'\" t
.\"     Title: OPENSLIDE-FORMATS
.\"    Author: [see the "AUTHORS" section]
.\" Generator: DocBook XSL Stylesheets v1.79.1 <http://docbook.sf.net/>
.\"      Date: 07/29/2016
.\"    Manual: File Formats
.\"    Source: OpenSlide 3.4.1+dfsg
.\"  Language: English
.\"
.TH "OPENSLIDE\-FORMATS" "3" "07/29/2016" "OpenSlide 3\&.4\&.1+dfsg" "File Formats"
.\" -----------------------------------------------------------------
.\" * Define some portability stuff
.\" -----------------------------------------------------------------
.\" ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.\" http://bugs.debian.org/507673
.\" http://lists.gnu.org/archive/html/groff/2009-02/msg00013.html
.\" ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.ie \n(.g .ds Aq \(aq
.el       .ds Aq '
.\" -----------------------------------------------------------------
.\" * set default formatting
.\" -----------------------------------------------------------------
.\" disable hyphenation
.nh
.\" disable justification (adjust text to left margin only)
.ad l
.\" -----------------------------------------------------------------
.\" * MAIN CONTENT STARTS HERE *
.\" -----------------------------------------------------------------
.SH "NAME"
openslide-formats \- Reference supported formats
.SH "DESCRIPTION"
.SH "LIST OF KNOWN PROPERTIES"
.SS "Properties generated by OpenSlide"
.PP
openslide\&.background\-color
.RS 4
The background color of the slide, given as an RGB hex triplet\&. This property is not always present\&.
.RE
.PP
openslide\&.bounds\-height
.RS 4
The height of the rectangle bounding the non\-empty region of the slide\&. This property is not always present\&.
.RE
.PP
openslide\&.bounds\-width
.RS 4
The width of the rectangle bounding the non\-empty region of the slide\&. This property is not always present\&.
.RE
.PP
openslide\&.bounds\-x
.RS 4
The X coordinate of the rectangle bounding the non\-empty region of the slide\&. This property is not always present\&.
.RE
.PP
openslide\&.bounds\-y
.RS 4
The Y coordinate of the rectangle bounding the non\-empty region of the slide\&. This property is not always present\&.
.RE
.PP
openslide\&.comment
.RS 4
A free\-form text comment\&.
.RE
.PP
openslide\&.mpp\-x
.RS 4
Microns per pixel in the X dimension of level 0\&. May not be present or accurate\&.
.RE
.PP
openslide\&.mpp\-y
.RS 4
Microns per pixel in the Y dimension of level 0\&. May not be present or accurate\&.
.RE
.PP
openslide\&.objective\-power
.RS 4
Magnification power of the objective\&. Often inaccurate; sometimes missing\&.
.RE
.PP
openslide\&.quickhash\-1
.RS 4
A non\-cryptographic hash of a subset of the slide data\&. It can be used to uniquely identify a particular virtual slide, but cannot be used to detect file corruption or modification\&.
.RE
.PP
openslide\&.vendor
.RS 4
The name of the vendor backend\&.
.RE
.SS "Properties for TIFF\-based formats"
.PP
tiff\&.Artist
.RS 4
The contents of the TIFF Artist tag\&.
.RE
.PP
tiff\&.Copyright
.RS 4
The contents of the TIFF Copyright tag\&.
.RE
.PP
tiff\&.DateTime
.RS 4
The contents of the TIFF DateTime tag\&.
.RE
.PP
tiff\&.DocumentName
.RS 4
The contents of the TIFF DocumentName tag\&.
.RE
.PP
tiff\&.HostComputer
.RS 4
The contents of the TIFF HostComputer tag\&.
.RE
.PP
tiff\&.ImageDescription
.RS 4
The contents of the TIFF ImageDescripton tag\&.
.RE
.PP
tiff\&.Make
.RS 4
The contents of the TIFF Make tag\&.
.RE
.PP
tiff\&.Model
.RS 4
The contents of the TIFF Model tag\&.
.RE
.PP
tiff\&.ResolutionUnit
.RS 4
The contents of the TIFF ResolutionUnit tag\&.
.RE
.PP
tiff\&.Software
.RS 4
The contents of the TIFF Software tag\&.
.RE
.PP
tiff\&.XPosition
.RS 4
The contents of the TIFF XPosition tag\&.
