2.1.

Patients with MMD

The surgeries included in this study were conducted at the Department of Neurosurgery, Zhongnan Hospital of Wuhan University. The study retrospectively analyzed MMD patients who underwent side-to-side superficial temporal artery-middle cerebral artery (SS-STA-MCA) bypass surgery performed by the same neurosurgeon (J.Z.) between October 2021 and October 2022. Diagnosis of MMD in all the patients was established with digital subtraction angiography (DSA).⁸ All patients met the diagnostic criteria set by the Research Committee on Spontaneous Occlusion of the Circle of Willis, Ministry of Health, Labour and Welfare, Japan.

The study protocol received approval from the institutional review board at Zhongnan Hospital of Wuhan University and was conducted by the revised 1983 Declaration of Helsinki. Written informed consent was obtained from all patients.

Inclusion criteria were as follows: (1) diagnosis of MMD according to the MMD diagnostic guidelines, (2) completion of LSCI, (3) no history of vascular intervention, and (4) adult MMD patients. Forty-eight patients met the eligibility criteria, as indicated in Table 1.

Table 1

Information on clinical data of adult patients with MMD.

2.2.

Surgical Procedures

The surgical technique employed for S-S STA-MCA anastomoses followed the standard procedure outlined in previous articles.⁸ The patient is positioned supine, with the head rotated toward the contralateral side. Following a frontotemporal craniotomy, the STA-MCA (M4) anastomosis is performed. Intraoperative confirmation of patency is achieved using indocyanine green video angiography (ICG-VA), and LSCI recordings are captured before and after the anastomosis. Throughout the surgery, mean arterial pressure is maintained at a constant level of 90 to 100 mmHg, and the end-expiratory carbon dioxide (${\mathrm{CO}}_2$) concentration is maintained between 38 and 42 mmHg. Patients were identified as experiencing symptomatic hyperperfusion when they presented with transient neurological deficits (TNDs) manifesting as a constellation of symptoms, including but not limited to aphasia, limb numbness, diminished muscle strength, hemiparesis, seizures, dysarthria, intense headaches, and facial paralysis.^15,16 These manifestations, indicative of the regions impacted by postoperative hyperperfusion, were meticulously evaluated and recorded by an experienced neurosurgeon.

2.3.

Instrumentation of LSCI

In our study, LSCI was performed using a laser speckle device (SIM BFI HR Pro, SIM Opto-Technology Co., Ltd., Wuhan, China).^17–19 The instrument operates at a laser wavelength of 785 nm, with an operating distance of 20 cm and a field of view of $42\times 31\mathrm{mm}$. The camera exposure time was set at 20 ms. The instrument is positioned in advance before the procedure begins and does not interfere with the sterile hood of the microscope or the surgeon’s normal operation.

During a craniotomy, the microscope is positioned above the patient at the neurosurgeon’s discretion, and the laser illumination is activated to display and record laser speckle blood flow images. The system can provide a real-time display of white light images [Figs. 1(a) and 1(b) upper side] and laser speckle blood flow images [Figs. 1(a) and 1(b) lower side] of the surgical area.

Fig. 1

Illustration of intraoperative blood flow imaging and ROI analysis conducted using LSCI. (a) White light and BFI maps obtained within the operative field before anastomosis. (b) White light and BFI maps obtained within the operative field after anastomosis. (c) Magnified images of the ROI location (white circle) in panel (a). (d) Segmented vascular structures in panel (c). (e) Parenchyma BFI in panel (d). (f) Distribution of blood flow velocity index along the radial direction of the vessel in the yellow area of panel (c) before and after anastomosis.

2.4.

Image Analysis

Raw laser speckle images were processed by calculating the speckle contrast $(K)$ within a $7\times 7\text{pixels}$ window using the equation $K=\sigma /⟨I⟩$, where $\sigma$ is the standard deviation of speckle intensity and $⟨I⟩$ is the mean speckle intensity. The blood flow index (BFI), defined as calculated $1/K^2$, is commonly utilized as an approximation for blood flow perfusion.^20,21

To quantify the changes in cortical microcirculation perfusion before and after the SS-STA-MCA anastomosis, regions of interest (ROIs) of the same size were selected after data collection for all cases was completed, as depicted by the white circle in Fig. 1(a). To ensure consistency of the locations of the ROI across pre-anastomosis and post-anastomosis blood flow images, image registration was performed by carefully comparing the morphology of the blood vessels within the ROIs. Figure 1(c) corresponds to the ROI of Figs. 1(a) and 1(b). Figure 1(f) illustrates the radial direction of BFI along the section for the selected vessels before and after anastomosis indicated by the yellow arrow in Fig. 1(c). It is clearly shown that BFI exhibits a significant increase after anastomosis, encompassing both intravascular and parenchymal regions (maximum BFI: pre-anastomosis 272.72 versus post-anastomosis 1472.7). Furthermore, the vessel diameter in the same section also increased after anastomosis [Fig. 1(f)]. Using a single threshold to binarize the BFI led to an underestimation of the vascular structure when the CBF index was relatively low. Therefore, we utilized the Hessian Frangi algorithm to extract morphological changes and enhance the vascular structure of the vessels.^22,23 The enhanced image was segmented using the same threshold value, allowing differentiation of the blood flow structure from the parenchymal region.

