Find research datasets worth reusing
Search datasets from major research repositories and use ShareScore to quickly assess how well each record supports discovery, access, and reuse.
285
datasets available to search
ShareScore release 0.9.0
Dataset results
285 results for “Machine learning dataset”
A large expert-curated cryo-EM image dataset for machine learning protein particle picking
<p>Cryo-electron microscopy (cryo-EM) is a powerful technique for determining the structures of biological macromolecular complexes. Picking single-protein particles from cryo-EM micrographs is a crucial step in reconstructing protein structures. However, the widely used template-based particle picking process is labor-intensive and time-consuming. Though machine learning and artificial intelligence (AI) based particle picking can potentially automate the process, its development is hindered by lack of large, high-quality labelled training data. To address this bottleneck, we present CryoPPP, a large, diverse, expert-curated cryo-EM image dataset for protein particle picking and analysis. It consists of labelled cryo-EM micrographs (images) of 34 representative protein datasets selected from the Electron Microscopy Public Image Archive (EMPIAR). The dataset is 2.6 terabytes and includes 9,893 high-resolution micrographs with labelled protein particle coordinates. The labelling process was rigorously validated through 2D particle class validation and 3D density map validation with the gold standard. The dataset is expected to greatly facilitate the development of both AI and classical methods for automated cryo-EM protein particle picking.</p>
2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning: Slices 3,001-4,000 (reference reconstructions and segmentations)
<p>This upload contains the reference reconstructions and segmentation of slices 3,001 – 4,000 from the data collection described in</p> <p>Maximilian B. Kiss, Sophia B. Coban, K. Joost Batenburg, Tristan van Leeuwen, and Felix Lucka “"2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning", <a href="https://doi.org/10.1038/s41597-023-02484-6"><em>Sci Data</em> <strong>10</strong>, 576 (2023)</a> or <a href="https://arxiv.org/abs/2306.05907">arXiv:2306.05907 (2023)</a></p> <p>Abstract:<br> "Recent research in computational imaging largely focuses on developing machine learning (ML) techniques for image reconstruction, which requires large-scale training datasets consisting of measurement data and ground-truth images. However, suitable experimental datasets for X-ray Computed Tomography (CT) are scarce, and methods are often developed and evaluated only on simulated data. We fill this gap by providing the community with a versatile, open 2D fan-beam CT dataset suitable for developing ML techniques for a range of image reconstruction tasks. To acquire it, we designed a sophisticated, semi-automatic scan procedure that utilizes a highly-flexible laboratory X-ray CT setup. A diverse mix of samples with high natural variability in shape and density was scanned slice-by-slice (5000 slices in total) with high angular and spatial resolution and three different beam characteristics: A high-fidelity, a low-dose and a beam-hardening-inflicted mode. In addition, 750 out-of-distribution slices were scanned with sample and beam variations to accommodate robustness and segmentation tasks. We provide raw projection data, reference reconstructions and segmentations based on an open-source data processing pipeline."</p> <p>The data collection has been acquired using a highly flexible, programmable and custom-built X-ray CT scanner, the FleX-ray scanner, developed by <a href="https://info.tescan.com/micro-ct">TESCAN-XRE NV,</a> located in the FleX-ray Lab at the <a href="https://www.cwi.nl/en/">Centrum Wiskunde & Informatica (CWI)</a> in Amsterdam, Netherlands. It consists of a cone-beam microfocus X-ray point source (limited to 90 kV and 90 W) that projects polychromatic X-rays onto a 14-bit CMOS (complementary metal-oxide semiconductor) flat panel detector with CsI(Tl) scintillator (Dexella 1512NDT) and 1536-by-1944 pixels, <span class="math-tex">\(74.8\mu m^2\)</span> each. To create a 2D dataset, a fan-beam geometry was mimicked by only reading out the central row of the detector. Between source and detector there is a rotation stage, upon which samples can be mounted. The machine components (i.e., the source, the detector panel, and the rotation stage) are mounted on translation belts that allow the moving of the components independently from one another.</p> <p>Please refer to the paper for all further technical details.</p> <p>The complete dataset can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8014758">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8014766">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8014787">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8014829">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8014874">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8014907">OOD</a>.<br> The reference reconstructions and segmentations can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8017583">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8017604">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8017612">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8017618">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8017624">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8017653">OOD</a>.</p> <p>The corresponding Python scripts for loading, pre-processing, reconstructing and segmenting the projection data in the way described in the paper can be found on <a href="https://github.com/mbkiss/2DeteCTcodes">github</a>. A machine-readable file with the used scanning parameters and instrument data for each acquisition mode as well as a script loading it can be found on the GitHub repository as well.</p> <p>Note: It is advisable to use the graphical user interface when decompressing the .zip archives. If you experience a zipbomb error when unzipping the file on a Linux system rerun the command with the UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE environment variable by setting in your .bashrc “export UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE”.</p> <p>For more information or guidance in using the data collection, please get in touch with</p> <p> Maximilian.Kiss [at] cwi.nl</p> <p> Felix.Lucka [at] cwi.nl</p>
2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning: Slices OOD (reference reconstructions and segmentations)
<p>This upload contains the reference reconstructions and segmentation of the out-of-distribution slices (OOD) from the data collection described in</p> <p>Maximilian B. Kiss, Sophia B. Coban, K. Joost Batenburg, Tristan van Leeuwen, and Felix Lucka “"2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning", <a href="https://doi.org/10.1038/s41597-023-02484-6"><em>Sci Data</em> <strong>10</strong>, 576 (2023)</a> or <a href="https://arxiv.org/abs/2306.05907">arXiv:2306.05907 (2023)</a></p> <p>Abstract:<br> "Recent research in computational imaging largely focuses on developing machine learning (ML) techniques for image reconstruction, which requires large-scale training datasets consisting of measurement data and ground-truth images. However, suitable experimental datasets for X-ray Computed Tomography (CT) are scarce, and methods are often developed and evaluated only on simulated data. We fill this gap by providing the community with a versatile, open 2D fan-beam CT dataset suitable for developing ML techniques for a range of image reconstruction tasks. To acquire it, we designed a sophisticated, semi-automatic scan procedure that utilizes a highly-flexible laboratory X-ray CT setup. A diverse mix of samples with high natural variability in shape and density was scanned slice-by-slice (5000 slices in total) with high angular and spatial resolution and three different beam characteristics: A high-fidelity, a low-dose and a beam-hardening-inflicted mode. In addition, 750 out-of-distribution slices were scanned with sample and beam variations to accommodate robustness and segmentation tasks. We provide raw projection data, reference reconstructions and segmentations based on an open-source data processing pipeline."</p> <p>The data collection has been acquired using a highly flexible, programmable and custom-built X-ray CT scanner, the FleX-ray scanner, developed by <a href="https://info.tescan.com/micro-ct">TESCAN-XRE NV,</a> located in the FleX-ray Lab at the <a href="https://www.cwi.nl/en/">Centrum Wiskunde & Informatica (CWI)</a> in Amsterdam, Netherlands. It consists of a cone-beam microfocus X-ray point source (limited to 90 kV and 90 W) that projects polychromatic X-rays onto a 14-bit CMOS (complementary metal-oxide semiconductor) flat panel detector with CsI(Tl) scintillator (Dexella 1512NDT) and 1536-by-1944 pixels, <span class="math-tex">\(74.8\mu m^2\)</span> each. To create a 2D dataset, a fan-beam geometry was mimicked by only reading out the central row of the detector. Between source and detector there is a rotation stage, upon which samples can be mounted. The machine components (i.e., the source, the detector panel, and the rotation stage) are mounted on translation belts that allow the moving of the components independently from one another.