Evaluating a Transparent Cranial Window for In Vivo Observation of µECoG Arrays on the Macaque Visual Cortex

High-density micro-electrocorticography (µECoG) provides the spatiotemporal resolution necessary to probe columnar-level cortical architecture. However, chronic multimodal interfacing is fundamentally challenged by aggressive post-surgical tissue responses that rapidly obscure optical access, in addition to localization uncertainties due to brain shift. In this study, established a transparent cranial window interface on the macaque visual cortex. To evaluate the feasibility of this approach, we primarily focused on assessing strategies to mitigate dural tissue regrowth and documenting the inherent biological complications (e.g., hemorrhage), while exploring the potential for direct, in vivo visual localization of µECoG arrays.

A custom transparent chamber assembly integrated with a 64-channel µECoG array was implanted in the visual cortex of two rhesus macaques (n = 2). We compared two dural interface designs-a floating Tecoflex sheet versus a bonded silicone ring-to optimize optical clarity and interface stability. Physical stability of the electrodes relative to vascular landmarks was quantified. Electrophysiological performance was longitudinally evaluated using impedance monitoring, visual evoked potentials, and support vector machine (SVM) neural decoding to discriminate between red-green and black-white grating stimuli, with statistical significance assessed via bootstrap tests.

Regarding the biological and electrical interfaces, while long-term optical maintenance proved challenging due to tissue regrowth or hemorrhage, the bonded silicone ring design effectively mitigated peripheral tissue invasion. Electrophysiologically, array performance captured the dynamic biological transitions of the interface. Stable impedance profiles and statistically significant gamma-band (30-80 Hz) responses-modulating from initial widespread coverage (up to 100%) to restricted subsets (e.g., 16% at five months)-alongside robust SVM classification accuracy (p < 0.001) were tracked longitudinally. Additionally, as a preliminary observation in a single subject, the transparent window permitted the precise identification of electrode positions relative to cortical vasculature, revealing minor physical displacements (median = 0.20 mm) within the initial two weeks post-implantation.

Comparing two dural interface designs highlights the mechanical and biological trade-offs required for chronic optical access, emphasizing that optimizing the mechanical compliance of the dural seal is a critical consideration for long-term multimodal interfaces. Furthermore, this interface shows the potential to overcome conventional localization limitations through early-stage visual co-registration. Although maintaining permanent optical clarity remains a fundamental challenge due to aggressive tissue responses, the system supports longitudinal high-density recording, providing a critical platform for multimodal studies bridging macroscopic network dynamics and microscopic cellular activity.