Figure 1(d) depicts the blood flow structure before and after anastomosis. Notably, new flow structures emerge after anastomosis [white dashed box in Fig. 1(d)]. The ratio of the blood flow structure area within the ROI to the total ROI area was designated as regional blood flow structuring (rBFS, $\mathrm{rBFS}=S_{\text{vessel}}/S_{\mathrm{ROI}}$); here, $S_{\text{vessel}}$ represents the area of the blood flow structure within the ROI, and $S_{\mathrm{ROI}}$ denotes the total area of the ROI. The parameter $\mathrm{\Delta }{\mathrm{rBFS}}_{\text{Post}/\mathrm{Pre}}={\mathrm{rBFS}}_{\text{Post}}/{\mathrm{rBFS}}_{\mathrm{Pre}}$ serves as an indicator of vessel dilation following anastomosis.

Figure 1(e) displays the BFI of the cortical regions after excluding vascular structures. The average BFI in the parenchyma regions was defined as regional cerebral blood flow (rCBF, $\mathrm{rCBF}=\sum _{\text{Parenchyma}}\mathrm{BFI}/S_{\text{Parenchyma}}$); here, $S_{\text{Parenchyma}}$ represents the parenchyma area excluding the blood flow structure. The parameter $\mathrm{\Delta }{\mathrm{rCBF}}_{\text{Post}/\mathrm{Pre}}={\mathrm{rCBF}}_{\text{Post}}/{\mathrm{rCBF}}_{\mathrm{Pre}}$ was introduced to characterize the ratio of changes in cortical perfusion.

2.5.

Statistical Analysis

This study utilized IBM SPSS Statistics V.22.0 for statistical analysis. Continuous variables with normal distribution were presented as mean ± standard deviation; non-normal variables were reported as median (interquartile range). To compare the parameters before and after anastomosis within each patient, a paired $t$-test was utilized. Categorical variables were analyzed in contingency tables with Fisher’s exact test. To control the false discovery rate, the false discovery rate (FDR) correction was applied. Receiver operating characteristic (ROC) analysis identified optimal cutoffs of rCBF and $\mathrm{\Delta }{\mathrm{rCBF}}_{\text{Post}/\mathrm{Pre}}$ for predicting CHS. Independent $t$-tests were performed for rCBF and rBFS across different subtypes and sexes, given that the data followed a normal distribution. Levene’s test was used to assess the equality of variances, and Welch’s correction was applied when variances were unequal. For non-normally distributed $\mathrm{\Delta }{\mathrm{rCBF}}_{\text{Post}/\mathrm{Pre}}$ and $\mathrm{\Delta }{\mathrm{rBFS}}_{\text{Post}/\mathrm{Pre}}$, the Mann-Whitney $U$-test was employed. Pearson correlation coefficients determined the strength of correlations between variables. Analysis of covariance assessed differences in slopes and intercepts of fitted curves. Results with $p<0.05$ were considered statistically significant.

3.1.

Perfusion Changes in the Cortical Microcirculation Before and After Anastomosis

Statistical analysis was conducted on 48 patients to examine the changes in rBFS and rCBF before and after anastomosis. As shown in Fig. 2(a), a significant increase in rBFS was observed after anastomosis compared to before anastomosis $(p<0.001)$. Figure 2(b) demonstrated a significant increase in rCBF $(p<0.001)$. Both rBFS and rCBF exhibited increases after anastomosis in all patients. Figure 2(c) revealed a significant negative correlation between rBFSPre and $\mathrm{\Delta }{\mathrm{rBFS}}_{\text{Post}/\mathrm{Pre}}$ (Pearson $r=-0.40$, $p=0.005$). Figure 2(d) showed a significant negative correlation between ${\mathrm{rCBF}}_{\mathrm{Pre}}$ and $\mathrm{\Delta }{\mathrm{rCBF}}_{\text{Post}/\mathrm{Pre}}$ (Pearson $r=-0.63$, $p<0.001$).