</p> <p>Please refer to the paper for all further technical details.</p> <p>The complete dataset can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8014758">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8014766">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8014787">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8014829">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8014874">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8014907">OOD</a>.<br> The reference reconstructions and segmentations can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8017583">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8017604">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8017612">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8017618">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8017624">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8017653">OOD</a>.</p> <p>The corresponding Python scripts for loading, pre-processing, reconstructing and segmenting the projection data in the way described in the paper can be found on <a href="https://github.com/mbkiss/2DeteCTcodes">github</a>. A machine-readable file with the used scanning parameters and instrument data for each acquisition mode as well as a script loading it can be found on the GitHub repository as well.</p> <p>Note: It is advisable to use the graphical user interface when decompressing the .zip archives. If you experience a zipbomb error when unzipping the file on a Linux system rerun the command with the UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE environment variable by setting in your .bashrc “export UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE”.</p> <p>For more information or guidance in using the data collection, please get in touch with</p> <p> Maximilian.Kiss [at] cwi.nl</p> <p> Felix.Lucka [at] cwi.nl</p>
2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning: Slices 1,001-2,000 (reference reconstructions and segmentations)
<p>This upload contains the reference reconstructions and segmentation of slices 1,001 – 2,000 from the data collection described in</p> <p>Maximilian B. Kiss, Sophia B. Coban, K. Joost Batenburg, Tristan van Leeuwen, and Felix Lucka “"2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning", <a href="https://doi.org/10.1038/s41597-023-02484-6"><em>Sci Data</em> <strong>10</strong>, 576 (2023)</a> or <a href="https://arxiv.org/abs/2306.05907">arXiv:2306.05907 (2023)</a></p> <p>Abstract:<br> "Recent research in computational imaging largely focuses on developing machine learning (ML) techniques for image reconstruction, which requires large-scale training datasets consisting of measurement data and ground-truth images. However, suitable experimental datasets for X-ray Computed Tomography (CT) are scarce, and methods are often developed and evaluated only on simulated data. We fill this gap by providing the community with a versatile, open 2D fan-beam CT dataset suitable for developing ML techniques for a range of image reconstruction tasks. To acquire it, we designed a sophisticated, semi-automatic scan procedure that utilizes a highly-flexible laboratory X-ray CT setup. A diverse mix of samples with high natural variability in shape and density was scanned slice-by-slice (5000 slices in total) with high angular and spatial resolution and three different beam characteristics: A high-fidelity, a low-dose and a beam-hardening-inflicted mode. In addition, 750 out-of-distribution slices were scanned with sample and beam variations to accommodate robustness and segmentation tasks. We provide raw projection data, reference reconstructions and segmentations based on an open-source data processing pipeline."</p> <p>The data collection has been acquired using a highly flexible, programmable and custom-built X-ray CT scanner, the FleX-ray scanner, developed by <a href="https://info.tescan.com/micro-ct">TESCAN-XRE NV,</a> located in the FleX-ray Lab at the <a href="https://www.cwi.nl/en/">Centrum Wiskunde & Informatica (CWI)</a> in Amsterdam, Netherlands. It consists of a cone-beam microfocus X-ray point source (limited to 90 kV and 90 W) that projects polychromatic X-rays onto a 14-bit CMOS (complementary metal-oxide semiconductor) flat panel detector with CsI(Tl) scintillator (Dexella 1512NDT) and 1536-by-1944 pixels, <span class="math-tex">\(74.8\mu m^2\)</span> each. To create a 2D dataset, a fan-beam geometry was mimicked by only reading out the central row of the detector. Between source and detector there is a rotation stage, upon which samples can be mounted. The machine components (i.e., the source, the detector panel, and the rotation stage) are mounted on translation belts that allow the moving of the components independently from one another.</p> <p>Please refer to the paper for all further technical details.</p> <p>The complete dataset can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8014758">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8014766">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8014787">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8014829">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8014874">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8014907">OOD</a>.<br> The reference reconstructions and segmentations can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8017583">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8017604">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8017612">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8017618">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8017624">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8017653">OOD</a>.</p> <p>The corresponding Python scripts for loading, pre-processing, reconstructing and segmenting the projection data in the way described in the paper can be found on <a href="https://github.com/mbkiss/2DeteCTcodes">github</a>. A machine-readable file with the used scanning parameters and instrument data for each acquisition mode as well as a script loading it can be found on the GitHub repository as well.</p> <p>Note: It is advisable to use the graphical user interface when decompressing the .zip archives. If you experience a zipbomb error when unzipping the file on a Linux system rerun the command with the UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE environment variable by setting in your .bashrc “export UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE”.</p> <p>For more information or guidance in using the data collection, please get in touch with</p> <p> Maximilian.Kiss [at] cwi.nl</p> <p> Felix.Lucka [at] cwi.nl</p>
2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning: Slices 4,001-5,000
<p>This upload contains slices 4,001 – 5,000 from the data collection described in</p> <p>Maximilian B. Kiss, Sophia B. Coban, K. Joost Batenburg, Tristan van Leeuwen, and Felix Lucka “"2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning", <a href="https://doi.org/10.1038/s41597-023-02484-6"><em>Sci Data</em> <strong>10</strong>, 576 (2023)</a> or <a href="https://arxiv.org/abs/2306.05907">arXiv:2306.05907 (2023)</a></p> <p>Abstract:<br> "Recent research in computational imaging largely focuses on developing machine learning (ML) techniques for image reconstruction, which requires large-scale training datasets consisting of measurement data and ground-truth images. However, suitable experimental datasets for X-ray Computed Tomography (CT) are scarce, and methods are often developed and evaluated only on simulated data. We fill this gap by providing the community with a versatile, open 2D fan-beam CT dataset suitable for developing ML techniques for a range of image reconstruction tasks. To acquire it, we designed a sophisticated, semi-automatic scan procedure that utilizes a highly-flexible laboratory X-ray CT setup. A diverse mix of samples with high natural variability in shape and density was scanned slice-by-slice (5000 slices in total) with high angular and spatial resolution and three different beam characteristics: A high-fidelity, a low-dose and a beam-hardening-inflicted mode. In addition, 750 out-of-distribution slices were scanned with sample and beam variations to accommodate robustness and segmentation tasks. We provide raw projection data, reference reconstructions and segmentations based on an open-source data processing pipeline."