Post by Subhrajit D.

Proud to share our new 𝑵𝒂𝒕𝒖𝒓𝒆 𝑴𝒆𝒕𝒉𝒐𝒅𝒔 paper introducing 𝐂𝐥𝐨𝐮𝐝𝐒𝐜𝐨𝐩𝐞, a cloud connected, multicontrast miniscope designed for 𝐧𝐞𝐮𝐫𝐨𝐬𝐮𝐫𝐯𝐞𝐢𝐥𝐥𝐚𝐧𝐜𝐞 in freely behaving animals. What if the brain could be studied not in short experimental snapshots, but continuously, as behavior naturally unfolds? That is the vision behind CloudScope. In this work, we designed and developed CloudScope to move beyond brief observation windows and open a new view into some of neuroscience’s most complex questions: how neural, vascular, and cellular processes interact across natural behavior, seizure related neurovascular disruption, and brain tumor evolution. A key advance was applying deep learning to a type of dataset rarely explored with AI: 𝐜𝐨𝐧𝐭𝐢𝐧𝐮𝐨𝐮𝐬, 𝐥𝐨𝐰 𝐟𝐫𝐞𝐪𝐮𝐞𝐧𝐜𝐲, 𝐰𝐢𝐝𝐞 𝐟𝐢𝐞𝐥𝐝 𝐜𝐚𝐥𝐜𝐢𝐮𝐦 𝐢𝐦𝐚𝐠𝐢𝐧𝐠 acquired across an entire day. Using established computer vision models, we decoded natural behavior from neurosurveillance data, while sanity checks confirmed that the model learned 𝐦𝐞𝐚𝐧𝐢𝐧𝐠𝐟𝐮𝐥 𝐧𝐞𝐮𝐫𝐨𝐛𝐞𝐡𝐚𝐯𝐢𝐨𝐫𝐚𝐥 𝐬𝐢𝐠𝐧𝐚𝐭𝐮𝐫𝐞𝐬 rather than simply memorizing the data. Beyond behavior prediction, CloudScope enabled mapping of neurovascular changes during natural behavior, characterization of seizure related disruption and recovery, and longitudinal tracking of cellular and microvascular changes during brain tumor evolution. We believe CloudScope and neurosurveillance represent an important step toward understanding how brain function changes across natural behavior, disease progression, and therapeutic response, helping us address some of neuroscience’s most complex questions in 𝐡𝐞𝐚𝐥𝐭𝐡 𝐚𝐧𝐝 𝐝𝐢𝐬𝐞𝐚𝐬𝐞. Grateful to Dr. Amit Banerjee and Dr. Arvind P. Pathak for their guidance throughout this work, and to the entire team for making this multidisciplinary effort possible. Link to paper: https://lnkd.in/eDGzA3hM Link to Editor's Research Brief: https://lnkd.in/eVWHxE8n #AIxNeuro #DeepLearning #CloudScope #Neurosurveillance #Neurovascular #Imaging #Neuroimaging #Miniscope #Microscope #PreclinicalImaging #Seizure #BrainCancer #BrainTumor