Menu
en-USen-GB
en-USEnglish (United States)en-GBEnglish (United Kingdom)
 

Automated Image Stitching Module

You are here / Software  /  MCID Core  /  Automated Image Stitching Module

 

Introduction

Biologists are used to viewing cellular detail at high magnification. To gain an appreciation of larger-scale structure (metastructure), the slide is moved about while images of different aspects of structure are compared.

Meta structure is more easily appreciated and communicated when a single image shows large areas of the specimen. Metastructure images can be made by reducing magnification so that the field of view is larger. While this is fine if cellular detail is not required, there are limitations:

  • Resolution is reduced with low NA objectives.
  • Low power objectives tend to introduce geometric distortion.
  • Each camera pixel represents a larger portion of the specimen, reducing image resolution.
  • Even a 1X objective may not accommodate the metastructure of interest. This is particularly true with digital cameras, which see only a small part (typically 30%) of the eyepiece field of view.

A traditional alternative to low magnification is the dreaded photomontage, in which multiple photomicrographs are cut and pasted together. Anyone who has spent hours making photomontages tends to look for other methods. This has lead to widespread adoption of digital montaging systems (Fig. 1).

What to Look For in a Montaging System

More advanced montaging systems automate the process, and yield much better images. Key things to look for are listed below.

  • Motor stage control algorithms correct positioning errors from field to field, so that these errors do not accumulate (Fig. 2).
  • Alignment algorithms perform precise edge mapping of the tiles, and construct the montage automatically. The best systems do this during the montaging process, so that the user can see the large image build in real time.
  • An autofocus system acquires large areas without user attention. Some montaging systems also include image combination (yielding great depth of field) and image deconvolution (digital confocal) functions (Fig. 3).
  • Intensity and color correction algorithms remove spatial variations in microscope illumination and camera detection. Left uncorrected, these variations tend to result in a "patchwork" montage, with each tile visible (Fig. 4).
  • Multimodal capabilities. Alignment, intensity and color correction algorithms must be able to accommodate fluorescence, and various brightfield modes of operation.
Figure 2:
Nine microscope fields of Papaver nudicaule, montaged with a 4X objective and a color video camera. In panel A, no attempt has been made to align the edges of each individual tile. The resulting image misalignment can clearly be seen. Panel B shows the same set of fields, acquired using MCID Elite automated alignment feature.
Tiling image 2
Figure 3:
Twenty five microscope fields (A,C) acquired and montaged with an AIS Montaging system (20X objective, counterstained Golgi preparation, section thickness approximately 60 um). High-resolution views of the inset areas are shown in panels B and D. The montage at top contains a single focal plane. Note that only some of the material is in focus. The montage at bottom uses image combination to increase depth of field. All of the cell material is in focus.
Tiling image 3
Figure 4:
Sixteen fields (rat cerebellum, 10X objective) acquired with an MCID Elite system, without (A) or with (B) corrections for uneven illumination and camera response. Notice the patchwork effect in A, while panel B is seamless.
Tiling image 4

The best montaging systems create seamless, focused, and large-scale images, completely automatically. There should not be any limitation on the number of fields, or on the type of microscopy being performed. Our own montaging systems, for example, are used for everything from demonstrating neural organization to making multicolor fluorescence images of entire tumors.

Montages can be visually spectacular. The eye is seduced by what looks like an impossibly sharp photomicrograph, showing large-scale organization far exceeding what we are used to seeing from a microscope. While digital montages are created for their scientific value, they are also a joy to behold and many can be appreciated as art.



Overcoming the Limitations of Digital Microscopy

Compare a digital microscopy image with a photomicrograph. The photomicrograph contains more detail than the digital image. Because most electronic cameras use a relatively small number of pixels (up to 4 million is typical), they cannot resolve all of the detail supplied by the optics. In contrast, 35 mm film has far higher resolution (equivalent to about twelve million pixels within a 35 mm frame). Therefore, digital images tend to compromise resolution, while film photomicrographs are very close to what we see through the eyepieces.

If a digital imaging system is to match photomicrography, it must represent the specimen with an adequate number of pixels. There are a number of ways to increase the pixel count.