.RE
.PP
tiff\&.XResolution
.RS 4
The contents of the TIFF XResolution tag\&.
.RE
.PP
tiff\&.YPosition
.RS 4
The contents of the TIFF YPosition tag\&.
.RE
.PP
tiff\&.YResolution
.RS 4
The contents of the TIFF YResolution tag\&.
.RE
.SS "Vendor\-specific properties"
.PP
A list of vendor\-specific properties can be found on the
\m[blue]\fBpages for each vendor format\fR\m[]\&\s-2\u[1]\d\s+2\&.
.SH "APERIO FORMAT"
.PP
Format
.RS 4
single\-file pyramidal tiled TIFF, with non\-standard metadata and compression
.RE
.PP
File extensions
.RS 4
\&.svs,
\&.tif
.RE
.PP
OpenSlide vendor backend
.RS 4
aperio
.RE
.SS "Vendor Documentation"
.PP
\m[blue]\fBhttp://www\&.aperio\&.com/documents/api/Aperio_Digital_Slides_and_Third\-party_data_interchange\&.pdf\fR\m[]
.SS "Detection"
.PP
Aperio slides are stored in single\-file TIFF format\&. OpenSlide will detect a file as Aperio if:
.sp
.RS 4
.ie n \{\
\h'-04' 1.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  1." 4.2
.\}
The file is TIFF\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 2.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  2." 4.2
.\}
The initial image is tiled\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 3.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  3." 4.2
.\}
The
ImageDescription
tag starts with
Aperio\&.
.RE
.SS "Relevant TIFF tags"
.TS
allbox tab(:);
lB lB.
T{
Tag
T}:T{
Description
T}
.T&
l l
l l.
T{
ImageDescription
T}:T{
Stores some important key\-value pairs and other information, see below
T}
T{
Compression
T}:T{
May be 33003 or 33005, which represent specific kinds of JPEG 2000 compression, see below
T}
.TE
.sp 1
.SS "Extra data stored in ImageDescription"
.PP
For tiled images, the
ImageDescription
tag contains some dimensional downsample information as well as what look like offsets\&. Additionally, vertical line\-delimited key\-value pairs are stored, in at least the full\-resolution image\&. A key\-value pair is equals\-delimited\&. These key\-values are stored as properties starting with \(lqaperio\&.\(rq\&. Currently, OpenSlide does not use any of the information present in these key\-value fields\&.
.PP
For stripped images, the
ImageDescription
tag may contain a name, followed by a carriage return\&. This is used for naming the associated images\&. The second image in the file does not have a name, though it is an associated image\&.
.SS "TIFF Image Directory Organization"
.PP
\m[blue]\fBhttp://www\&.aperio\&.com/documents/api/Aperio_Digital_Slides_and_Third\-party_data_interchange\&.pdf\fR\m[]
page 14:
.PP
The first image in an SVS file is always the baseline image (full resolution)\&. This image is always tiled, usually with a tile size of 240x240 pixels\&. The second image is always a thumbnail, typically with dimensions of about 1024x768 pixels\&. Unlike the other slide images, the thumbnail image is always stripped\&. Following the thumbnail there may be one or more intermediate \(lqpyramid\(rq images\&. These are always compressed with the same type of compression as the baseline image, and have a tiled organization with the same tile size\&.
.PP
Optionally at the end of an SVS file there may be a slide label image, which is a low resolution picture taken of the slide\(cqs label, and/or a macro camera image, which is a low resolution picture taken of the entire slide\&. The label and macro images are always stripped\&.
.SS "JPEG 2000 (compression types 33003 or 33005)"
.PP
Some Aperio files use compression type 33003 or 33005\&. Images using this compression need to be decoded as a JPEG 2000 codestream\&. For 33003: YCbCr format, possibly with a chroma subsampling of 4:2:2\&. For 33005: RGB format\&. Note that the TIFF file may not encode the colorspace or subsampling parameters in the
PhotometricInterpretation
field, nor the
YCbCrSubsampling
field, even though the TIFF standard seems to require this\&. The correct subsampling can be found in the JPEG 2000 codestream\&.
.SS "Associated Images"
.PP
thumbnail
.RS 4
the second image in the file
.RE
.PP
label
.RS 4
optional, the name \(lqlabel\(rq is given in
ImageDescription
.RE
.PP
macro
.RS 4
optional, the name \(lqmacro\(rq is given in
ImageDescription
.RE
.SS "Known Properties"
.PP
All key\-value data encoded in the
ImageDescription
TIFF field is represented as properties prefixed with \(lqaperio\&.\(rq\&.
