How auto W/L is implemented in 3DSlicer?

Scorbinwen · October 16, 2019, 9:12am

Operating system: Windows
Slicer version: 4.8.1
Expected behavior:NO
Actual behavior:NO
%E6%8D%95%E8%8E%B7
How auto W/L is implemented in 3DSlicer?
I’m fanscinated with the auto W/L module in 3DSlicer,which satisfies my requirement for visualizing MR volume.But I want to use this functionality for batch processing,so I want to know the exact algorithm.
I want know whether this functionality can be implemented using ITK apis.
Programmatic advice is welcome.
Thank you

lassoan · October 16, 2019, 2:24pm

We define window/level based on the image histogram (using vtkImageHistogramStatistics):

github.com

Slicer/Slicer/blob/9cd58e8ebd623c554a73881c5a7c1e8cf05f5361/Libs/MRML/Core/vtkMRMLScalarVolumeDisplayNode.cxx#L733-L770


if (this->HistogramStatistics == nullptr)
  {
  this->HistogramStatistics = vtkImageHistogramStatistics::New();


  // Set automatic window/level to include the entire intensity range
  // (except top/bottom 0.1%, to not let a very thin tail of the intensity
  // distribution to decrease the image contrast too much).
  // While in CT and sometimes in MRI, there may be a large empty area
  // outside the reconstructed image, which could be suppressed
  // by a larger lower percentile value, it would make the method
  // too specific to particular imaging modalities and could lead to
  // suboptimal results for other types of images.
  // Therefore, we choose small, symmetric percentile values here
  // and maybe add modality-specific methods later (e.g., for CT
  // images we could set lower value to -1000HU).
  this->HistogramStatistics->SetAutoRangePercentiles(0.1, 99.9);


  // Percentiles are very low (0.1%), so there is no need for
  // range expansion.
  this->HistogramStatistics->SetAutoRangeExpansionFactors(0.0, 0.0);

This file has been truncated. show original

cpinter · October 16, 2019, 3:50pm

I’m really curious what do you mean by this Because W/L is a strictly visualization feature that has no persistent effect on the data, but batch processing typically changes the data. Can you please shed some light on it for us?

Scorbinwen · October 17, 2019, 3:17am

Oh,Sorry about my unclear statement,I want to convert the MRI volumes to PNG images,while maximally maintaining the MRI image’s detatils.Luckily, auto W/L module in 3DSlicer just satisfies my very requirement.But I know this W/L visualization feature has no persistent effect on the data,so I try to write a program to process these MRI data based on the auto W/L algorithm.
Thank you for your kind reply, @Iassoan has solved my problem.
Thank you both very much~~

Scorbinwen · October 17, 2019, 3:50am

Thank you for your kind reply,
I will try this algorithm.
Thank you~~

Chris_Rorden · October 18, 2019, 3:15pm

I would be very cautious about scaling medical imaging data to PNG. While PNG can support 16-bit grayscale, most implementations only support 8-bit (e.g. 256 shades of gray). While that may be sufficient with many machine learning algorithms (e.g. facial recognition of photographs) I would be very wary of using 8-bits to represent medical imaging data. Since you mention MR scans, in fMRI scans a 1% signal change is huge. Most of the brightness comes from constant properties (e.g. amount of water in the voxel), and we are seeking a very small but reliable change in signal. You really want to be careful to preserve the precision of your data. As a minimum, you might want to consider first doing your desired low/high pass filter, then subtracting each image from the mean, and looking at scaling these residual values from 0…255.

Scorbinwen · November 25, 2019, 6:29am

Thank you for your advice,I’m reconsidering my problems.

shai-ikko · December 13, 2023, 1:02pm

Since I was just looking for this, and the code link is no longer valid, here’s an updated (and more durable, hopefully) link:

github.com

Slicer/Slicer/blob/v5.6.1/Libs/MRML/Core/vtkMRMLScalarVolumeDisplayNode.cxx#L749-L786


      
          if (this->HistogramStatistics == nullptr)
            {
            this->HistogramStatistics = vtkImageHistogramStatistics::New();
          
            // Set automatic window/level to include the entire intensity range
            // (except top/bottom 0.1%, to not let a very thin tail of the intensity
            // distribution to decrease the image contrast too much).
            // While in CT and sometimes in MRI, there may be a large empty area
            // outside the reconstructed image, which could be suppressed
            // by a larger lower percentile value, it would make the method
            // too specific to particular imaging modalities and could lead to
            // suboptimal results for other types of images.
            // Therefore, we choose small, symmetric percentile values here
            // and maybe add modality-specific methods later (e.g., for CT
            // images we could set lower value to -1000HU).
            this->HistogramStatistics->SetAutoRangePercentiles(0.1, 99.9);
          
            // Percentiles are very low (0.1%), so there is no need for
            // range expansion.
            this->HistogramStatistics->SetAutoRangeExpansionFactors(0.0, 0.0);

This file has been truncated. show original

Topic		Replies	Views
Window width/level range calculation Support python	3	1067	August 2, 2021
Recent improvements in window/level management Support feature	10	7147	August 31, 2020
General questions about Microscopy image processing and & visualization using 3D Slicer Support	6	395	March 3, 2020
Custom window/level presets Feature requests	1	1267	December 12, 2018
Consultation on the feasibility of developing an extension for 3DSlicer Support segmentation , python , virtual-reality	8	378	June 24, 2021

How auto W/L is implemented in 3DSlicer?

Related Topics