</p> <p>The data collection has been acquired using a highly flexible, programmable and custom-built X-ray CT scanner, the FleX-ray scanner, developed by <a href="https://info.tescan.com/micro-ct">TESCAN-XRE NV,</a> located in the FleX-ray Lab at the <a href="https://www.cwi.nl/en/">Centrum Wiskunde & Informatica (CWI)</a> in Amsterdam, Netherlands. It consists of a cone-beam microfocus X-ray point source (limited to 90 kV and 90 W) that projects polychromatic X-rays onto a 14-bit CMOS (complementary metal-oxide semiconductor) flat panel detector with CsI(Tl) scintillator (Dexella 1512NDT) and 1536-by-1944 pixels, <span class="math-tex">\(74.8\mu m^2\)</span> each. To create a 2D dataset, a fan-beam geometry was mimicked by only reading out the central row of the detector. Between source and detector there is a rotation stage, upon which samples can be mounted. The machine components (i.e., the source, the detector panel, and the rotation stage) are mounted on translation belts that allow the moving of the components independently from one another.</p> <p>Please refer to the paper for all further technical details.</p> <p>The complete dataset can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8014758">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8014766">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8014787">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8014829">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8014874">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8014907">OOD</a>.<br> The reference reconstructions and segmentations can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8017583">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8017604">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8017612">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8017618">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8017624">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8017653">OOD</a>.</p> <p>The corresponding Python scripts for loading, pre-processing, reconstructing and segmenting the projection data in the way described in the paper can be found on <a href="https://github.com/mbkiss/2DeteCTcodes">github</a>. A machine-readable file with the used scanning parameters and instrument data for each acquisition mode as well as a script loading it can be found on the GitHub repository as well.</p> <p>Note: It is advisable to use the graphical user interface when decompressing the .zip archives. If you experience a zipbomb error when unzipping the file on a Linux system rerun the command with the UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE environment variable by setting in your .bashrc “export UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE”.</p> <p>For more information or guidance in using the data collection, please get in touch with</p> <p> Maximilian.Kiss [at] cwi.nl</p> <p> Felix.Lucka [at] cwi.nl</p>
2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning: Slices 3,001-4,000
<p>This upload contains slices 3,001 – 4,000 from the data collection described in</p> <p>Maximilian B. Kiss, Sophia B. Coban, K. Joost Batenburg, Tristan van Leeuwen, and Felix Lucka “"2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning", <a href="https://doi.org/10.1038/s41597-023-02484-6"><em>Sci Data</em> <strong>10</strong>, 576 (2023)</a> or <a href="https://arxiv.org/abs/2306.05907">arXiv:2306.05907 (2023)</a></p> <p>Abstract:<br> "Recent research in computational imaging largely focuses on developing machine learning (ML) techniques for image reconstruction, which requires large-scale training datasets consisting of measurement data and ground-truth images. However, suitable experimental datasets for X-ray Computed Tomography (CT) are scarce, and methods are often developed and evaluated only on simulated data. We fill this gap by providing the community with a versatile, open 2D fan-beam CT dataset suitable for developing ML techniques for a range of image reconstruction tasks. To acquire it, we designed a sophisticated, semi-automatic scan procedure that utilizes a highly-flexible laboratory X-ray CT setup. A diverse mix of samples with high natural variability in shape and density was scanned slice-by-slice (5000 slices in total) with high angular and spatial resolution and three different beam characteristics: A high-fidelity, a low-dose and a beam-hardening-inflicted mode. In addition, 750 out-of-distribution slices were scanned with sample and beam variations to accommodate robustness and segmentation tasks. We provide raw projection data, reference reconstructions and segmentations based on an open-source data processing pipeline."</p> <p>The data collection has been acquired using a highly flexible, programmable and custom-built X-ray CT scanner, the FleX-ray scanner, developed by <a href="https://info.tescan.com/micro-ct">TESCAN-XRE NV,</a> located in the FleX-ray Lab at the <a href="https://www.cwi.nl/en/">Centrum Wiskunde & Informatica (CWI)</a> in Amsterdam, Netherlands. It consists of a cone-beam microfocus X-ray point source (limited to 90 kV and 90 W) that projects polychromatic X-rays onto a 14-bit CMOS (complementary metal-oxide semiconductor) flat panel detector with CsI(Tl) scintillator (Dexella 1512NDT) and 1536-by-1944 pixels, <span class="math-tex">\(74.8\mu m^2\)</span> each. To create a 2D dataset, a fan-beam geometry was mimicked by only reading out the central row of the detector. Between source and detector there is a rotation stage, upon which samples can be mounted. The machine components (i.e., the source, the detector panel, and the rotation stage) are mounted on translation belts that allow the moving of the components independently from one another.</p> <p>Please refer to the paper for all further technical details.</p> <p>The complete dataset can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8014758">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8014766">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8014787">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8014829">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8014874">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8014907">OOD</a>.<br> The reference reconstructions and segmentations can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8017583">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8017604">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8017612">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8017618">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8017624">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8017653">OOD</a>.</p> <p>The corresponding Python scripts for loading, pre-processing, reconstructing and segmenting the projection data in the way described in the paper can be found on <a href="https://github.com/mbkiss/2DeteCTcodes">github</a>. A machine-readable file with the used scanning parameters and instrument data for each acquisition mode as well as a script loading it can be found on the GitHub repository as well.</p> <p>Note: It is advisable to use the graphical user interface when decompressing the .zip archives. If you experience a zipbomb error when unzipping the file on a Linux system rerun the command with the UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE environment variable by setting in your .bashrc “export UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE”.</p> <p>For more information or guidance in using the data collection, please get in touch with</p> <p> Maximilian.Kiss [at] cwi.nl</p> <p> Felix.Lucka [at] cwi.nl</p>
2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning: Slices 1-1,000