.PP
openslide\&.mpp\-x
.RS 4
normalized
aperio\&.MPP
.RE
.PP
openslide\&.mpp\-y
.RS 4
normalized
aperio\&.MPP
.RE
.PP
openslide\&.objective\-power
.RS 4
normalized
aperio\&.AppMag
.RE
.SS "Test Data"
.PP
\m[blue]\fBhttp://openslide\&.cs\&.cmu\&.edu/download/openslide\-testdata/Aperio/\fR\m[]
.SH "HAMAMATSU FORMAT"
.PP
Format
.RS 4
multi\-file JPEG/NGR with proprietary metadata and index file formats, and single\-file TIFF\-like format with proprietary metadata
.RE
.PP
File extensions
.RS 4
\&.vms,
\&.vmu,
\&.ndpi
.RE
.PP
OpenSlide vendor backend
.RS 4
hamamatsu
.RE
.SS "Detection"
.PP
OpenSlide will detect a file as Hamamatsu if:
.sp
.RS 4
.ie n \{\
\h'-04' 1.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  1." 4.2
.\}
The file given is a INI\-style text file\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 2.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  2." 4.2
.\}
It has a
[Virtual Microscope Specimen]
(VMS) or
[Uncompressed Virtual Microscope Specimen]
(VMU) group\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 3.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  3." 4.2
.\}
If VMS, there are at least 1 row and 1 column of JPEG images (NoJpegColumns
and
NoJpegRows)\&.
.RE
.PP
or if:
.sp
.RS 4
.ie n \{\
\h'-04' 1.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  1." 4.2
.\}
The file has a TIFF directory structure\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 2.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  2." 4.2
.\}
TIFF tag 65420 is present\&.
.RE
.SS "Overview"
.PP
The Hamamatsu format has three variants\&. VMS and VMU consist of an index file, 2 or more image files, and (in the case of VMS) an \(lqoptimisation\(rq file\&. NDPI consists of a single TIFF\-like file with some custom TIFF tags\&. VMS and NDPI contain JPEG images; VMU contains NGR images (a custom uncompressed format)\&.
.PP
Multiple focal planes are ignored, only focal plane 0 is read\&.
.PP
JPEG does not allow for files larger than 65535 pixels on a side\&. In VMS, multiple JPEG files are used to encode large images\&. To avoid having many files, VMS uses close to maximum size (65K by 65K) JPEG files\&. NDPI, instead, stuffs large levels into a single JPEG and sets the overflowed width/height fields to 0\&.
.PP
Unfortunately, JPEG provides very poor support for random\-access decoding of parts of a file\&. To get around this, JPEG restart markers are placed at regular intervals, and these offsets are specified in the optimisation file (in VMS) or in a TIFF tag (in NDPI)\&. With restart markers identified, OpenSlide can treat JPEG as a tiled format, where the height is the height of an MCU row, and the width is the number of MCUs per row divided by the restart marker interval times the width of an MCU\&. (This often leads to oddly\-shaped and inefficient tiles of 4096x8, for example\&.)
.PP
Unfortunately, the VMS optimisation file does not give the location of every restart marker, only the ones found at the beginning of an MCU row\&. It also seems that the file ends early, and does not give the location of the restart marker at the last MCU row of the last image file\&.
.PP
Thus, the optimisation file can only be taken as a hint, and cannot be trusted\&. The entire set of JPEG files must be scanned for restart markers in order to facilitate random access\&. OpenSlide does this lazily as needed, and also in a background thread that runs only when OpenSlide is otherwise idle\&.
.PP
The VMS map file is a lower\-resolution version of the other images, and can be used to make a 2\-level JPEG pyramid\&. JPEG also allows for lower\-resolution decoding, so further pyramid levels are synthesized from each JPEG file\&.
.SS "VMS File"
.PP
The
\&.vms
file is the main index file for the VMS format\&. It is a Windows INI\-style key\-value pair file, with sections\&. Only keys in the
Virtual Microscope Specimen
group are read by OpenSlide\&.
.PP
Here are known keys from the file:
.TS
allbox tab(:);
lB lB.
