https://www.na-mic.org/w/api.php?action=feedcontributions&feedformat=atom&user=FolkaNAMIC Wiki - User contributions [en]2026-04-11T17:33:20ZUser contributionsMediaWiki 1.33.0https://www.na-mic.org/w/index.php?title=Project_Week_25/CNN_for_PseudoCT_Generation_from_T1T2_MR&diff=96893Project Week 25/CNN for PseudoCT Generation from T1T2 MR2017-06-30T10:14:19Z<p>Folka: /* Project Description */</p>
<hr />
<div>__NOTOC__<br />
<br />
Back to [[Project_Week_25#Projects|Projects List]]<br />
<br />
==Key Investigators==<br />
<!-- Key Investigator bullet points --><br />
*[http://www.imagenglab.com/pileggi_29.html Giampaolo Pileggi] (Magna Graecia University, Italy/German Cancer Research Center (DKFZ), Germany)<br />
*[http://www.imagenglab.com/newsite/paolo_zaffino/ Paolo Zaffino] (Magna Graecia University, Italy)<br />
*[http://www.imagenglab.com/newsite/salvatore_scaramuzzino/ Salvatore Scaramuzzino] (Magna Graecia University/ASL Vercelli, Italy)<br />
*[http://www.imagenglab.com/newsite/mf_spadea/ Maria Francesca Spadea] (Magna Graecia University, Italy)<br />
*[http://isgwww.cs.uni-magdeburg.de/cas/team.php Gino Gulamhussene] (University of Magdeburg, Germany)<br />
*[http://isgwww.cs.uni-magdeburg.de/isg/meyer.html Anneke Meyer] (University of Magdeburg, Germany)<br />
<br />
==Project Description==<br />
This tool allows the user to generate an HU map (Pseudo-CT) from T1/T2 input MRI. The tool is still in the early development stages with good early results in terms of Mean Absolute Error and Bias.<br />
<br />
{| class="wikitable"<br />
! style="text-align: left; width:27%" | Objective<br />
! style="text-align: left; width:27%" | Approach and Plan<br />
! style="text-align: left; width:27%" | Progress and Next Steps<br />
|- style="vertical-align:top;"<br />
|<!-- Objective bullet points --><br />
The main object is to understand which topology of CNN is the most suited for the Pseudo-CT generation task.<br />
|<!-- Approach and Plan bullet points --><br />
Test of different structures with training on low-res images in order to speed-up computational time.<br />
Analysis of the different output with MAE and Bias metrics<br />
|<!-- Progress and Next steps (fill out at the end of project week), bullet points --><br />
The most suited CNN to be used for this task consisted of an edited U-Net that does not make use of the final Dense Layer, useful for segmentation but not for regression purposes.<br />
Also unpooling layer was used in the decoding part, with zero-filling instead of repetitions for the missing values.<br />
Next planned steps will be the fine tuning of the CNN in order to lower the MAE and BIAS, the translation to TensorFlow (now the code uses Theano as a backend) and the data parallelization. <br />
|}<br />
<br />
==Illustrations==<br />
[[File:MRI2CT.png]]<br />
<br />
==Background and References==<br />
<!-- Use this space for information that may help people better understand your project, like links to papers, source code, or data --></div>Folkahttps://www.na-mic.org/w/index.php?title=Project_Week_25/CNN_for_PseudoCT_Generation_from_T1T2_MR&diff=96832Project Week 25/CNN for PseudoCT Generation from T1T2 MR2017-06-30T09:18:33Z<p>Folka: /* Project Description */</p>
<hr />
<div>__NOTOC__<br />
<br />
Back to [[Project_Week_25#Projects|Projects List]]<br />
<br />
==Key Investigators==<br />
<!-- Key Investigator bullet points --><br />
*[http://www.imagenglab.com/pileggi_29.html Giampaolo Pileggi] (Magna Graecia University, Italy/German Cancer Research Center (DKFZ), Germany)<br />
*[http://www.imagenglab.com/newsite/paolo_zaffino/ Paolo Zaffino] (Magna Graecia University, Italy)<br />
*[http://www.imagenglab.com/newsite/salvatore_scaramuzzino/ Salvatore Scaramuzzino] (Magna Graecia University/ASL Vercelli, Italy)<br />
*[http://www.imagenglab.com/newsite/mf_spadea/ Maria Francesca Spadea] (Magna Graecia University, Italy)<br />
*[http://isgwww.cs.uni-magdeburg.de/cas/team.php Gino Gulamhussene] (University of Magdeburg, Germany)<br />
*[http://isgwww.cs.uni-magdeburg.de/isg/meyer.html Anneke Meyer] (University of Magdeburg, Germany)<br />
<br />
==Project Description==<br />
This tool allows the user to generate an HU map (Pseudo-CT) from T1/T2 input MRI. The tool is still in the early development stages with good early results in terms of Mean Absolute Error and Bias.<br />
<br />
{| class="wikitable"<br />
! style="text-align: left; width:27%" | Objective<br />
! style="text-align: left; width:27%" | Approach and Plan<br />
! style="text-align: left; width:27%" | Progress and Next Steps<br />
|- style="vertical-align:top;"<br />
|<!-- Objective bullet points --><br />
The main object is to understand which topology of CNN is the most suited for the Pseudo-CT generation task.<br />
|<!-- Approach and Plan bullet points --><br />
Test of different structures with training on low-res images in order to speed-up computational time.<br />
Analysis of the different output with MAE and Bias metrics<br />
|<!-- Progress and Next steps (fill out at the end of project week), bullet points --><br />
The most suited CNN to be used resulted an edited U-Net that does not make use of the final Dense Layer, useful for segmentation but not for regression purpose.<br />
Also unpooling layer was used in the decoding part, with zero-filling instead of repetitions. <br />
|}<br />
<br />
==Illustrations==<br />
[[File:MRI2CT.png]]<br />
<br />
==Background and References==<br />
<!-- Use this space for information that may help people better understand your project, like links to papers, source code, or data --></div>Folka