Datasets available for downloading:
Fibroblasts (PROLIF/SS)
3D confocal images of nuclei and nucleoli in primary human fibroblasts, under proliferating and serum-starved conditions.
PC3 (EPI/EMT)
3D confocal images of nuclei and nucleoli in the PC3 human prostate cancer cell line, in epithelial and epithelial-to-mesenchymal transision states.
Caco-2
3D confocal images of nuclei in the Caco-2 human colon cancer cell line.
Astrocytes (NHA/VPA)
3D confocal images of primary human hippocampal astrocyte nuclei under untreated (NHA) and
valproic acid-treated (VPA) states, across 3 timepoints.
LICENSE:
Copyright © 2016- Regents of the University of Michigan
Creative Commons Attribution 4.0 International LicenseThe Fibroblast dataset consists of 3D fluorescence microscopy images of primary human fibroblasts (newborn male). The dataset contains fibroblasts in two different phenotypic states: cell cycle G0/G1 synchronized by serum-starvation (SS) and proliferating (PROLIF). Cells were labeled with DAPI (4',6-diamidino-2-phenylindole) for the nuclei, and fibrillarin antibody (fibrillarin) and ethidium bromide (EtBr) for nucleoli. Original 3-channel volumes were aquired using Zeiss LSM 710 laser scanning confocal microscope with a 63x PLAN/Apochromat 1.4NA DIC objective and then cut into Zx1024x1024 sub-volumes with channels separated into c0, c1, and c2, correspondingly. The dataset also includes segmentation results: 965 nuclear masks (498 SS and 470 PROLIF) and 2,181 nucleolar masks (1,151 SS and 1,030 PROLIF).
Sample preparation protocols: PDF
WARNING: each directory will take up to 20-30GB after extraction.
157 (1.25 GB)
158 (1.38 GB)
159 (1.42 GB)
163 (1.75 GB)
SS:160 (1.43 GB)
161 (1.43 GB)
162 (1.50 GB)
164 (1.53 GB)
165 (1.25 GB)
166 (1.22 GB)
167 (1.59 GB)
Citation:
Kalinin AA, Allyn-Feuer A, Ade A, Fon GV, Meixner W, Dilworth D, de Wet JR, Higgins GA, Zheng G, Creekmore A, Wiley JW, Verdone JE, Veltri RW, Pienta KJ, Coffey DS, Athey BD, Dinov ID. 2018. 3D cell nuclear morphology: microscopy imaging dataset and voxel-based morphometry classification results. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). doi: 10.1109/CVPRW.2018.00304
The PC3 dataset consists of 3D fluorescence microscopy images of human prostate cancer cell line PC3. Slides were cultured in two different phenotypic states: epithelial (EPI) and epithelial-to-mesenchymal transition (EMT). Then cells were labeled with DAPI (4',6-diamidino-2-phenylindole) for the nuclei, and fibrillarin antibody (fibrillarin) and ethidium bromide (EtBr) for nucleoli. Original 3-channel volumes were aquired using Zeiss LSM 710 laser scanning confocal microscope with a 63x PLAN/Apochromat 1.4NA DIC objective and then cut into Zx1024x1024 sub-volumes with channels separated into c0, c1, and c2, correspondingly. The dataset also includes segmentation results: 458 nuclear (310 EPI and 148 EMT) and 1,101 nucleolar (649 EPI and 452 EMT) masks.
Sample preparation protocols: PDF
Citation:
Kalinin AA, Allyn-Feuer A, Ade A, Fon GV, Meixner W, Dilworth D, de Wet JR, Higgins GA, Zheng G, Creekmore A, Wiley JW, Verdone JE, Veltri RW, Pienta KJ, Coffey DS, Athey BD, Dinov ID. 2018. 3D cell nuclear morphology: microscopy imaging dataset and voxel-based morphometry classification results. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). doi: 10.1109/CVPRW.2018.00304
The Caco2 dataset consists of 103 3D fluorescence microscopy images of human colon cancer cell line Caco-2. Cells were labeled with 4 different fluorophores: DAPI (4',6-diamidino-2-phenylindole) for the nuclei, HES1 Alexa 647 (1:200; Abcam), NR3C1 Alexa 488 (1:150; Cell signaling) and OCLN Alexa 594 (1:600; Thermo Fisher). Original multichannel volumes were aquired using laser scanning confocal microscope (see the paper for details). Images were saved in Analyze image format, in which each 3D image is represented by a pair of files: .img and .hdr.
WARNING: Full dataset takes ~120GB after unpacking.
NOTE: Any 3D image in either Nifti and Analyze formats can be viewed interactively in web-browser using SOCR BrainViewer web-app. To use BrainViewer, drag and drop an image (.nii.gz or .img+.hdr) into a browser window or press 'Select files' button. Use controls to switch between 2D and 3D views.
Single image in Nifti format (2 MB)
Sample images in Nifti format (243 MB)
Full dataset in Analyze format (14 GB)
Citation:
Zheng G, Kalinin AA, Dinov ID, Meixner W, Zhu S, Wiley JW. 2018. Hypothesis: Caco-2 cell rotational 3D mechanogenomic Turing patterns has clinical implications to colon crypts. Journal of Cellular and Molecular Medicine, 22(12), 6380-6385. doi: 10.1111/jcmm.13853
The Astrocytes dataset consists of 3D fluorescence microscopy images of primary human astrocytes (hippocampal). The dataset contains fibroblasts imaged at three timepoints (days 3, 5, and 7) and in two different phenotypic states: normal human astrocytes cutured in DMSO (NHA) and astrocytes treated with 1.5 mM valproic acid (VPA). Cell nuclei were labeled with DAPI (4',6-diamidino-2-phenylindole) and imaged using Zeiss LSM 710 laser scanning confocal microscope with a 63x PLAN/Apochromat 1.4 NA DIC objective. Original 3D volumes were then cut into Zx1024x1024 sub-volumes and deconvolved using the Lucy-Richardson deconvolution algorithm (see paper for details). The dataset also includes segmented nuclei: 512 NHA masks and 320 VPA masks.
Day 3: images and masks (7.6 GB total)
Day 5: images and masks (8.5 GB total)
Day 7: images and masks (4.5 GB total)
VPA:Day 3: images and masks (4.8 GB total)
Citation:
Kalinin AA, Hou X, Ade AS, Fon GV, Meixner W Higgins GA, Zheng G, Sexton JZ, Wan X, Athey BD, O’Meara MJ, Dinov ID. 2018. Valproic acid-induced changes of 4D nuclear morphology in astrocyte cells. Molecular Biology of the Cell, 2(18), 1624-1633. doi: 10.1091/mbc.E20-08-0502