T{
Key
T}:T{
Description
T}
.T&
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l.
T{
NoLayers
T}:T{
Number of layers, currently must be 1 to be accepted
T}
T{
NoJpegColumns
T}:T{
Number of JPEG files across, given in ImageFile attributes
T}
T{
NoJpegRows
T}:T{
Number of JPEG files down, given in ImageFile attributes
T}
T{
ImageFile
T}:T{
Semantically equivalent to ImageFile(0,0,0), though not specified that way\&. The image in position (0,0,0) of the set of images
T}
T{
ImageFile(x,y)
T}:T{
Semantically equivalent to ImageFile(0,x,y), though not specified that way\&. The image in position (0,x,y) of the set of images
T}
T{
ImageFile(z,x,y)
T}:T{
Where x and y are non\-negative integers\&. Both x and y cannot be 0\&. z is a positive integer\&. These are the images that make up the virtual slide, as a concatenation of JPEG images\&. x and y specify the location of each JPEG, z specifies the focal plane
T}
T{
MapFile
T}:T{
A lower\-resolution version of all the ImageFiles
T}
T{
OptimisationFile
T}:T{
File specifying some of the restart marker offsets in each ImageFile
T}
T{
AuthCode
T}:T{
Unknown
T}
T{
SourceLens
T}:T{
Apparently the objective power
T}
T{
PhysicalWidth
T}:T{
Width of the main image in nm
T}
T{
PhysicalHeight
T}:T{
Height of the main image in nm
T}
T{
LayerSpacing
T}:T{
Unknown
T}
T{
MacroImage
T}:T{
Image file for the \(lqmacro\(rq associated image
T}
T{
PhysicalMacroWidth
T}:T{
Width of the macro image in nm
T}
T{
PhysicalMacroHeight
T}:T{
Height of the macro image in nm, sometimes with a trailing semicolon
T}
T{
XOffsetFromSlideCentre
T}:T{
Distance in X from the center of the entire slide (i\&.e\&., the macro image) to the center of the main image, in nm
T}
T{
YOffsetFromSlideCentre
T}:T{
Distance in Y from the center of the entire slide to the center of the main image, in nm
T}
.TE
.sp 1
.SS "VMU File"
.PP
The
\&.vmu
file is the main index file for the VMU format\&. Only keys in the
Uncompressed Virtual Microscope Specimen
group are read by OpenSlide\&.
.PP
Here are known keys from the file:
.TS
allbox tab(:);
lB lB.
T{
Key
T}:T{
Description
T}
.T&
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l.
T{
NoLayers
T}:T{
(see VMS above)
T}
T{
ImageFile
T}:T{
(see VMS above)
T}
T{
ImageFile(x,y)
T}:T{
(see VMS above)
T}
T{
ImageFile(z,x,y)
T}:T{
(see VMS above)
T}
T{
MapFile
T}:T{
(see VMS above)
T}
T{
MapScale
T}:T{
Seems to be the downsample factor of the map
T}
T{
AuthCode
T}:T{
(see VMS above)
T}
T{
SourceLens
T}:T{
(see VMS above)
T}
T{
PixelWidth
T}:T{
Width of the image in pixels
T}
T{
PixelHeight
T}:T{
Height of the image in pixels
T}
T{
PhysicalWidth
T}:T{
(see VMS above)
T}
T{
PhysicalHeight
T}:T{
(see VMS above)
T}
T{
LayerSpacing
T}:T{
(see VMS above)
T}
T{
LayerOffset
T}:T{
Unknown
T}
T{
MacroImage
T}:T{
(see VMS above)
T}
T{
PhysicalMacroWidth
T}:T{
(see VMS above)
T}
T{
PhysicalMacroHeight
T}:T{
(see VMS above)
T}
T{
XOffsetFromSlideCentre
T}:T{
(see VMS above)
T}
T{
YOffsetFromSlideCentre
T}:T{
(see VMS above)
T}
T{
Reference
T}:T{
Unknown
T}
T{
BitsPerPixel
T}:T{
Bits per pixel, currently expected to be 36
T}
T{
PixelOrder
T}:T{
Currently expected to be RGB
T}
T{
Creator
T}:T{
String describing the software creating this image
T}
T{
IlluminationMode
T}:T{
Unknown
T}
T{
ExposureMultiplier
T}:T{
Unknown, possibly the multiplier used to scale to 15 bits?
