion volumes measured with ADC and histology (R(2) = 0.99). Bland-Altman analysis showed that the agreement of ADC-derived lesion volumes with histology was superior to the agreement of FPD-CBCT-derived lesion volumes with histology.

Conclusions: We correlated FPD-CBCT measurements of CBV and MR ADC lesion volumes with histologically assessed infarct volume. As expected, ADC is a very accurate and precise method for determining the extent of infarction. FPD-CBCT CBV lesion volumes are correlated to the size of the infarct. Improvement of FPD-CBCT image quality provides an opportunity to establish quantitative CBV measurement in the angiography suite.

Fig 1.

Representative of DSA and TOF imaging canine stroke modeling: angiographic data of a single subject obtained during embolic stroke induction. A, DSA of the left ICA before introducing the autologous clot (ventral projection). B, The injected blood clot is seen on radiography due to the presence of barium sulfate (arrow). C and D, DSA (C, ventral projection) and 3D time-of-flight image (D, ventral projection) of the right ICA confirm occlusion of the distal left ICA and proximal MCA (arrows).

Fig 2.

Representative ADC, MTT, and CBV for canine stroke: example of image data for a single subject. DWI (A), MR perfusion (MTT map) (B), and FPD-CBCT–assessed CBV (C) all show abnormality of the left hemisphere due to the ischemic insult. Squares on the CBV maps (C) indicate the VOIs that were used to calculate CBV values used in equation 3.

Fig 3.

Representative histologic results: An example of histology by TTC (A) and Fluoro-Jade C staining (B) of the same subject and approximate section as shown in Fig 1. Ischemic regions observed with MR and CBCT imaging are confirmed by absence of TTC staining and by fluorescent green−stained degenerated neurons (magnification, ×20).

Fig 4.

Linear regression analysis. Lesion volumes in FPD-CBCT−assessed CBV data were measured by using various lesion thresholds with t_lesion=CBV_h−kσ. A−E, An optimal threshold was established by linear regression analysis of the CBV lesion volumes by using various k-values in relation to lesion volumes obtained with histology. F, The results were compared with the regression analysis of lesion volumes assessed with ADC versus lesion volumes measured with histology. The qualities of the linear fits are summarized in Table 2.

Fig 5.

Results of Bland-Altman analysis performed to evaluate the agreement of lesion volumes assessed with FPD-CBCT CBV (A) and ADC (B) with histologically measured infarct volumes. Mean differences and 1.96 × SD are indicated by the solid and dashed lines, respectively.