Supplementary MaterialS2

Title:Measuring liver triglyceride content in mice: non-invasive magnetic resonance methods as an alternative to histopathology

Journal name: Magnetic Resonance Materials in Physics, Biology and Medicine

Authors:

JurgenH. Runge, MD —Corresponding author ()

P.J. Bakker, Msc

I.C. Gaemers, Msc, PhD

J. Verheij, MD, PhD

T.B.M. Hakvoort, Msc, PhD

R. Ottenhoff Bsc

J. Stoker, MD, PhD

A.J. Nederveen, PhD

Institution:

AcademicMedicalCenter

Department of Radiology

University of Amsterdam

Meibergdreef 9 - G1-228

1105 AZ Amsterdam

The Netherlands

Oil Red-O-staining:

Oil Red-O (ORO)-staining was performed according to the staining procedure used by our center’s Department of Pathology. The procedure was as follows:

Cryosections were:

  1. cut at 7µm at -18°C using a cryotome and transferred to slides;
  2. driedovernight and fixated for 20’ at room temperature with formol-macrodex;
  3. thoroughly rinsed in distilled water;
  4. dipped twice in 60% 2-propanol;
  5. stained for 25 minutes in OROworking solution (see below);
  6. dipped ten times in 60% 2-propanol to remove excess ORO solution;
  7. thoroughly rinsed with distilled water;
  8. counterstained with hematoxylin for 1’;
  9. rinsed with tap water to blue sections for 5’;
  10. thoroughly rinsed with distilled water for 5’;
  11. and finally covered with Aquatex and a cover slip.

The ORO working solution was composed of:

  1. 60volume-% of ORO stock solution (see below)
  2. 40 volume-% of distilled water

These were added together, mixed well, kept for 10 minutes at room temperature and finally filtered, just prior to the staining procedure.

The ORO stock solution was composed of:

  1. 5 gram of Oil Red O powder
  2. 1000 ml of 2-propanol

These were added together, mixed well and kept overnight in a stove at 56°C. The mixture was allowed to cool down to room temperature before filtering and was kept at room temperature until an ORO working solution was needed.

Digital Image Analysis of ORO-slides

Digital-Image Analysis of ORO sections (DIA-ORO) was performed with photographs acquired with an Olympus BX-41 microscope equipped with an UC30 camera and AnalySIS getIT 5.1 (Olympus Soft Imaging Solutions GmbH, Münster, Germany). Two, six and six randomly chosen fields at x100, x200 and x400 magnification respectively were captured to allow averaging and obtaining average values for each mouse. At each magnification, background photographs (blanks) were obtained by taking a photograph without a sample, to allow adjustments for variation in image intensity. Images were analyzed using Wolfram Mathematica 8 (Wolfram Research, Champaign, IL) as follows:

  1. Images and corresponding (to magnification) blanks were loaded into the program.
  1. For each image, the corresponding blank was subtracted from the image with the ImageSubtract command.
  1. A Gaussian-filtered image was then created by with the GaussianFilter command, with a kernel of pixel radius 50.
  1. Using the images from step 2 and step 3, the next step involved creating color separated images for each RGB-channel:
  2. Image_step3 was subtracted from Image_step2;
  3. To increase differences between different pixels, the RGB values of Image_step4a were multiplied by 5, using the ImageMultiply command.
  4. The ColorSeparate command was applied to Image_step4b to create the three separated color channels.
  1. In the next step, a binary image that identifies fat clustersas 1 was created as follows:
  2. First, the red channel was subtracted from the green channel.
  3. Next, the blue channel was subtracted from Image_step5a:
  4. Then, the Binarize command was applied to Image_step5b with the threshold at 0.5, replacing all pixels 0.5 with 1 and others with 0.
  5. Finally, with the DeleteSmallComponents command and 20 as maximum element size, small components (that represent staining errors and not actual fat clusters) were deleted from Image_step5c.
  1. In the last step, fat pixel area, number of clusters and median and mean cluster size were extracted as follows:
  2. The Dimensions and ImageData commands were applied to the binarized Image_step5d to allow calculation of total pixel area.
  3. The Total, Flatten and ImageData commands were applied to Image_step5d to count the total numbers of pixels with value 1 (i.e. fat vesicle pixels).
  4. Percentage was calculated by dividing the number of selected pixels (6b) by the total number of pixels (6a) and multiplying by 100.
  5. The Max and MorphologicalComponents commands were applied to Image_step5d to identify and count the number of separate fat vesicles.
  6. The Flatten and Count commands were applied on the separated fat vesicles to count the number of pixels per separate fat vesicle of which the mean and median values were subsequently calculated.