Dynamic organization of visual cortical networks revealed by machine learning applied to massive spiking datasets

Опубликована: Апрель 4, 2024

Complex cognitive functions in a mammalian brain are distributed across many anatomically and functionally distinct areas rely on highly dynamic routing of neural activity the network. While modern electrophysiology methods enable recording spiking from increasingly large neuronal populations at cellular level, development probabilistic to extract these inter-area interactions is lagging. Here, we introduce an unsupervised machine learning model that infers connectivity recorded population synchrony their activity. As opposed traditional decoding models reveal dynamics whole population, produces cellular-level cell-type specific functional otherwise omitted analysis. The evaluated ground truth synthetic data compared alternative ensure quality quantification predictions. Our strategy incorporates two sequential stages – extraction static structure network followed by inference temporal changes connection strength. This two-stage architecture enables detailed statistical criteria be developed evaluate confidence predictions comparison with descriptive methods. We applied analyze large-scale in-vivo recordings visual cortices. discovery patterns local long-range circuits cortex temporally varying strength feedforward feedback drives during sensory stimulation. approach provides conceptual link between slow brain-wide studied neuroimaging fast enabled may help uncover often overlooked dimensions code.

Язык: Английский

---

Dynamic organization of visual cortical networks revealed by machine learning applied to massive spiking datasets

Опубликована: Апрель 4, 2024

Complex cognitive functions in a mammalian brain are distributed across many anatomically and functionally distinct areas rely on highly dynamic routing of neural activity the network. While modern electrophysiology methods enable recording spiking from increasingly large neuronal populations at cellular level, development probabilistic to extract these inter-area interactions is lagging. Here, we introduce an unsupervised machine learning model that infers connectivity recorded population synchrony their activity. As opposed traditional decoding models reveal dynamics whole population, produces cellular-level cell-type specific functional otherwise omitted analysis. The evaluated ground truth synthetic data compared alternative ensure quality quantification predictions. Our strategy incorporates two sequential stages – extraction static structure network followed by inference temporal changes connection strength. This two-stage architecture enables detailed statistical criteria be developed evaluate confidence predictions comparison with descriptive methods. We applied analyze large-scale in-vivo recordings visual cortices. discovery patterns local long-range circuits cortex temporally varying strength feedforward feedback drives during sensory stimulation. approach provides conceptual link between slow brain-wide studied neuroimaging fast enabled may help uncover often overlooked dimensions code.

Язык: Английский

---

Towards Robotic Mapping of a Honeybee Comb

Опубликована: Июль 1, 2024

Язык: Английский

---

Large-Volume Electron Microscopy

Neuromethods, Год журнала: 2024, Номер unknown, С. 81 - 103

Опубликована: Янв. 1, 2024

---

Aligning and Restoring Imperfect ssEM Images for Continuity Reconstruction

Lecture notes in computer science, Год журнала: 2024, Номер unknown, С. 543 - 552

Опубликована: Янв. 1, 2024

Язык: Английский

---