<p>This upload contains slices 1 – 1,000 from the data collection described in</p> <p>Maximilian B. Kiss, Sophia B. Coban, K. Joost Batenburg, Tristan van Leeuwen, and Felix Lucka “"2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning", <a href="https://doi.org/10.1038/s41597-023-02484-6"><em>Sci Data</em> <strong>10</strong>, 576 (2023)</a> or <a href="https://arxiv.org/abs/2306.05907">arXiv:2306.05907 (2023)</a></p> <p>Abstract:<br> "Recent research in computational imaging largely focuses on developing machine learning (ML) techniques for image reconstruction, which requires large-scale training datasets consisting of measurement data and ground-truth images. However, suitable experimental datasets for X-ray Computed Tomography (CT) are scarce, and methods are often developed and evaluated only on simulated data. We fill this gap by providing the community with a versatile, open 2D fan-beam CT dataset suitable for developing ML techniques for a range of image reconstruction tasks. To acquire it, we designed a sophisticated, semi-automatic scan procedure that utilizes a highly-flexible laboratory X-ray CT setup. A diverse mix of samples with high natural variability in shape and density was scanned slice-by-slice (5000 slices in total) with high angular and spatial resolution and three different beam characteristics: A high-fidelity, a low-dose and a beam-hardening-inflicted mode. In addition, 750 out-of-distribution slices were scanned with sample and beam variations to accommodate robustness and segmentation tasks. We provide raw projection data, reference reconstructions and segmentations based on an open-source data processing pipeline."</p> <p>The data collection has been acquired using a highly flexible, programmable and custom-built X-ray CT scanner, the FleX-ray scanner, developed by <a href="https://info.tescan.com/micro-ct">TESCAN-XRE NV,</a> located in the FleX-ray Lab at the <a href="https://www.cwi.nl/en/">Centrum Wiskunde & Informatica (CWI)</a> in Amsterdam, Netherlands. It consists of a cone-beam microfocus X-ray point source (limited to 90 kV and 90 W) that projects polychromatic X-rays onto a 14-bit CMOS (complementary metal-oxide semiconductor) flat panel detector with CsI(Tl) scintillator (Dexella 1512NDT) and 1536-by-1944 pixels, <span class="math-tex">\(74.8\mu m^2\)</span> each. To create a 2D dataset, a fan-beam geometry was mimicked by only reading out the central row of the detector. Between source and detector there is a rotation stage, upon which samples can be mounted. The machine components (i.e., the source, the detector panel, and the rotation stage) are mounted on translation belts that allow the moving of the components independently from one another.</p> <p>Please refer to the paper for all further technical details.</p> <p>The complete dataset can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8014758">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8014766">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8014787">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8014829">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8014874">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8014907">OOD</a>.<br> The reference reconstructions and segmentations can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8017583">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8017604">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8017612">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8017618">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8017624">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8017653">OOD</a>.</p> <p>The corresponding Python scripts for loading, pre-processing, reconstructing and segmenting the projection data in the way described in the paper can be found on <a href="https://github.com/mbkiss/2DeteCTcodes">github</a>. A machine-readable file with the used scanning parameters and instrument data for each acquisition mode as well as a script loading it can be found on the GitHub repository as well.</p> <p>Note: It is advisable to use the graphical user interface when decompressing the .zip archives. If you experience a zipbomb error when unzipping the file on a Linux system rerun the command with the UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE environment variable by setting in your .bashrc “export UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE”.</p> <p>For more information or guidance in using the data collection, please get in touch with</p> <p> Maximilian.Kiss [at] cwi.nl</p> <p> Felix.Lucka [at] cwi.nl</p>
2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning: Slices 1-1,000 (reference reconstructions and segmentations)
<p>This upload contains the reference reconstructions and segmentation of slices 1 – 1,000 from the data collection described in</p> <p>Maximilian B. Kiss, Sophia B. Coban, K. Joost Batenburg, Tristan van Leeuwen, and Felix Lucka “"2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning", <a href="https://doi.org/10.1038/s41597-023-02484-6"><em>Sci Data</em> <strong>10</strong>, 576 (2023)</a> or <a href="https://arxiv.org/abs/2306.05907">arXiv:2306.05907 (2023)</a></p> <p>Abstract:<br> "Recent research in computational imaging largely focuses on developing machine learning (ML) techniques for image reconstruction, which requires large-scale training datasets consisting of measurement data and ground-truth images. However, suitable experimental datasets for X-ray Computed Tomography (CT) are scarce, and methods are often developed and evaluated only on simulated data. We fill this gap by providing the community with a versatile, open 2D fan-beam CT dataset suitable for developing ML techniques for a range of image reconstruction tasks. To acquire it, we designed a sophisticated, semi-automatic scan procedure that utilizes a highly-flexible laboratory X-ray CT setup. A diverse mix of samples with high natural variability in shape and density was scanned slice-by-slice (5000 slices in total) with high angular and spatial resolution and three different beam characteristics: A high-fidelity, a low-dose and a beam-hardening-inflicted mode. In addition, 750 out-of-distribution slices were scanned with sample and beam variations to accommodate robustness and segmentation tasks. We provide raw projection data, reference reconstructions and segmentations based on an open-source data processing pipeline."</p> <p>The data collection has been acquired using a highly flexible, programmable and custom-built X-ray CT scanner, the FleX-ray scanner, developed by <a href="https://info.tescan.com/micro-ct">TESCAN-XRE NV,</a> located in the FleX-ray Lab at the <a href="https://www.cwi.nl/en/">Centrum Wiskunde & Informatica (CWI)</a> in Amsterdam, Netherlands. It consists of a cone-beam microfocus X-ray point source (limited to 90 kV and 90 W) that projects polychromatic X-rays onto a 14-bit CMOS (complementary metal-oxide semiconductor) flat panel detector with CsI(Tl) scintillator (Dexella 1512NDT) and 1536-by-1944 pixels, <span class="math-tex">\(74.8\mu m^2\)</span> each. To create a 2D dataset, a fan-beam geometry was mimicked by only reading out the central row of the detector. Between source and detector there is a rotation stage, upon which samples can be mounted. The machine components (i.e., the source, the detector panel, and the rotation stage) are mounted on translation belts that allow the moving of the components independently from one another.</p> <p>Please refer to the paper for all further technical details.</p> <p>The complete dataset can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8014758">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8014766">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8014787">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8014829">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8014874">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8014907">OOD</a>.<br> The reference reconstructions and segmentations can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8017583">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8017604">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8017612">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8017618">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8017624">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8017653">OOD</a>.</p> <p>The corresponding Python scripts for loading, pre-processing, reconstructing and segmenting the projection data in the way described in the paper can be found on <a href="https://github.com/mbkiss/2DeteCTcodes">github</a>. A machine-readable file with the used scanning parameters and instrument data for each acquisition mode as well as a script loading it can be found on the GitHub repository as well.</p> <p>Note: It is advisable to use the graphical user interface when decompressing the .zip archives. If you experience a zipbomb error when unzipping the file on a Linux system rerun the command with the UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE environment variable by setting in your .bashrc “export UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE”.</p> <p>For more information or guidance in using the data collection, please get in touch with</p> <p> Maximilian.Kiss [at] cwi.nl</p> <p> Felix.Lucka [at] cwi.nl</p>
2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning: Slices OOD
