3D Microvascular Image Data and Labels for Machine Learning

These images and associated binary labels were collected from collaborators across multiple universities to serve as a diverse representation of biomedical images of vessel structures, for use in the training and validation of machine learning tools for vessel segmentation. The dataset contains images from a variety of imaging modalities, at different resolutions, using difference sources of contrast and featuring different organs/ pathologies. This data was use to train, test and validated a foundational model for 3D vessel segmentation, tUbeNet, which can be found on github. The paper descripting the training and validation of the model can be found here. Filenames are structured as follows:  <br>Data - [Modality]_[species Organ]_[resolution].tif Labels - [Modality]_[species Organ]_[resolution]_labels.tif <br>Sub-volumes of larger dataset - [Modality]_[species Organ]_subvolume[dimensions in pixels].tif <br>Manual labelling of blood vessels was carried out using Amira (2020.2, Thermo-Fisher, UK).   <br><b>Training data:</b><b> </b>opticalHREM_murineLiver_2.26x2.26x1.75um.tif: A high resolution episcopic microscopy (HREM) dataset, acquired in house by staining a healthy mouse liver with Eosin B and imaged using a standard HREM protocol.  <i>NB: </i><i>25% of this image volume was with</i><i>h</i><i>e</i><i>l</i><i>d from training, for use as test data</i><i>.</i><i> </i>CT_murineTumour_20x20x20um.tif: X-ray microCT images of a microvascular cast, taken from a subcutaneous mouse model of colorectal cancer (acquired in house). <i>NB: </i><i>25% of this image volume was </i><i>withhe</i><i>ld</i><i> from training, for use as test data</i><i>.</i><i> </i>RSOM_murineTumour_20x20um.tif: Raster-Scanning Optoacoustic Mesoscopy (RSOM) data from a subcutaneous tumour model (provided by Emma Brown, Bohndiek Group, University of Cambridge). The image data has undergone filtering to reduce the background ​(Brown et al., 2019)​.  OCTA_humanRetina_24x24um.tif: retinal angiography data obtained using Optical Coherence Tomography Angiography (OCT-A) (provided by Dr Ranjan Rajendram, Moorfields Eye Hospital). <b>Test data:</b><b> </b>MRI_porcineLiver_0.9x0.9x5mm.tif: T1-weighted Balanced Turbo Field Echo Magnetic Resonance Imaging (MRI) data from a machine-perfused porcine liver, acquired in-house. <b>Test Data</b><b> </b>MFHREM_murineTumourLectin_2.76x2.76x2.61um.tif: a subcutaneous colorectal tumour mouse model was imaged in house using Multi-fluorescence HREM in house, with Dylight 647 conjugated lectin staining the vasculature ​(Walsh et al., 2021)​. The image data has been processed using an asymmetric deconvolution algorithm described by ​Walsh et al., 2020​. NB:<i> </i><i>A sub-volume of 480x480x640 voxels was manually labelled (</i><i>MFHREM_murineTumourLectin_subvolume480x480x640</i><i>.tif</i><i>).</i><i> </i> MFHREM_murineBrainLectin_0.85x0.85x0.86um.tif: an MF-HREM image of the cortex of a mouse brain, stained with Dylight-647 conjugated lectin, was acquired in house ​(Walsh et al., 2021)​. The image data has been downsampled and processed using an asymmetric deconvolution algorithm described by ​Walsh et al., 2020​. <i>NB: </i><i>A sub-volume of </i><i>1000x1000x99</i><i> voxels was manually labelled.</i><i> This sub-volume is provided at full resolution and without preprocessing (</i><i>MFHREM_murineBrainLectin_subvol_0.57x0.57x0.86um</i><i>.tif</i><i>).</i><i> </i> 2Photon_murineOlfactoryBulbLectin_0.2x0.46x5.2um.tif: two-photon data of mouse olfactory bulb blood vessels, labelled with sulforhodamine 101, was kindly provided by Yuxin Zhang at the Sensory Circuits and Neurotechnology Lab, the Francis Crick Institute ​(Bosch et al., 2022)​.  <i>NB: </i><i>A sub-volume of 500x500x79 voxel was manually labelled (</i><i>2Photon_murineOlfactoryBulbLectin_subvolume500x500x79</i><i>.tif</i><i>).</i><i> </i><br><b>References</b><b>:</b><b> </b>​​Bosch, C., Ackels, T., Pacureanu, A., Zhang, Y., Peddie, C. J., Berning, M., Rzepka, N., Zdora, M. C., Whiteley, I., Storm, M., Bonnin, A., Rau, C., Margrie, T., Collinson, L., &amp; Schaefer, A. T. (2022). Functional and multiscale 3D structural investigation of brain tissue through correlative in vivo physiology, synchrotron microtomography and volume electron microscopy. Nature Communications 2022 13:1, 13(1), 1–16. https://doi.org/10.1038/s41467-022-30199-6 ​Brown, E., Brunker, J., &amp; Bohndiek, S. E. (2019). Photoacoustic imaging as a tool to probe the tumour microenvironment. DMM Disease Models and Mechanisms, 12(7). https://doi.org/10.1242/DMM.039636 ​Walsh, C., Holroyd, N. A., Finnerty, E., Ryan, S. G., Sweeney, P. W., Shipley, R. J., &amp; Walker-Samuel, S. (2021). Multifluorescence High-Resolution Episcopic Microscopy for 3D Imaging of Adult Murine Organs. Advanced Photonics Research, 2(10), 2100110. https://doi.org/10.1002/ADPR.202100110 ​Walsh, C., Holroyd, N., Shipley, R., &amp; Walker-Samuel, S. (2020). Asymmetric Point Spread Function Estimation and Deconvolution for Serial-Sectioning Block-Face Imaging. Communications in Computer and Information Science, 1248 CCIS, 235–249. https://doi.org/10.1007/978-3-030-52791-4_19 ​