Fig. 2

Changes in rBFS and rCBF before and after anastomosis in 48 patients. (a) There was a significant increase in rBFS after anastomosis compared to the pre-anastomosis state. (b) There was a significant increase in rCBF after anastomosis compared to before anastomosis. (c) ${\mathrm{rBFS}}_{\mathrm{Pre}}$ showed a significant negative correlation with $\mathrm{\Delta }{\mathrm{rBFS}}_{\text{Post}/\mathrm{Pre}}$. (d) ${\mathrm{rCBF}}_{\mathrm{Pre}}$ showed a significant negative correlation with $\mathrm{\Delta }{\mathrm{rCBF}}_{\text{Post}/\mathrm{Pre}}$. (*$p<0.05$, **$p<0.01$, ***$p<0.001$).

Patients were grouped based on the presence or absence of CHS following surgery. Among these, seven patients exhibited transient symptoms of CHS postoperatively. According to the data presented in Table 2, a significant difference was observed in the ${\mathrm{rCBF}}_{\mathrm{Pre}}$ between patients with CHS and those who did not. $\mathrm{\Delta }{\mathrm{rCBF}}_{\text{Post}/\mathrm{Pre}}$ also showed significant variation. Conversely, ${\mathrm{rCBF}}_{\text{Post}}$ and rBFS values showed no significant divergence between the two groups. Clinical variables such as sex, presence of hypertension, and the Suzuki stage did not reveal any significant differences between the groups.

Table 2

Comparison of basic characteristics and hemodynamic data.

Figure 3 illustrates the ROC curve analysis employed to evaluate the predictive power of ${\mathrm{rCBF}}_{\mathrm{Pre}}$ and $\mathrm{\Delta }{\mathrm{rCBF}}_{\text{Post}/\mathrm{Pre}}$ concerning the onset of postoperative CHS. The optimal balance between sensitivity and specificity, as quantified by the Youden index, is achieved at cutoff values of 95.63 for ${\mathrm{rCBF}}_{\mathrm{Pre}}$ and 3.805 for $\mathrm{\Delta }{\mathrm{rCBF}}_{\text{Post}/\mathrm{Pre}}$. The areas under the curve (AUC) for ${\mathrm{rCBF}}_{\mathrm{Pre}}$ and $\mathrm{\Delta }{\mathrm{rCBF}}_{\text{Post}/\mathrm{Pre}}$ stand at 0.753 and 0.738, respectively.

Fig. 3

ROC curve of ${\mathrm{rCBF}}_{\mathrm{Pre}}$ and $\mathrm{\Delta }{\mathrm{rCBF}}_{\text{Post}/\mathrm{Pre}}$ for predicting CHS.

3.2.

Differences in Cortical Microcirculatory Perfusion Between Hemorrhagic and Ischemic

Statistical analysis examined the changes in rCBF and rBFS before and after anastomosis in patients with hemorrhagic and ischemic MMD in SS-STA-MAC. Figure 4 illustrates that rCBF and rBFS increased significantly after anastomosis in both ischemic and hemorrhagic MMD patients (ischemic rCBF, $p<0.001$; ischemic rBFS, $p<0.001$; hemorrhagic rCBF, $p<0.001$; hemorrhagic rBFS, $p<0.001$).

Fig. 4

Comparison of rCBF and rBFS before and after anastomosis in hemorrhagic and ischemic MMD patients. (a) Statistics of rCBF for hemorrhagic and ischemic MMD patients. (b) Statistics of rBFS for hemorrhagic and ischemic MMD patients. (c) Statistics of $\mathrm{\Delta }{\mathrm{rCBF}}_{\text{Post}/\mathrm{Pre}}$ and $\mathrm{\Delta }{\mathrm{rBFS}}_{\text{Post}/\mathrm{Pre}}$ for hemorrhagic and ischemic MMD patients. (d) Curve fitting of the scatter plot illustrating the relationship between ${\mathrm{rCBF}}_{\mathrm{Pre}}$ and ${\mathrm{rCBF}}_{\text{Post}}$. (e) Curve fitting of the scatter plot illustrating the relationship between ${\mathrm{rCBF}}_{\mathrm{Pre}}$ and ${\mathrm{rCBF}}_{\text{Post}}$. (*$p<0.05$, **$p<0.01$, ***$p<0.001$).

In Fig. 4(a), the ischemic ${\mathrm{rCBF}}_{\mathrm{Pre}}$ was significantly lower than the hemorrhagic ${\mathrm{rCBF}}_{\mathrm{Pre}}$ (ischemic $116.26\pm 41.57$ versus hemorrhagic $122.35\pm 68.58$, $p=0.036$). However, there was no significant difference in rCBF between hemorrhagic and ischemic patients post-anastomosis. In Figs. 4(b)–4(c), there were no significant differences in rBFS between ischemic and hemorrhagic MMD patients.