<p>This upload contains the out-of-distribution slices (OOD) from the data collection described in</p> <p>Maximilian B. Kiss, Sophia B. Coban, K. Joost Batenburg, Tristan van Leeuwen, and Felix Lucka “"2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning", <a href="https://doi.org/10.1038/s41597-023-02484-6"><em>Sci Data</em> <strong>10</strong>, 576 (2023)</a> or <a href="https://arxiv.org/abs/2306.05907">arXiv:2306.05907 (2023)</a></p> <p>Abstract:<br> "Recent research in computational imaging largely focuses on developing machine learning (ML) techniques for image reconstruction, which requires large-scale training datasets consisting of measurement data and ground-truth images. However, suitable experimental datasets for X-ray Computed Tomography (CT) are scarce, and methods are often developed and evaluated only on simulated data. We fill this gap by providing the community with a versatile, open 2D fan-beam CT dataset suitable for developing ML techniques for a range of image reconstruction tasks. To acquire it, we designed a sophisticated, semi-automatic scan procedure that utilizes a highly-flexible laboratory X-ray CT setup. A diverse mix of samples with high natural variability in shape and density was scanned slice-by-slice (5000 slices in total) with high angular and spatial resolution and three different beam characteristics: A high-fidelity, a low-dose and a beam-hardening-inflicted mode. In addition, 750 out-of-distribution slices were scanned with sample and beam variations to accommodate robustness and segmentation tasks. We provide raw projection data, reference reconstructions and segmentations based on an open-source data processing pipeline."</p> <p>The data collection has been acquired using a highly flexible, programmable and custom-built X-ray CT scanner, the FleX-ray scanner, developed by <a href="https://info.tescan.com/micro-ct">TESCAN-XRE NV,</a> located in the FleX-ray Lab at the <a href="https://www.cwi.nl/en/">Centrum Wiskunde & Informatica (CWI)</a> in Amsterdam, Netherlands. It consists of a cone-beam microfocus X-ray point source (limited to 90 kV and 90 W) that projects polychromatic X-rays onto a 14-bit CMOS (complementary metal-oxide semiconductor) flat panel detector with CsI(Tl) scintillator (Dexella 1512NDT) and 1536-by-1944 pixels, <span class="math-tex">\(74.8\mu m^2\)</span> each. To create a 2D dataset, a fan-beam geometry was mimicked by only reading out the central row of the detector. Between source and detector there is a rotation stage, upon which samples can be mounted. The machine components (i.e., the source, the detector panel, and the rotation stage) are mounted on translation belts that allow the moving of the components independently from one another.</p> <p>Please refer to the paper for all further technical details.</p> <p>The complete dataset can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8014758">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8014766">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8014787">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8014829">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8014874">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8014907">OOD</a>.<br> The reference reconstructions and segmentations can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8017583">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8017604">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8017612">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8017618">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8017624">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8017653">OOD</a>.</p> <p>The corresponding Python scripts for loading, pre-processing, reconstructing and segmenting the projection data in the way described in the paper can be found on <a href="https://github.com/mbkiss/2DeteCTcodes">github</a>. A machine-readable file with the used scanning parameters and instrument data for each acquisition mode as well as a script loading it can be found on the GitHub repository as well.</p> <p>Note: It is advisable to use the graphical user interface when decompressing the .zip archives. If you experience a zipbomb error when unzipping the file on a Linux system rerun the command with the UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE environment variable by setting in your .bashrc “export UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE”.</p> <p>For more information or guidance in using the data collection, please get in touch with</p> <p> Maximilian.Kiss [at] cwi.nl</p> <p> Felix.Lucka [at] cwi.nl</p>
2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning: Slices 2,001-3,000 (reference reconstructions and segmentations)
<p>This upload contains the reference reconstructions and segmentation of slices 2,001 – 3,000 from the data collection described in</p> <p>Maximilian B. Kiss, Sophia B. Coban, K. Joost Batenburg, Tristan van Leeuwen, and Felix Lucka “"2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning", <a href="https://doi.org/10.1038/s41597-023-02484-6"><em>Sci Data</em> <strong>10</strong>, 576 (2023)</a> or <a href="https://arxiv.org/abs/2306.05907">arXiv:2306.05907 (2023)</a></p> <p>Abstract:<br> "Recent research in computational imaging largely focuses on developing machine learning (ML) techniques for image reconstruction, which requires large-scale training datasets consisting of measurement data and ground-truth images. However, suitable experimental datasets for X-ray Computed Tomography (CT) are scarce, and methods are often developed and evaluated only on simulated data. We fill this gap by providing the community with a versatile, open 2D fan-beam CT dataset suitable for developing ML techniques for a range of image reconstruction tasks. To acquire it, we designed a sophisticated, semi-automatic scan procedure that utilizes a highly-flexible laboratory X-ray CT setup. A diverse mix of samples with high natural variability in shape and density was scanned slice-by-slice (5000 slices in total) with high angular and spatial resolution and three different beam characteristics: A high-fidelity, a low-dose and a beam-hardening-inflicted mode. In addition, 750 out-of-distribution slices were scanned with sample and beam variations to accommodate robustness and segmentation tasks. We provide raw projection data, reference reconstructions and segmentations based on an open-source data processing pipeline."</p> <p>The data collection has been acquired using a highly flexible, programmable and custom-built X-ray CT scanner, the FleX-ray scanner, developed by <a href="https://info.tescan.com/micro-ct">TESCAN-XRE NV,</a> located in the FleX-ray Lab at the <a href="https://www.cwi.nl/en/">Centrum Wiskunde & Informatica (CWI)</a> in Amsterdam, Netherlands. It consists of a cone-beam microfocus X-ray point source (limited to 90 kV and 90 W) that projects polychromatic X-rays onto a 14-bit CMOS (complementary metal-oxide semiconductor) flat panel detector with CsI(Tl) scintillator (Dexella 1512NDT) and 1536-by-1944 pixels, <span class="math-tex">\(74.8\mu m^2\)</span> each. To create a 2D dataset, a fan-beam geometry was mimicked by only reading out the central row of the detector. Between source and detector there is a rotation stage, upon which samples can be mounted. The machine components (i.e., the source, the detector panel, and the rotation stage) are mounted on translation belts that allow the moving of the components independently from one another.</p> <p>Please refer to the paper for all further technical details.</p> <p>The complete dataset can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8014758">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8014766">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8014787">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8014829">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8014874">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8014907">OOD</a>.<br> The reference reconstructions and segmentations can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8017583">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8017604">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8017612">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8017618">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8017624">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8017653">OOD</a>.</p> <p>The corresponding Python scripts for loading, pre-processing, reconstructing and segmenting the projection data in the way described in the paper can be found on <a href="https://github.com/mbkiss/2DeteCTcodes">github</a>. A machine-readable file with the used scanning parameters and instrument data for each acquisition mode as well as a script loading it can be found on the GitHub repository as well.</p> <p>Note: It is advisable to use the graphical user interface when decompressing the .zip archives. If you experience a zipbomb error when unzipping the file on a Linux system rerun the command with the UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE environment variable by setting in your .bashrc “export UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE”.</p> <p>For more information or guidance in using the data collection, please get in touch with</p> <p> Maximilian.Kiss [at] cwi.nl</p> <p> Felix.Lucka [at] cwi.nl</p>
2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning: Slices 2,001-3,000