T}
T{
GainRed
T}:T{
Unknown
T}
T{
GainGreen
T}:T{
Unknown
T}
T{
GainBlue
T}:T{
Unknown
T}
T{
FocalPlaneTolerance
T}:T{
Unknown
T}
T{
NMP
T}:T{
Unknown
T}
T{
MacroIllumination
T}:T{
Unknown
T}
T{
FocusOffset
T}:T{
Unknown
T}
T{
RefocusInterval
T}:T{
Unknown
T}
T{
CubeName
T}:T{
Unknown
T}
T{
HardwareModel
T}:T{
Name of the hardware
T}
T{
HardwareSerial
T}:T{
Serial number of the hardware
T}
T{
NoFocusPoints
T}:T{
Unknown
T}
T{
FocusPoint0X
T}:T{
Unknown
T}
T{
FocusPoint0Y
T}:T{
Unknown
T}
T{
FocusPoint0Z
T}:T{
Unknown
T}
T{
FocusPoint1X
T}:T{
Unknown
T}
T{
FocusPoint1Y
T}:T{
Unknown
T}
T{
FocusPoint1Z
T}:T{
Unknown
T}
T{
FocusPoint2X
T}:T{
Unknown
T}
T{
FocusPoint2Y
T}:T{
Unknown
T}
T{
FocusPoint2Z
T}:T{
Unknown
T}
T{
FocusPoint3X
T}:T{
Unknown
T}
T{
FocusPoint3Y
T}:T{
Unknown
T}
T{
FocusPoint3Z
T}:T{
Unknown
T}
T{
NoBlobPoints
T}:T{
Unknown
T}
T{
BlobPoint0Blob
T}:T{
Unknown
T}
T{
BlobPoint0FocusPoint
T}:T{
Unknown
T}
T{
BlobPoint1Blob
T}:T{
Unknown
T}
T{
BlobPoint1FocusPoint
T}:T{
Unknown
T}
T{
BlobPoint2Blob
T}:T{
Unknown
T}
T{
BlobPoint2FocusPoint
T}:T{
Unknown
T}
T{
BlobPoint3Blob
T}:T{
Unknown
T}
T{
BlobPoint3FocusPoint
T}:T{
Unknown
T}
T{
BlobMapWidth
T}:T{
Unknown
T}
T{
BlobMapHeight
T}:T{
Unknown
T}
.TE
.sp 1
.SS "NDPI File"
.PP
NDPI uses a TIFF\-like structure, but libtiff cannot read the headers of an NDPI file\&. This is because NDPI specifies the
RowsPerStrip
as the height of the file, and after doing out the multiplication, this typically overflows libtiff and it refuses to open the file\&. Also, the TIFF tags are not stored in sorted order\&.
.PP
NDPI stores an image pyramid in TIFF directory entries\&. In some files, the lower\-resolution pyramid levels contain no restart markers\&. The macro image, and sometimes an active\-region map, seems to come last\&.
.PP
JPEG files in NDPI are not necessarily valid\&. If
ImageWidth
or
ImageHeight
exceeds the JPEG limit of 65535, then the width or height as stored in the JPEG file is 0\&. libjpeg will refuse to read the header of such a file, so the JPEG data stream must be altered when fed into libjpeg\&.
.PP
NDPI is based on the classic TIFF format, which does not support files larger than 4 GB\&. However, NDPI files can be larger than 4 GB\&. NDPI generally handles this by overflowing the corresponding TIFF fields, requiring the reader to guess the high\-order bits\&. This affects TIFF Value Offsets with pointers to out\-of\-line values, as well as the value of the
StripOffsets
field\&. Some TIFF fields (e\&.g\&.
Software) have the same Value Offset in every directory; for these, no concatenation of high\-order bits is necessary\&. For the others (primarily field 65426) it seems reasonable to select high\-order bits which place the value at the largest offset below the directory itself, since the TIFF directory is positioned after the data it points to\&. NDPI always stores next\-directory offsets (in the TIFF header and at the end of each directory) as 64\-bit quantities, even though TIFF specifies them as 32 bits; this is possible because the TIFF standard places them at the end of their parent data structures\&.
.PP
It is not clear whether NDPI can support individual directories larger than 4 GB\&. Such files would require additional inferences for the
StripByteCounts
field, for Value Offsets that are identical across directories, and for the optimisation entries\&.
