WSL > Produkte > Roxas > PreparationImaging

Sample Preparation and Image Acquisition

Success of automated image analysis depends on quality images taken from well-prepared samples. It is thus a sensible investment of time to improve sample preparation and image capturing. Read below for some rough guidelines.

A. Sample Preparation

Approach 1: Thin sections

Use: Thin sections are the approach of choice when dealing with short tree cores, and cross-sections of tree branches, stems of shrubs and roots of herbaceous plants. They provide optimal contrast and detail.
Sample preparation: See the herb-chronology web pages for advice on sample preparation. Recent advancements in microtome technology also enabled to prepare thin sections of tree cores.

Approach 2: Wood surface

Use: Directly preparing the wood surface may be best in large tree cores of species with medium to large anatomical structures such as in ring-porous species.
Sample preparation: Make sure the surface of the core is plain and clean, and conduit lumina are not filled with dust. Using a core microtome such as the one provided by H. Gaertner (WSL) produces the best results. Alternatively, core surface may be sanded. After cutting or sanding dust should be removed from conduit lumina using air or water blast. Try to improve contrast, e.g. by painting the surface with black ink and filling the lumina with chalk powder. The goal is to get the conduit lumina in another color and/or brightness than the surrounding matrix tissue (cell walls, parenchyma cells). Ideally, the conduit lumina appear in a homogenous color throughout the image. The more overlap in color and/or brightness between conduit lumina and matrix tissue, the more manual editing will be required after automatic ROXAS analysis.

B. Image Acquisition

Approach 1: Microscope (transmitted and incident light)

Use: A transmitted-light microscope is used for thin sections and an incident light microscope for on-surface imaging.
Some general considerations: There is a common trade-off between image resolution and field of view, even if resolution of microscope cameras are improving fast. To combine high resolution with the advantage of having a single image per sample it is recommended to take overlapping sub-images and stitch them together (see below).
Procedure: Place the sample on a XY microscope stage. In tree cores, make sure the core is aligned either fully horizontally or vertically to the direction of the stage movements to avoid offsets (“zigzag”) in the stitched image. Take overlapping images (ca. 25% overlap) of the sample. JPG with a low compression level (>=98%) is sufficient if storage capacity is an issue.
Image stitching: Common tools as the one provided by Adobe Photoshop © offer little control, are relatively slow and result in image distortions that invalidate quantitative measurements. I recommend using the affordable tool PTGui (www.ptgui.com). In its original application it is used to create panorama images of large scenes taken with a camera rotating on a tripod. However, with the appropriate settings and handling PTGui will produce distortion-free images (even if images were taken with a non-orthoscopic optical system) in a fast, controllable and reliable manner. The process is highly automated and does not require provisional alignment of sub-images. Images composed of more than 100 sub-images can be created. A manual by the author of these web pages about how to best use PTGui in quantitative wood anatomy can be obtained here.

Approach 2: Scanner

Use: When the sample surface of a tree core was prepared (see above) scanning may be most efficient.
Procedure: Use a high-resolution scanner. The tree core should be aligned fully vertically. This will allow to optimally trim the image, which is better for storage and ROXAS analysis. Also make sure the tree core has full contact with the scanner along its entire length to obtain focused images.

C. Image Resolution

Determine the pixel/micron resolution to obtain measurements in the proper units. Therefore, an image of a reference object of known size (e.g. a stage micrometer) may be taken using the same magnification as for the samples. Measure the dimension of the reference object in pixels, and divide the obtained number by the dimension of the reference object in metric units.

Important: Determine the pixel/micron resolution after image stitching to make sure any removed image distortion is accounted for.

	DE FR IT EN Print Eidg. Forschungsanstalt für Wald, Schnee und Landschaft WSL
	Sample Preparation and Image Acquisition Success of automated image analysis depends on quality images taken from well-prepared samples. It is thus a sensible investment of time to improve sample preparation and image capturing. Read below for some rough guidelines. . A. Sample Preparation Approach 1: Thin sections Use: Thin sections are the approach of choice when dealing with short tree cores, and cross-sections of tree branches, stems of shrubs and roots of herbaceous plants. They provide optimal contrast and detail. Sample preparation: See the herb-chronology web pages for advice on sample preparation. Recent advancements in microtome technology also enabled to prepare thin sections of tree cores. Approach 2: Wood surface Use: Directly preparing the wood surface may be best in large tree cores of species with medium to large anatomical structures such as in ring-porous species. Sample preparation: Make sure the surface of the core is plain and clean, and conduit lumina are not filled with dust. Using a core microtome such as the one provided by H. Gaertner (WSL) produces the best results. Alternatively, core surface may be sanded. After cutting or sanding dust should be removed from conduit lumina using air or water blast. Try to improve contrast, e.g. by painting the surface with black ink and filling the lumina with chalk powder. The goal is to get the conduit lumina in another color and/or brightness than the surrounding matrix tissue (cell walls, parenchyma cells). Ideally, the conduit lumina appear in a homogenous color throughout the image. The more overlap in color and/or brightness between conduit lumina and matrix tissue, the more manual editing will be required after automatic ROXAS analysis. B. Image Acquisition Approach 1: Microscope (transmitted and incident light) Use: A transmitted-light microscope is used for thin sections and an incident light microscope for on-surface imaging. Some general considerations: There is a common trade-off between image resolution and field of view, even if resolution of microscope cameras are improving fast. To combine high resolution with the advantage of having a single image per sample it is recommended to take overlapping sub-images and stitch them together (see below). Procedure: Place the sample on a XY microscope stage. In tree cores, make sure the core is aligned either fully horizontally or vertically to the direction of the stage movements to avoid offsets (“zigzag”) in the stitched image. Take overlapping images (ca. 25% overlap) of the sample. JPG with a low compression level (>=98%) is sufficient if storage capacity is an issue. Image stitching: Common tools as the one provided by Adobe Photoshop © offer little control, are relatively slow and result in image distortions that invalidate quantitative measurements. I recommend using the affordable tool PTGui (www.ptgui.com). In its original application it is used to create panorama images of large scenes taken with a camera rotating on a tripod. However, with the appropriate settings and handling PTGui will produce distortion-free images (even if images were taken with a non-orthoscopic optical system) in a fast, controllable and reliable manner. The process is highly automated and does not require provisional alignment of sub-images. Images composed of more than 100 sub-images can be created. A manual by the author of these web pages about how to best use PTGui in quantitative wood anatomy can be obtained here. Approach 2: Scanner Use: When the sample surface of a tree core was prepared (see above) scanning may be most efficient. Procedure: Use a high-resolution scanner. The tree core should be aligned fully vertically. This will allow to optimally trim the image, which is better for storage and ROXAS analysis. Also make sure the tree core has full contact with the scanner along its entire length to obtain focused images. C. Image Resolution Determine the pixel/micron resolution to obtain measurements in the proper units. Therefore, an image of a reference object of known size (e.g. a stage micrometer) may be taken using the same magnification as for the samples. Measure the dimension of the reference object in pixels, and divide the obtained number by the dimension of the reference object in metric units. Important: Determine the pixel/micron resolution after image stitching to make sure any removed image distortion is accounted for.