  • High-resolution cameras. High-resolution cameras (e.g. 4096 x 4096 pixels) are becoming more cost effective. As such cameras mature, we can expect digital photomicrographs to more closely approximate film. However, the camera is limited to the portion of the microscope field projected onto the detector. It cannot go beyond a single microscope field. There is also a limited variety of cameras (color and low light capabilities can be difficult), they function quite slowly (transferring all of those pixels to from the camera to the computer takes time), and they tend to be quite costly.
  • Higher magnification. If we magnify the specimen, each part of it occupies more pixels so resolution is better (Fig. 5). Of course, this strategy also has a negative aspect in that it curtails the field of view.
  • Higher magnification + montage. We can use higher magnification to improve resolution, and retain field of view by montaging. This approach is cost-effective and allows the use of a very broad variety of cameras. We can even combine montaging with high-resolution cameras to achieve the best possible results.
Figure 5:
Rat hippocampus acquired as a single image with a 4X objective (top) or montaged (24 fields) with a 10X objective (bottom) using an AIS Montaging system. Full resolution views of the inset areas are shown at right. The 10X montage exhibits much better detail (C).
Tiling image 5


The MCID Core Tiled Field Mapping System (TFM)

We offer an advanced montaging system, which we call the Tiled Field Mapping system. TFM is available as part of the MCID Elite or AIS image analyzers. It offers a unique feature set.

Integration with a powerful image analyzer: TFM is an integral part of our state-of-the-art; multi-functional MCID™ Elite image analysis system, and is also available as an application-specific AIS™ Montaging system. Both systems provide a natural and convenient interface for working with large images, including image enhancement, archiving, printing, manipulation and quantification.

Autoalignment: Spatial errors accumulate as a motor stage moves over larger distances. The edges of two adjacent fields might be within a pixel, but the error will be multiple pixels three fields away. TFM includes algorithms that perform automated alignment of the discrete fields, and feedback regulation of stage precision as the montaging proceeds.

Intensity/color correction: Cameras, microscope optics and illumination systems all contribute to intensity variations (e.g. 20%) across a field of view. There are also color variations that might not be obvious in a single field, but any intensity and/or color variation that repeats from tile to tile will form a pattern that is very obvious in a montage. Therefore, TFM corrects both intensity and color errors, automatically.

Autofocus: Residual tilt in the microscope stage is not noticed within a single field, but is very obvious across multiple fields of view. Point-to-point variations in specimen depth a present both within and across fields. TFM includes automatic corrections and accommodations which achieve sharply focused images.

  • A tilt correction function adjusts for stage misalignment.
  • Software autofocus makes a number of exposures and calculates the best plane of focus for each tile.
  • A hardware autofocus system (laser detector, stage with fast Z positioning) is an available option. It operates in conjunction with the software autofocus and speeds the focus process.
  • Image combination convolves any number of planes of focus into a single sharp image. The image combination feature is particularly useful for thick specimens.

Any size image: Unlike many other montaging systems, TFM does not limit the number of tiles that can be acquired. Start it up and leave it running. As a general rule, expect about 10 seconds/tile. We have routinely made montages of about 8000 x 8000 pixels, containing more than 200 microscope fields.

Fluorescence montages: TFM uses any of the monochrome or color cameras that the MCID/AIS image analyzers support. These include cameras suitable for low light imaging of fluorescent or luminescent samples (Fig. 6).

Any microscope: TFM can be used with any microscope, upright or inverted using one of many leading motorised stage kits from Prior, Ludl & Marzhauser. Visit our stage section for more detailed information on the options available.

Figure 6:
Fluorescence montage consisting of 28microscope fields (10X objective, monochrome camera, three color excitation presented with a computer-controlled filter wheel) acquired with an MCID Elite system. The image is a xenograft of a human cervical cancer cell line grown in mouse. The red areas are regions of hypoxia, the green are blood vessels and the blue is Hoechst- labeled DNA.
Tiling image 6

Theme colors

 

Skin Layout

  • Opened
  • Border
  • Boxed

Texture