Curve fitting was performed to analyze the relationship between ${\mathrm{rCBF}}_{\mathrm{Pre}}$ versus ${\mathrm{rCBF}}_{\mathrm{Post}}$ and ${\mathrm{rBFS}}_{\text{Pre}}$ versus ${\mathrm{rBFS}}_{\text{Post}}$ in patients with ischemic and hemorrhagic MMD [Figs. 4(d)–4(e)]. The Pearson correlation coefficient ($r$) was calculated and tested for significance. In Fig. 4(d), it can be observed that there was no significant correlation between ${\mathrm{rCBF}}_{\mathrm{Pre}}$ and ${\mathrm{rCBF}}_{\mathrm{Post}}$ in ischemic patients ($r=0.36$), whereas a significant correlation was found in patients with hemorrhagic MMD ($r=0.65$, $p=0.003$). The slopes of the two curves differed significantly ($p<0.001$), and the distribution of rCBF was more concentrated in patients with ischemic MMD compared to those with hemorrhagic MMD.

In Fig. 4(e), a significant correlation was observed between ${\mathrm{rBFS}}_{\mathrm{Pre}}$ and ${\mathrm{rBFS}}_{\text{Post}}$ in both hemorrhagic and ischemic MMD patients (ischemic, $r=0.65$, $p<0.001$; hemorrhagic, $r=0.74$, $p<0.001$). The slope of ${\mathrm{rBFS}}_{\text{Post}}$ in ischemic patients was significantly higher than that in hemorrhagic patients (ischemic, 1.016; hemorrhagic, 0.70; $p<0.001$). A higher rBFS slope indicates greater changes in ${\mathrm{rBFS}}_{\text{Post}}$.

3.3.

Differences in Cortical Microcirculatory Perfusion Between Males and Females

Figure 5 presents the statistical analysis of rCBF and rBFS before and after the anastomosis in adults with MMD of different sexes. In Fig. 5(a), both sexes showed a significant increase in rCBF and rBFS after surgery ($p<0.001$ for both sexes). Female patients had significantly higher rCBF than male patients ($p=0.043$). After surgery, female patients continued to have significantly higher rCBF than males ($p=0.035$). In Fig. 5(b), there was no significant difference in ${\mathrm{rBFS}}_{\mathrm{Pre}}$ between male and female patients ($0.24\pm 0.015$ in females versus $0.24\pm 0.019$ in males). However, after the anastomosis, rBFS was significantly higher in females than in males ($0.27\pm 0.014$ in females versus $0.26\pm 0.021$ in males; $p=0.005$). In Fig. 5(c), there were no significant differences in $\mathrm{\Delta }{\mathrm{rBFS}}_{\text{Post}/\mathrm{Pre}}$ and $\mathrm{\Delta }{\mathrm{rCBF}}_{\text{Post}/\mathrm{Pre}}$ between sexes ($\mathrm{\Delta }{\mathrm{rBFS}}_{\text{Post}/\mathrm{Pre}}$: $110.22\pm 5.38$ in females versus $108.99\pm 6.84$ in males; $\mathrm{\Delta }{\mathrm{rCBF}}_{\text{Post}/\mathrm{Pre}}$: $305.08\pm 143.34$ in females versus $259.78\pm 96.42$ in males).

Fig. 5

Comparison of rCBF and rBFS before and after anastomosis in male and female MMD patients. (a) Statistics of rCBF before and after anastomosis for male and female MMD patients. (b) Statistics of rBFS before and after anastomosis for male and female MMD patients. (c) Statistics of $\mathrm{\Delta }{\mathrm{rCBF}}_{\text{Post}/\mathrm{Pre}}$ and $\mathrm{\Delta }{\mathrm{rBFS}}_{\text{Post}/\mathrm{Pre}}$ for male and female MMD patients. (d) Curve fitting of the scatter plot illustrating the relationship between ${\mathrm{rCBF}}_{\mathrm{Pre}}$ and ${\mathrm{rCBF}}_{\text{Post}}$. (e) Curve fitting of the scatter plot illustrating the relationship between ${\mathrm{rCBF}}_{\mathrm{Pre}}$ and ${\mathrm{rCBF}}_{\text{Post}}$. (*$p<0.05$, **$p<0.01$, ***$p<0.001$).

Curve fitting was performed for ${\mathrm{rCBF}}_{\mathrm{Pre}}$ versus ${\mathrm{rCBF}}_{\text{Post}}$ and ${\mathrm{rBFS}}_{\mathrm{Pre}}$ versus ${\mathrm{rBFS}}_{\text{Post}}$ for both sexes [Figs. 5(d)–5(e)]. The Pearson correlation coefficient ($r$) was calculated and tested for significance, and all four curves showed significant correlations. In Figs. 5(d)–5(e), the fitted curves of rCBF and rBFS for females had higher constant terms than those for males.