Difference between revisions of "2016 Winter Project Week/Projects/ImageRestoration"

From NAMIC Wiki
Jump to: navigation, search
(Created page with '__NOTOC__ <gallery> Image:PW-MIT2015.png|Projects List Image:WMH_T1.png|Clinical Stroke Image </gallery> ==Key Investigators== - Adrian Dal…')
 
 
(10 intermediate revisions by 2 users not shown)
Line 1: Line 1:
 
__NOTOC__
 
__NOTOC__
 
<gallery>
 
<gallery>
Image:PW-MIT2015.png|[[2016_Winter_Project_Week#Projects|Projects List]]
+
Image:PW-MIT2016.png|[[2016_Winter_Project_Week#Projects|Projects List]]
 
Image:WMH_T1.png|Clinical Stroke Image
 
Image:WMH_T1.png|Clinical Stroke Image
 +
Image:gmm_interp.png|Running result: Original (linear-interp) data | our gmm-based interp | "true" high res data
 +
Image:gmmResult2.png|More results show our ability to improve images, but only on small areas.
 
</gallery>
 
</gallery>
  
 
==Key Investigators==
 
==Key Investigators==
- Adrian Dalca, Katie Bouman, Polina Golland, MIT
+
*Adrian Dalca (MIT)
 +
*Katie Bouman (MIT)
 +
*Polina Golland (MIT)
  
 
==Project Description==
 
==Project Description==
Line 16: Line 20:
 
To improve results for large datasets of clinical-quality data, we are investigating restoration methods without training datasets. Here, we are using a patch-based Gaussian Mixture Model approach with MRF priors and utilizing only the current dataset, without an external training dataset.  
 
To improve results for large datasets of clinical-quality data, we are investigating restoration methods without training datasets. Here, we are using a patch-based Gaussian Mixture Model approach with MRF priors and utilizing only the current dataset, without an external training dataset.  
  
 +
<strong>Essentially,  we explore a model where for a given location in all volumes of a dataset, we model those image patches as drawn from a particular mixture model.</strong>
  
<div style="width: 27%; float: left; padding-right: 3%;">
+
{| class="wikitable"
<h3>Objective</h3>
+
! style="text-align: left; width:27%" |  Objective
* We will investigate a current model for learning and using Patch-based Gaussian Mixture Model to restore sparse-slice data..
+
! style="text-align: left; width:27%" |  Approach and Plan
</div>
+
! style="text-align: left; width:27%" |  Progress and Next Steps
<div style="width: 27%; float: left; padding-right: 3%;">
+
|- style="vertical-align:top;"
<h3>Approach, Plan</h3>
+
|
 
+
<!-- Objective bullet points -->
</div>
+
* We will investigate a current model for learning and using Patch-based Gaussian Mixture Model to restore sparse-slice data.
<div style="width: 27%; float: left; padding-right: 3%;">
+
|
<h3>Progress</h3>
+
<!-- Add a bulleted list of key points -->
</div>
+
* We've implemented a small version of the model without regularization and that only works on a small section of a volume. We'll try to deploy it on large-scale dataset on entire volumes.
</div>
+
|
 +
<!-- Fill this out at the end of Project Week; describe what you did this week and what you plan to do next -->
 +
* We've built infrastructure for deploying the training at each location in parallele
 +
* We've used our lab's cluster to deploy training with various parameter at each of 18,000 locations
 +
* We've coded reconstruction modules to rebuild each region, but still need to combine each of the reconstructions.
 +
|}

Latest revision as of 14:09, 8 January 2016

Home < 2016 Winter Project Week < Projects < ImageRestoration

Key Investigators

  • Adrian Dalca (MIT)
  • Katie Bouman (MIT)
  • Polina Golland (MIT)

Project Description

Most synthesis, in-painting or super-resolution methods require a training dataset which includes the desired-quality images. Unfortunately, in the clinical setting this is often now available.

Due to the low quality of clinical images (often with many artifacts, 7mm thick slices, etc), most standard algorithms, such as those for registration, segmentation, analysis, will fail.

To improve results for large datasets of clinical-quality data, we are investigating restoration methods without training datasets. Here, we are using a patch-based Gaussian Mixture Model approach with MRF priors and utilizing only the current dataset, without an external training dataset.

Essentially, we explore a model where for a given location in all volumes of a dataset, we model those image patches as drawn from a particular mixture model.

Objective Approach and Plan Progress and Next Steps
  • We will investigate a current model for learning and using Patch-based Gaussian Mixture Model to restore sparse-slice data.
  • We've implemented a small version of the model without regularization and that only works on a small section of a volume. We'll try to deploy it on large-scale dataset on entire volumes.
  • We've built infrastructure for deploying the training at each location in parallele
  • We've used our lab's cluster to deploy training with various parameter at each of 18,000 locations
  • We've coded reconstruction modules to rebuild each region, but still need to combine each of the reconstructions.