"dataset for Closed-loop Vision Guided Hooking of Flexible Microelectrodes with Reliability Enhancement"

"This dataset contains images and grountruth masks for closed-loop automated hooking. Below is the abstract of the manusript.Multi-thread flexible microelectrodes are promising brain-computer interfaces because of their high signal fidelity and biocompatibility. Due to the small elastic modulus of microelectrodes, a microneedle is usually used to thread into the engaging holes of microelectrodes, transport them to cranial windows and insert them into brain. However, the occlusion between microneedles and microelectrodes poses a challenge to detection of threading results and microelectrode detachment, leading to unreliable open-loop visual guidance. To solve the problem, this paper proposes a closed-loop vision guidance method. First, we introduce a complete microelectrode segmentation target, which is more robust to occlusion than the conventional microelectrode hole segmentation target. Second, we develop an instance segmentation and key point extraction framework that leverages the occlusion relationship between microneedles and microelectrodes to enhance prediction accuracy. Finally, a closed-loop visual feedback strategy is designed based on predicted masks and key points. Experimental result demonstrates that the proposed method achieves accurate detection of the abnormal events and automated adjustment, enhancing the reliability of microelectrode hooking."