<p>This upload contains slices 2,001 – 3,000 from the data collection described in</p> <p>Maximilian B. Kiss, Sophia B. Coban, K. Joost Batenburg, Tristan van Leeuwen, and Felix Lucka “"2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning", <a href="https://doi.org/10.1038/s41597-023-02484-6"><em>Sci Data</em> <strong>10</strong>, 576 (2023)</a> or <a href="https://arxiv.org/abs/2306.05907">arXiv:2306.05907 (2023)</a></p> <p>Abstract:<br> "Recent research in computational imaging largely focuses on developing machine learning (ML) techniques for image reconstruction, which requires large-scale training datasets consisting of measurement data and ground-truth images. However, suitable experimental datasets for X-ray Computed Tomography (CT) are scarce, and methods are often developed and evaluated only on simulated data. We fill this gap by providing the community with a versatile, open 2D fan-beam CT dataset suitable for developing ML techniques for a range of image reconstruction tasks. To acquire it, we designed a sophisticated, semi-automatic scan procedure that utilizes a highly-flexible laboratory X-ray CT setup. A diverse mix of samples with high natural variability in shape and density was scanned slice-by-slice (5000 slices in total) with high angular and spatial resolution and three different beam characteristics: A high-fidelity, a low-dose and a beam-hardening-inflicted mode. In addition, 750 out-of-distribution slices were scanned with sample and beam variations to accommodate robustness and segmentation tasks. We provide raw projection data, reference reconstructions and segmentations based on an open-source data processing pipeline."</p> <p>The data collection has been acquired using a highly flexible, programmable and custom-built X-ray CT scanner, the FleX-ray scanner, developed by <a href="https://info.tescan.com/micro-ct">TESCAN-XRE NV,</a> located in the FleX-ray Lab at the <a href="https://www.cwi.nl/en/">Centrum Wiskunde & Informatica (CWI)</a> in Amsterdam, Netherlands. It consists of a cone-beam microfocus X-ray point source (limited to 90 kV and 90 W) that projects polychromatic X-rays onto a 14-bit CMOS (complementary metal-oxide semiconductor) flat panel detector with CsI(Tl) scintillator (Dexella 1512NDT) and 1536-by-1944 pixels, <span class="math-tex">\(74.8\mu m^2\)</span> each. To create a 2D dataset, a fan-beam geometry was mimicked by only reading out the central row of the detector. Between source and detector there is a rotation stage, upon which samples can be mounted. The machine components (i.e., the source, the detector panel, and the rotation stage) are mounted on translation belts that allow the moving of the components independently from one another.</p> <p>Please refer to the paper for all further technical details.</p> <p>The complete dataset can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8014758">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8014766">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8014787">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8014829">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8014874">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8014907">OOD</a>.<br> The reference reconstructions and segmentations can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8017583">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8017604">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8017612">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8017618">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8017624">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8017653">OOD</a>.</p> <p>The corresponding Python scripts for loading, pre-processing, reconstructing and segmenting the projection data in the way described in the paper can be found on <a href="https://github.com/mbkiss/2DeteCTcodes">github</a>. A machine-readable file with the used scanning parameters and instrument data for each acquisition mode as well as a script loading it can be found on the GitHub repository as well.</p> <p>Note: It is advisable to use the graphical user interface when decompressing the .zip archives. If you experience a zipbomb error when unzipping the file on a Linux system rerun the command with the UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE environment variable by setting in your .bashrc “export UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE”.</p> <p>For more information or guidance in using the data collection, please get in touch with</p> <p> Maximilian.Kiss [at] cwi.nl</p> <p> Felix.Lucka [at] cwi.nl</p>
2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning: Slices 1,001-2,000
<p>This upload contains slices 1,001 – 2,000 from the data collection described in</p> <p>Maximilian B. Kiss, Sophia B. Coban, K. Joost Batenburg, Tristan van Leeuwen, and Felix Lucka “"2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning", <a href="https://doi.org/10.1038/s41597-023-02484-6"><em>Sci Data</em> <strong>10</strong>, 576 (2023)</a> or <a href="https://arxiv.org/abs/2306.05907">arXiv:2306.05907 (2023)</a></p> <p>Abstract:<br> "Recent research in computational imaging largely focuses on developing machine learning (ML) techniques for image reconstruction, which requires large-scale training datasets consisting of measurement data and ground-truth images. However, suitable experimental datasets for X-ray Computed Tomography (CT) are scarce, and methods are often developed and evaluated only on simulated data. We fill this gap by providing the community with a versatile, open 2D fan-beam CT dataset suitable for developing ML techniques for a range of image reconstruction tasks. To acquire it, we designed a sophisticated, semi-automatic scan procedure that utilizes a highly-flexible laboratory X-ray CT setup. A diverse mix of samples with high natural variability in shape and density was scanned slice-by-slice (5000 slices in total) with high angular and spatial resolution and three different beam characteristics: A high-fidelity, a low-dose and a beam-hardening-inflicted mode. In addition, 750 out-of-distribution slices were scanned with sample and beam variations to accommodate robustness and segmentation tasks. We provide raw projection data, reference reconstructions and segmentations based on an open-source data processing pipeline."</p> <p>The data collection has been acquired using a highly flexible, programmable and custom-built X-ray CT scanner, the FleX-ray scanner, developed by <a href="https://info.tescan.com/micro-ct">TESCAN-XRE NV,</a> located in the FleX-ray Lab at the <a href="https://www.cwi.nl/en/">Centrum Wiskunde & Informatica (CWI)</a> in Amsterdam, Netherlands. It consists of a cone-beam microfocus X-ray point source (limited to 90 kV and 90 W) that projects polychromatic X-rays onto a 14-bit CMOS (complementary metal-oxide semiconductor) flat panel detector with CsI(Tl) scintillator (Dexella 1512NDT) and 1536-by-1944 pixels, <span class="math-tex">\(74.8\mu m^2\)</span> each. To create a 2D dataset, a fan-beam geometry was mimicked by only reading out the central row of the detector. Between source and detector there is a rotation stage, upon which samples can be mounted. The machine components (i.e., the source, the detector panel, and the rotation stage) are mounted on translation belts that allow the moving of the components independently from one another.</p> <p>Please refer to the paper for all further technical details.</p> <p>The complete dataset can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8014758">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8014766">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8014787">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8014829">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8014874">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8014907">OOD</a>.<br> The reference reconstructions and segmentations can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8017583">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8017604">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8017612">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8017618">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8017624">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8017653">OOD</a>.</p> <p>The corresponding Python scripts for loading, pre-processing, reconstructing and segmenting the projection data in the way described in the paper can be found on <a href="https://github.com/mbkiss/2DeteCTcodes">github</a>. A machine-readable file with the used scanning parameters and instrument data for each acquisition mode as well as a script loading it can be found on the GitHub repository as well.</p> <p>Note: It is advisable to use the graphical user interface when decompressing the .zip archives. If you experience a zipbomb error when unzipping the file on a Linux system rerun the command with the UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE environment variable by setting in your .bashrc “export UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE”.</p> <p>For more information or guidance in using the data collection, please get in touch with</p> <p> Maximilian.Kiss [at] cwi.nl</p> <p> Felix.Lucka [at] cwi.nl</p>
