Fast Improvement of TEM Images with Low-Dose Electrons by Deep Learning

<p>This is a dataset of High-Dose-Electron (HDE) images and Low-Dose-Electron (LDE) images taken with a transmission electron microscopy used in&nbsp; <a href="https://doi.org/10.1017/S1431927621013799">H. Katsuno, Y. Kimura, T. Yamazaki and I Takigawa, Microsc. Microanal. <strong>28</strong> (2022), pp 138--144</a>&nbsp;(<a href="https://arxiv.org/abs/2106.01718">arXiv:2106.01718</a>).</p> <p>There are two LDE images for each HDE image.</p> <p>cf) HDE image is 0001.tif and corresponding LDE images are 0002.tif and 0003.tif.</p> <p>&nbsp;</p> <p>Equipment of TEM:</p> <p>field-emission gun (JEM-2100F, JEOL, Tokyo)</p> <p>OneView IS (Gatan, Inc., Pleasanton, CA, USA)</p> <p>&nbsp;</p> <p>Typical magnification was 25,000x and 30,000x.</p> <p>&nbsp;</p> <p>HDE image</p> <p>The resolution was 4096 x 4096 pixels and its&nbsp;exposure time was 5 s.</p> <p>The typical total doses was 10¹⁰&nbsp;e^-.</p> <p>&nbsp;</p> <p>LDE image</p> <p>The resolution was 512 x 512 pixels and its&nbsp;exposure time was 3.3 ms.</p> <p>The typical total doses was 10⁶&nbsp;e^-.</p> <p>&nbsp;</p> <table> <tbody> <tr> <td>Filename</td> <td>Material</td> <td>Total number of images</td> <td>Total number of a pair of HDE and LDE</td> </tr> <tr> <td>train1_Ni.zip</td> <td>Ni</td> <td>336</td> <td>224</td> </tr> <tr> <td>train2_FeNi.zip</td> <td>FeNi</td> <td>390</td> <td>260</td> </tr> <tr> <td>train3_SiC.zip</td> <td>SiC</td> <td>210</td> <td>140</td> </tr> <tr> <td>train4_Silicate</td> <td>Silicate</td> <td>264</td> <td>176</td> </tr> <tr> <td>train5_Alumina</td> <td>Alumina</td> <td>300</td> <td>200</td> </tr> <tr> <td>val1_Ni.zip</td> <td>Ni</td> <td>54</td> <td>36</td> </tr> <tr> <td>val2_FeNi.zip</td> <td>FeNi</td> <td>60</td> <td>40</td> </tr> <tr> <td>val3_SiC.zip</td> <td>SiC</td> <td>30</td> <td>20</td> </tr> <tr> <td>val4_Silicate</td> <td>Silicate</td> <td>36</td> <td>24</td> </tr> <tr> <td>val5_Alumina</td> <td>Alumina</td> <td>60</td> <td>40</td> </tr> </tbody> </table> <p>&nbsp;</p> <p>&nbsp;</p> <p>The ipynb file and model parameters for machine learning are located in <a href="https://github.com/hiroyasukatsuno/Fast-Improvement-Low-Dose-TEMimages">the GitHub page</a>.</p> <p>&nbsp;</p>