Conventional dynamic susceptibility contrast (DSC) MR perfusion imaging requires the use of gadolinium-based contrast agent, however, potential adverse effects preclude its use in some populations, and for repeated scanning of dynamic physiological processes ¹. Previous studies have shown that the analysis of resting state blood oxygenation level–dependent (BOLD) signal can identify cerebrovascular impairment in stroke ² and Moyamoya disease ³. In this study, we evaluate the feasibility of resting-state BOLD for simultaneous evaluation of hemodynamic compromise and functional brain connectivity in patients with chronic cerebrovascular diseases. We propose a novel iterative approach for temporal-shift analysis. The findings represent the first report of the effects of acetazolamide (ACZ), a vasodilator used in the interrogation of hemodynamic reserve, on temporal-shift assessments and functional connectivity of brain networks.Ten patients with chronic steno-occlusive disease of the anterior circulation (age, 32-70 years; mean, 49.7 years; 7 females) were evaluated. BOLD images were acquired using a gradient-echo echo-planar-imaging (EPI) sequence: TR/TE = 2000/30 ms, FA = 78°, FOV = 220 x 220 mm², matrix = 64 x 64, slice-thickness = 4 mm, 30 slices. The BOLD acquisition spanned 20 minutes with a total of 600 repeated image volumes. At 5 minutes following initiation of the BOLD scan, ACZ was slowly infused over 3-5 minutes. DSC perfusion was performed thereafter with the injection of Gadobenate Dimeglumine (MultiHance, Bracco, Milan, Italy) at 4 ml/s, and the time-to-maximum of the residue function map (Tmax) calculated. The first and the last 5 minutes of the BOLD data were used in the analysis, corresponding to pre- and post-ACZ phases of the examination. The optimal temporal offset of the BOLD signal that maximizes its correlation with a reference signal was calculated to produce the temporal-shift (TS) map (Fig. 1). Two reference signals were considered: the global mean signal over the entire brain and the average time series over a region of interest within the superior sagittal sinus (SSS). For the SSS approach, we proposed a novel iterative approach to optimize the calculation of TS maps: the average time course over voxels with zero time delay from previous iteration was calculated and employed as the reference signal for the next iteration. Regions with long Tmax (> 4 s) were defined and used as the reference/gold-standard. Spatial overlap between abnormal regions defined by the TS map and the Tmax map was evaluated using the Dice similarity coefficient ². The TS map was used to correct for functional connectivity analysis of the default mode network (DMN) and sensorimotor network (SMN). The same analyses were performed for both pre-ACZ and post-ACZ data and the results were compared.Fig 2 shows an example of the resting-state BOLD TS maps and the Tmax maps in a patient with severe stenosis of the left internal carotid artery. Brain regions with long Tmax were associated with a positive TS value (a delay in the time course with respect to the reference signal) in the temporal-shift analysis. TS maps can be used to correct for the functional analysis of the DMN and SMN to identify nodes of the brain network that would be missed without the correction (Fig 3). Also, the use of ACZ changed the results of functional-connectivity analysis. ANOVA found that TS maps obtained with SSS signal as reference showed higher similarity with Tmax maps when evaluated with Dice Similarity coefficient by comparison with those using global signal (P = 0.027) (Fig 4A). Compared with results from pre-ACZ data, similarity between TS and Tmax maps from post-ACZ analysis was lower (P = 0.009). ANOVA showed that the volumes of voxels identified to be within the respective brain networks increased significantly with TS correction for DMN (P = 0.012) and SMN (P = 0.036), suggesting TS correction can recover brain networks that would be otherwise missed due to hemodynamic compromise (Fig 4B and C).Temporal-shift analysis of the resting state BOLD signal can identify brain regions with hemodynamic compromise as measured by DSC Tmax maps among patients with chronic cerebrovascular diseases. The use of ACZ changes the behavior of temporal-shift analysis and functional connectivity analysis. Further development of this technique might provide a non-invasive alternative to contrast agent based MR perfusion imaging. Simultaneous characterizing of network connectivity and cerebral hemodynamics may provide a novel approach to risk stratification among patients with cerebrovascular diseases based on “function-physiology mismatch” by evaluating concordance and disparity between functional impairment and physiological measurements as estimated by resting state BOLD signal.