2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning: Slices 4,001-5,000 (reference reconstructions and segmentations)
<p>This upload contains the reference reconstructions and segmentation of slices 4,001 – 5,000 from the data collection described in</p> <p>Maximilian B. Kiss, Sophia B. Coban, K. Joost Batenburg, Tristan van Leeuwen, and Felix Lucka “"2DeteCT - A large 2D expandable, trainable, experimental Computed Tomography dataset for machine learning", <a href="https://doi.org/10.1038/s41597-023-02484-6"><em>Sci Data</em> <strong>10</strong>, 576 (2023)</a> or <a href="https://arxiv.org/abs/2306.05907">arXiv:2306.05907 (2023)</a></p> <p>Abstract:<br> "Recent research in computational imaging largely focuses on developing machine learning (ML) techniques for image reconstruction, which requires large-scale training datasets consisting of measurement data and ground-truth images. However, suitable experimental datasets for X-ray Computed Tomography (CT) are scarce, and methods are often developed and evaluated only on simulated data. We fill this gap by providing the community with a versatile, open 2D fan-beam CT dataset suitable for developing ML techniques for a range of image reconstruction tasks. To acquire it, we designed a sophisticated, semi-automatic scan procedure that utilizes a highly-flexible laboratory X-ray CT setup. A diverse mix of samples with high natural variability in shape and density was scanned slice-by-slice (5000 slices in total) with high angular and spatial resolution and three different beam characteristics: A high-fidelity, a low-dose and a beam-hardening-inflicted mode. In addition, 750 out-of-distribution slices were scanned with sample and beam variations to accommodate robustness and segmentation tasks. We provide raw projection data, reference reconstructions and segmentations based on an open-source data processing pipeline."</p> <p>The data collection has been acquired using a highly flexible, programmable and custom-built X-ray CT scanner, the FleX-ray scanner, developed by <a href="https://info.tescan.com/micro-ct">TESCAN-XRE NV,</a> located in the FleX-ray Lab at the <a href="https://www.cwi.nl/en/">Centrum Wiskunde & Informatica (CWI)</a> in Amsterdam, Netherlands. It consists of a cone-beam microfocus X-ray point source (limited to 90 kV and 90 W) that projects polychromatic X-rays onto a 14-bit CMOS (complementary metal-oxide semiconductor) flat panel detector with CsI(Tl) scintillator (Dexella 1512NDT) and 1536-by-1944 pixels, <span class="math-tex">\(74.8\mu m^2\)</span> each. To create a 2D dataset, a fan-beam geometry was mimicked by only reading out the central row of the detector. Between source and detector there is a rotation stage, upon which samples can be mounted. The machine components (i.e., the source, the detector panel, and the rotation stage) are mounted on translation belts that allow the moving of the components independently from one another.</p> <p>Please refer to the paper for all further technical details.</p> <p>The complete dataset can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8014758">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8014766">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8014787">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8014829">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8014874">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8014907">OOD</a>.<br> The reference reconstructions and segmentations can be found via the following links: <a href="https://doi.org/10.5281/zenodo.8017583">1-1000</a>, <a href="https://doi.org/10.5281/zenodo.8017604">1001-2000</a>, <a href="https://doi.org/10.5281/zenodo.8017612">2001-3000</a>, <a href="https://doi.org/10.5281/zenodo.8017618">3001-4000</a>, <a href="https://doi.org/10.5281/zenodo.8017624">4001-5000</a>, <a href="https://doi.org/10.5281/zenodo.8017653">OOD</a>.</p> <p>The corresponding Python scripts for loading, pre-processing, reconstructing and segmenting the projection data in the way described in the paper can be found on <a href="https://github.com/mbkiss/2DeteCTcodes">github</a>. A machine-readable file with the used scanning parameters and instrument data for each acquisition mode as well as a script loading it can be found on the GitHub repository as well.</p> <p>Note: It is advisable to use the graphical user interface when decompressing the .zip archives. If you experience a zipbomb error when unzipping the file on a Linux system rerun the command with the UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE environment variable by setting in your .bashrc “export UNZIP_DISABLE_ZIPBOMB_DETECTION=TRUE”.</p> <p>For more information or guidance in using the data collection, please get in touch with</p> <p> Maximilian.Kiss [at] cwi.nl</p> <p> Felix.Lucka [at] cwi.nl</p>
A dataset on in-situ electromagnetic wave integrity control of selective laser fusion printed parts using Machine Learning
<p><strong>DeepSLM Dataset</strong></p> <p>This dataset contains signals from metal parts printed in 3D using SLM (Selective laser melting) technology, produced as part of the DeepSLM project:<em> In-situ monitoring of Selective-Laser-Melted Ti6Al4V Parts Using Eddy Current Testing and Machine Learning</em>.</p> <p><em>Sensor and data collection</em></p> <p>An impedance-based non-destructive testing device is installed on the LPBF coating device. It is in the form of sensors with the following characteristics:</p> <p><em>Spectrum of analysis of the sensors</em></p> <ul> <li>Minimal length: 3-5 mm</li> <li>Minimal width acquisition: 2-3 mm (Without borders effects)</li> <li>Minimal number of layers: 50 layers</li> <li>Signal minimal penetration depth (TiAl<em>6</em>V<em>4</em>/875kHz) = 1.5 mm</li> </ul> <p>The sensors record the signal during printing, while simultaneously sending it to a computer via Bluetooth. The signals are first stored in the database. Next, a dataset is constructed from the extracted signals related to each part, and from the added semi-automatic annotations which describe the porosity of the part and layers.</p> <p><em>Annotations</em></p> <p>In the annotation process, there are 2 different types of labels and a metadata summary for these prints:</p> <p>1. Archimedean porosity measurement for the whole part<br>2. Layer porosity obtained after image processing applied on the resulting metallography images<br>3. Print-related metadata such as print parameters, part dimensions, layer thickness, etc</p> <p>Size: 66 parts<br>Set of printing parameters: 250</p> <p> --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------</p> <p><strong>Dataset organization</strong></p> <p>The dataset is composed of 7 different printings and is organized as follows:<br>- signals: contains collected signals<br>- metallography: contains porosity extracted by metallography<br>- archimedean: contains porosity extracted by archimedean weighing<br>- metallography-images: contains metallography process images<br>- metadata.csv: contains print metadata (print parameter, part name, strategy, etc.)</p> <p>Each printed part has a unique ID ranging from 0 to 65. Each sub-part has an ID which is the ID of the base part suffixed with the sub-part position: 14_1 is the first sub-part of part 14. This nomenclature is uniform throughout the dataset.</p> <p>The README.md file in the dataset provides more detailed information on the structure of the dataset and the format of each of the files making up the dataset.</p> <p> </p> <p><strong>Relative article</strong></p> <p><a title="https://doi.org/10.1007/978-3-031-47784-3_18" href="https://doi.org/10.1007/978-3-031-47784-3_18" target="_blank" rel="noreferrer noopener">https://doi.org/10.1007/978-3-031-47784-3_18</a></p> <p>Sallem, H., Ghorbel, H., Goffinet, E., Cinna, A., Pralong, J., Wicht, J., & Revaz, B. (2023, May). In-Situ Monitoring of Selective Laser Melted Ti–6Al–4V Parts Using Eddy Current Testing and Machine Learning. In <em>Advances In Additive Manufacturing Conference</em> (pp. 139-148). Cham: Springer Nature Switzerland.</p>
Dataset and scripts for publication "Property design of extruded magnesium-gadolinium alloys through machine learning"
<p>Data and scripts accompanying publication "Property design of extruded magnesium-gadolinium alloys through machine learning"</p>
Dataset for "Calibration of CAMS PM2.5 data over Hungary: A machine learning approach"
