Dataset from Cellpose3, two-photon calcium imaging from ribo-gcamp8s-expressing neurons

For testing denoising algorithms, we collected a new dataset of noisy images using in vivo two-photon imaging in mice expressing somatic jGCaMP8s (riboL1-jGCaMP8s mouse line) (Zhang et al, 2023, Wang et al 2023) as described in Stringer et al 2021. We imaged each neural field of view for 300+ time-frames at a frame rate of 30Hz. 12 fields of view had dense expression and 8 fields of view had sparse expression. As ground-truth for the segmentations, we used the average of 300 frames. Based on the average ROI size in the time-averaged image segmentation, we set the diameter to 17 for use for all denoising and segmentation.
If you use this dataset, please cite the Cellpose 3 paper: https://www.biorxiv.org/content/10.1101/2024.02.10.579780v2

为测试去噪算法，我们通过在表达体细胞型 jGCaMP8s（ribol1-jGCaMP8s 小鼠品系，Zhang 等人，2023年，Wang 等人，2023年所述）的小鼠体内进行活体双光子成像，收集了一个新的噪声图像数据集。我们以每帧30Hz的帧率对每个神经视野进行了300帧以上的成像。12个视野表现出密集的表达，而8个视野表现出稀疏的表达。作为分割的基准，我们采用了300帧的平均值。基于时间平均图像分割中的平均感兴趣区域（ROI）大小，我们将直径设定为17，用于所有去噪和分割操作。若使用本数据集，请引用 Cellpose 3 论文：https://www.biorxiv.org/content/10.1101/2024.02.10.579780v2。