» start page

 » WebMicroscope
     tutorial / help

 » Gleason grading
     background
    
tutorials

 » documentation
   
publications
   
supplemental info

 » about us / contact

 » start / end grading

 » interactive grading
    with instant feedback

 » interobserver varia-
    bility assessment

 » intraobserver
    agreement
 » save / load


Click to open
schematic reference
image
Image processing
Please refer to the published article for general information regarding the image
processing phase. Here you can find details on some topics that are only
briefly described in the article:

Sharpening: As described in the published article, sharpening of the
image tiles before the montage phase was found to improve final image
quality. Sharpening was conducted with a script-called image processing
program run from the command prompt (Irobot, Jasc software inc.).

Stitching: The sharpened image tiles were stitched into a single montage
file, using the ER-Mapper software (Earth Resource Mapping Pty, West
Perth, Australia) controlled by a custom script. The image tiles were stitched with
a ~ 5 um overlapping zone with partial  transparency to compensate for
minor alignment inaccuracies in the microscope stage movements.
Prior to stiching, the image tiles are positioned in a coordinate
system. This was accomplished through consecutive numbering of the image
tiles in the x and y directions by the KS400 microscope control software. The
numbering was used to assign exact coordinate information to the image tiles,
again with a custom script. Stitching of the image tiles is conducted only as an
algorithm, matemathically defining the exact location of each image tile in
the montage file.

Wavelet files: A compressed image is one that has been compressed to
reduce the image file size, while still providing an almost perfect version of
the original. There are various compression techniques. The most effective
technique is wavelet based, which is a method of processing, quantizing and
then encoding the image in a way that results in very high levels of compression,
with an almost perfect reconstruction of the original image. Wavelet compression
involves a way of analyzing an uncompressed image in a recursive fashion,
resulting in a series of higher resolution images, each adding to the information
content in lower resolution images. Because the compressed imagery is
constructed of multi-resolution wavelet levels, fast roaming and zooming on
the imagery is possible by selectively decompressing only the portion of
the image, at the level of detail currently being viewed.

Compression:  The montage files never physically exist as uncompressed
files, which would be almost unmanageable with mainstream workstations.
Instead the algorithmic representations of the montages are directly compressed
into wavelet files. The format used in the atlas and by the Image Web Server
software is ecw (enhanced compressed wavelets).

Automation of the image processing phase: The entire processing phase
was automated by one main custom VBScript (Windows Script Host). The
described software used for the various processing steps all operate with
their own inherent scripting languages. Custom sub-scripts written for these (KS
400 microscope control software, Irobot image processing software, ER-Mapper)
are sequentially initiated from the main script. The main script ends with automatic
uploading of the virtual slides to the image server. Information on the script
details can be obtained from the authors.
                                                                                                                                         Go back