<p>This study utilized air quality information from eleven particular checking locales arranged all through Hungary. The informational index contains in-situ estimations of particulate matter with a width of 2.5 micrometers or less (PM2.5). Moreover, the dataset incorporates matching PM2.5 gauges from the Copernicus Climate Checking Administration (CAMS) model, which makes it a valuable asset for correlation and adjustment. The dataset incorporates various seasons, considering an exhaustive assessment of PM tainting under different climatic circumstances.</p>
Dataset for publication: "Using explainable machine learning to interpret the effects of policies on air pollution: COVID-19 lockdown in London"
<p>This online repository offers supplementary datasets supporting the findings in the research article titled "Using explainable machine learning to interpret the effects of policies on air pollution: COVID-19 lockdown in London," published in Environmental Science & Technology. The dataset contains air quality data from different monitoring sites in London between 2016 and 2020 and weather data for the same period. Additionally, the dataset incorporates 136 features relating to London's Middle Layer Super Output Areas (MSOAs) in the year 2019, which can be used to identify the key factors contributing to the heterogeneous changes in air quality levels at different spatial locations during the pandemic. Please refer to the metadata file for detailed data sources and descriptions. </p>
BubbleML: A Multi-Physics Dataset and Benchmarks for Machine Learning
<p>In the field of phase change phenomena, the lack of accessible and diverse datasets suitable for machine learning (ML) training poses a significant challenge. Existing experimental datasets are often restricted, with limited availability and sparse ground truth data, impeding our understanding of this complex multi-physics phenomena. To bridge this gap, we present the <a href="https://github.com/HPCForge/BubbleML">BubbleML</a> Dataset which leverages physics-driven simulations to provide accurate ground truth information for various boiling scenarios, encompassing nucleate pool boiling, flow boiling, and sub-cooled boiling. This extensive dataset covers a wide range of parameters, including varying gravity conditions, flow rates, sub-cooling levels, and wall superheat, comprising 79 simulations. BubbleML is validated against experimental observations and trends, establishing it as an invaluable resource for ML research. Furthermore, we showcase its potential to facilitate exploration of diverse downstream tasks by introducing two benchmarks: (a) optical flow analysis to capture bubble dynamics, and (b) operator networks for learning temperature dynamics. The BubbleML dataset and its benchmarks serve as a catalyst for advancements in ML-driven research on multi-physics phase change phenomena, enabling the development and comparison of state-of-the-art techniques and models.</p>
GPRChinaSPEI1km: High spatial resolution and century-long SPEI datasets for China from 1901 to 2020 generated by machine learning
<p>The high spatial resolution and century-long Standardized Precipitation Evapotranspiration Index (SPEI) dataset with a spatial resolution of 0.0083 degrees (~1 km) was spatially downscaled from the global SPEI data with a 0.5 degrees spatial resolution (https://spei.csic.es/database.html) based on machine learning integrated with high spatial resolution climatic and topographic variables. The 1-km SPEI datasets are across the land areas of China from January 1901 to December 2020, including 1-month, 3-month, 6-month and 12-month SPEIs. The unit of the data is 0.01. The dataset was evaluated using the root zone soil moisture and the historical drought events, and the evaluation indicated that the high spatial resolution SPEI dataset is reliable.</p> <p>Data Information: </p> <p>GPRChinaSPEI1km: High spatial resolution and century-long SPEI datasets over China from 1901 to 2020 generated by machine learning</p> <p>Publication: </p> <p><span>He, Q., Wang, M., Liu, K., & Wang, B. (2025). High-resolution Standardized Precipitation Evapotranspiration Index (SPEI) reveals trends in drought and vegetation water availability in China. <em>Geography and Sustainability</em>, <em>6</em>(2), 100228. https://doi.org/10.1016/j.geosus.2024.08.007</span></p> <p></p> <p>----------------------------------------------------data description---------------------------------------------</p> <p>This is a gridded dataset for the Standardized Precipitation Evapotranspiration Index (SPEI) at a spatial resolution of 1 km over the main terrestrial lands of China for each month during 1901-2020, which is generated using the Gaussian process regression (GPR) based on the Global SPEI database (https://spei.csic.es/database.html) integrated with high spatial resolution climatic and topographic variables. Four timescales of SPEI were generated: 1-month (SPEI-1), 3-month (SPEI-3), 6-month (SPEI-6) and 12-month (SPEI-12). The details are as follows:</p> <p>Region: China</p> <p>Temporal Extent: January 1901 to December 2020</p> <p>Spatial resolution: 0.0083° (~1 km)</p> <p>Temporal resolution: month</p> <p>Timescales: 1-month, 3-month, 6-month and 12-month</p> <p>Data format: GeoTIFF</p> <p>Unit: unitless (0.01)</p> <p>Geographic coordinate system: WGS 1984</p> <p>---------------------------------------------------dataset filename---------------------------------------------</p> <p>The file name specifically shows the data information.</p> <p>For example,</p> <p>“SPEI_1_2020_1.tif” means “1-month SPEI of January 2020”.</p> <p>“SPEI_3_2020_1.tif” means “3-month SPEI of January 2020”.</p> <p>All the file names are formatted in “SPEI_timescale_year_month”</p> <p>timescale: 1, 3, 6 and 12 indicate 1-month, 3-month, 6-month and 12-month, respectively</p> <p>year: from 1901 to 2020</p> <p>month: from 1 to 12</p> <p>--------------------------------------------------storage information-------------------------------------------</p> <p>The high-resolution SPEI dataset is stored in TIFF format using WGS 1984 coordinate system. The data type is int16 with a scale factor of 0.01. The nodata value is -32768. The dataset requires multiplication by 0.01 during application to obtain the actual value ranges.</p> <p>The data were compressed into .rar format every 10 years for each timescale SPEI.</p>
Dataset for "Enabling Machine Learning Models in Alarm Fatigue Research: Creation of a Large Relevance-annotated Oxygen Saturation Alarm Data Set"
<p>Chromik and Flint et al. (2024) (under review) propose an algorithm that uses clinical alarm logs, an annotation guideline (Klopfenstein et al. 2023), and routinely collected intensive care data to create a data set of relevance-annotated oxygen saturation alarms. We provide the algorithm's source code and data set of annotated oxygen saturation alarms as supplementary material to the publication.</p> <ul> <li>The algorithm's implementation is open-source and can be re-used on similar data sets.</li> <li>Our implementation used airway management data mappings to identify airway devices (AD), ventilation devices (VD), and ventilation modes (VM). These mappings can be found here: <a href="../doi/10.5281/zenodo.7511031">https://zenodo.org/doi/10.5281/zenodo.7511031</a></li> <li>The data set suggests that the majority of oxygen saturation alarms in the intensive care unit is non-actionable.</li> <li>We are the first to provide such an extensive data set of annotated oxygen saturation alarms.</li> </ul>
ScienceDex guides
Understand access before you commit
These curated guides explain access requirements, typical timelines, costs, and reuse considerations for widely used research datasets.
Allen Brain Atlas
Allen Brain Atlas is an Allen Institute collection of brain map atlases, datasets, APIs, and analysis tools covering mouse, human, and non-human primate brain resources.
Annotated Behaviour and Observability Dataset (ABODe)
ABODe is a University of Edinburgh DataShare dataset for behavior classification in group-housed mice using home-cage video, identities, bounding boxes, ground-plate positions, and annotator labels.
DANDI Archive for NWB datasets
DANDI is a BRAIN Initiative archive for publishing and sharing neurophysiology data, including electrophysiology, optophysiology, and behavioral data packaged as NWB and related standards.
International Brain Laboratory public data
The International Brain Laboratory public data releases expose standardized mouse decision-making experiments, including Neuropixels recordings, widefield calcium imaging, behavior, and session metadata accessed through the ONE API.
OpenNeuro
OpenNeuro is a free, open platform for sharing neuroimaging datasets, with public search, dataset pages, and download paths for web, S3, DataLad, and the OpenNeuro CLI.