本数据集收录的图像及其配套二值标签由多所高校的合作者联合采集，旨在完整呈现血管结构生物医学图像的多样化样例，以支撑血管分割机器学习工具的训练与验证工作。

该数据集涵盖多种成像模态、不同分辨率的图像，采用差异化造影方案，并覆盖不同器官与病理特征的样本。本数据集曾用于训练、测试与验证一款用于三维血管分割的基础模型tUbeNet，该模型的开源代码托管于GitHub平台。相关描述模型训练与验证过程的论文可通过指定链接获取。

文件名格式规范如下：
图像文件：Data - [成像模态(Modality)]_[物种_器官]_[分辨率].tif
标签文件：Labels - [成像模态(Modality)]_[物种_器官]_[分辨率]_labels.tif
大型数据集的子体积块：[成像模态(Modality)]_[物种_器官]_subvolume[像素维度].tif

血管的人工标注工作使用Amira (2020.2, Thermo-Fisher, UK) 完成。

**训练数据集：**

opticalHREM_murineLiver_2.26x2.26x1.75um.tif：该样本为高分辨率体视显微镜（HREM）数据集，通过伊红B染色健康小鼠肝脏并采用标准HREM成像方案自主采集得到。*注：该图像体积的25%被预留作为测试数据，未参与训练。*

CT_murineTumour_20x20x20um.tif：该样本为小鼠结直肠癌皮下模型微血管铸型的X射线显微CT图像，由本实验室自主采集。*注：该图像体积的25%被预留作为测试数据，未参与训练。*

RSOM_murineTumour_20x20um.tif：该样本为光栅扫描光声显微内镜（RSOM）数据，来自皮下肿瘤模型，由剑桥大学Bohndiek课题组Emma Brown提供。图像数据已进行背景降噪处理（Brown et al., 2019）。

OCTA_humanRetina_24x24um.tif：该样本为光学相干断层扫描血管造影（OCTA）获取的视网膜血管造影数据，由Moorfields眼科医院Ranjan Rajendram博士提供。

**测试数据集：**

MRI_porcineLiver_0.9x0.9x5mm.tif：该样本为机器灌注猪肝脏的T1加权平衡快速场回波磁共振成像（MRI）数据，由本实验室自主采集。

MFHREM_murineTumourLectin_2.76x2.76x2.61um.tif：该样本为使用多荧光高分辨率体视显微镜自主采集的结直肠癌皮下小鼠模型图像，通过结合Dylight 647标记的凝集素对血管进行染色（Walsh et al., 2021）。图像数据已采用Walsh等人2020年提出的非对称反卷积算法进行处理。*注：其中一个尺寸为480×480×640体素的子体积块已完成人工标注（对应文件为MFHREM_murineTumourLectin_subvolume480x480x640.tif）。*

MFHREM_murineBrainLectin_0.85x0.85x0.86um.tif：该样本为使用多荧光高分辨率体视显微镜采集的小鼠大脑皮层图像，通过结合Dylight-647标记的凝集素进行染色（Walsh et al., 2021）。图像数据已进行下采样，并采用Walsh等人2020年提出的非对称反卷积算法处理。*注：其中一个尺寸为1000×1000×99体素的子体积块已完成人工标注。该子体积块以全分辨率且未经过预处理的形式提供，对应文件为MFHREM_murineBrainLectin_subvol_0.57x0.57x0.86um.tif。*

2Photon_murineOlfactoryBulbLectin_0.2x0.46x5.2um.tif：该样本为使用磺基罗丹明101标记的小鼠嗅球血管双光子成像数据，由弗朗西斯·克里克研究所神经环路与神经技术实验室的张宇欣慷慨提供（Bosch et al., 2022）。*注：其中一个尺寸为500×500×79体素的子体积块已完成人工标注（对应文件为2Photon_murineOlfactoryBulbLectin_subvolume500x500x79.tif）。*

**参考文献：**

1. Bosch, C. 等人 (2022). 通过关联在体生理学、同步辐射显微CT与体积电子显微镜对脑组织开展功能与多尺度三维结构研究. *Nature Communications*, 2022, 13(1): 1-16. https://doi.org/10.1038/s41467-022-30199-6

2. Brown, E., Brunker, J., & Bohndiek, S. E. (2019). 光声成像作为探查肿瘤微环境的工具. *DMM Disease Models and Mechanisms*, 12(7). https://doi.org/10.1242/DMM.039636

3. Walsh, C. 等人 (2021). 用于成年小鼠器官三维成像的多荧光高分辨率体视显微镜技术. *Advanced Photonics Research*, 2(10): e2100110. https://doi.org/10.1002/ADPR.202100110

4. Walsh, C., Holroyd, N., Shipley, R., & Walker-Samuel, S. (2020). 串行切片块面成像的非对称点扩散函数估计与反卷积. *Communications in Computer and Information Science*, 1248 CCIS: 235-249. https://doi.org/10.1007/978-3-030-52791-4_19