Abstract Body:

From GoPros to spy cameras, continuous imaging is driving unprecedented advancements in safety, security, and artificial intelligence. A similar approach for neuroimaging in preclinical disease models wherein brain function and its dysregulation can be interrogated continuously could usher in a new era for neuroscience. However, current neuroimaging systems are often bulky (e.g. MRI or PET scanners) or require animals to be anesthetized or head-fixed, which limits our ability to conduct continuous neuroimaging. “Miniscopes”¹ or miniaturized microscopes help overcome this barrier by enabling neuroimaging in freely moving animals, but are limited to a few hours of operation due to power and optical requirements². Additionally, most miniscopes can image neuronal activity^3,4 or vascular function^5-8, but not both, further diminishing their use. Collectively, these constraints prevent continuous functional/molecular neuroimaging in preclinical models of brain health and disease. To address this gap, we developed the “CloudScope” (Figs. 1a–c) by combining an internet of things (IoT) network architecture (Fig. 1a), with a 3D-printed multi-contrast miniscope design⁹ (Fig. 1b). The CloudScope (Fig. 1c, 3.5 grams) can conduct functional/molecular neuroimaging for > 12 hours, in freely behaving mice while allowing users to monitor/control experiments via the internet from anywhere in the world. CloudScope permits tracking of neuronal activity (Neu_ACT) or fluorescently tagged cells via fluorescence (FL) of calcium-bound GCaMP or GFP (Fig. 1d); changes in cerebral blood volume (CBV, using total hemoglobin content or HbT as a surrogate) and oxygenation (Hb_SAT) via intrinsic optical signals (IOS, Fig. 1e); and cerebral blood flow (CBF) via laser speckle contrast (LSC, Fig. 1f) over a 3×3 mm² field of view (FoV) at a spatial resolution of 6 microns every 5 seconds. We dub this unique functional neuroimaging capability “neurosurveillance” and showcase its potential in two preclinical disease models: seizures and brain tumors. Figs. 1g-n show its utility for interrogating seizure-induced neurovascular dysfunction. Instead of studying seizure initiation/cessation^10,11, we continuously interrogated the dramatic changes in Neu_ACT, CBV, CBF, and Hb_SAT, during a drug-induced seizure (via a GABA_A antagonist¹²) and its prolonged aftermath in two animals (A1 and A2). While both animals exhibited similar ictal Neu_ACT dynamics (Fig. 1g), corresponding ictal microvascular changes were substantially different (Figs. 1h-j), with A1 exhibiting greater vasoconstriction (Fig. 1h) and reduced Hb_SAT (Fig. 1j). CBF changes were dissimilar (Fig. 1i). Moreover, the first 60 minutes following each seizure were characterized by reduced Neu_ACT (Fig. 1k), sustained vasoconstriction (Fig. 1l) and reduced CBF (Fig. 1m), which may (i.e. A1) or may not (i.e. A2) result in lower Hb_SAT (Fig. 1n). Importantly, these data show that similar ictal or post-ictal Neu_ACT changes do not result in similar hemodynamic changes. Next, Figs. 1o-t demonstrate neurosurveillance of a brain tumor’s microenvironment (bTME, of a GL261-GFP tumor). Figs. 1m-o show tumor progression and concomitant vascular changes over 24 hr, which allows mapping of the expanding tumor front (Figs. 1o and r) and its directionality (Fig. 1s). To evaluate if the CBF within the bTME was coupled with the CBV, we computed the Pearson’s correlation coefficient (R_CBF,CBV) between their 24 hr time series, and found that the tumor and its progressing front showed CBF/CBV changes that were substantially more positively coupled than their counterparts in the surrounding non-tumor region (Fig. 1t). Collectively, we believe that CloudScope will enable molecular imaging of the brain with novel probes, multi-contrast neuroimaging of preclinical disease models and therapeutics, and empower global collaborations, ushering in a revolution in neuroscience.

Author