.PP
Here are the observed TIFF tags:
.TS
allbox tab(:);
lB lB.
T{
Tag
T}:T{
Description
T}
.T&
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l
l l.
T{
ImageWidth
T}:T{
Width of the image
T}
T{
ImageHeight
T}:T{
Height of the image
T}
T{
Make
T}:T{
\(lqHamamatsu\(rq
T}
T{
Model
T}:T{
\(lqNanoZoomer\(rq or \(lqC9600\-12\(rq, etc
T}
T{
XResolution
T}:T{
Seemingly correct X resolution, when interpreted with ResolutionUnit
T}
T{
YResolution
T}:T{
Seemingly correct Y resolution, when interpreted with ResolutionUnit
T}
T{
ResolutionUnit
T}:T{
Seemingly correct resolution unit
T}
T{
Software
T}:T{
\(lqNDP\&.scan\(rq, sometimes with a version number
T}
T{
StripOffsets
T}:T{
The offset of the JPEG file for this layer
T}
T{
StripByteCounts
T}:T{
The length of the JPEG file for this layer
T}
T{
65420
T}:T{
Always exists, always 1\&.  File format version?
T}
T{
65421
T}:T{
SourceLens, correctly downsampled for each entry\&. \-1 for macro image, \-2 for a map of non\-empty regions\&.
T}
T{
65422
T}:T{
XOffsetFromSlideCentre
T}
T{
65423
T}:T{
YOffsetFromSlideCentre
T}
T{
65424
T}:T{
Seemingly the Z offset from the center focal plane (in nm?)
T}
T{
65425
T}:T{
Unknown, always 0?
T}
T{
65426
T}:T{
Optimisation entries, as above
T}
T{
65427
T}:T{
Reference
T}
T{
65428
T}:T{
Unknown, AuthCode?
T}
T{
65430
T}:T{
Unknown, have seen 0\&.0
T}
T{
65433
T}:T{
Unknown, I have seen 1500 in this tag
T}
T{
65439
T}:T{
Unknown, perhaps some polygon ROI?
T}
T{
65440
T}:T{
Unknown, I have seen this: <0 0 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 1 9 1 10 1 11 1 12 1 13 1 14 1 15 1 16 1 17>
T}
T{
65441
T}:T{
Unknown, always 0?
T}
T{
65442
T}:T{
Scanner serial number
T}
T{
65443
T}:T{
Unknown, have seen 0 or 16
T}
T{
65444
T}:T{
Unknown, always 80?
T}
T{
65445
T}:T{
Unknown, have seen 0, 2, 10
T}
T{
65446
T}:T{
Unknown, always 0?
T}
T{
65449
T}:T{
ASCII metadata block, key=value pairs, not always present
T}
T{
65455
T}:T{
Unknown, have seen 13
T}
T{
65456
T}:T{
Unknown, have seen 101
T}
T{
65457
T}:T{
Unknown, always 0?
T}
T{
65458
T}:T{
Unknown, always 0?
T}
.TE
.sp 1
.SS "Optimisation File (only for VMS)"
.PP
The optimisation file contains a list of 32\- (or 64\- or 320\- ?) bit little endian values, giving the file offset into an MCU row, each offset starts at a 40\-byte alignment, and the last row (of the entire file, not each image) seems to be missing\&. The offsets are all packed into 1 file, even with multiple images\&. The order of images is left\-to\-right, top\-to\-bottom\&.
.SS "Map File (only for VMS/VMU)"
.PP
The VMS map file is a standard JPEG file\&. Its restart markers (if any) are not included in the optimisation file\&. The VMU map file is in NGR format\&. This file can be used to provide a lower\-resolution view of the slide\&.
.SS "Image Files (only for VMS/VMU)"
.PP
These files are given by the VMS/VMU
ImageFile
keys\&. They are assumed to have a height which is a multiple of the MCU height\&. They are assumed to have a width which is a multiple of MCUs per row divided by the restart interval\&.
.PP
For VMS, these files are in JPEG, for VMU they are in NGR format\&.
.SS "NGR Format"
.PP
The NGR file contains uncompressed 16\-bit RGB data, with a small header\&. The files we have encountered start with
GN, two more bytes, and then width, height, and column width in little endian 32\-bit format\&. The column width must divide evenly into the width\&. Column width is important, since NGR files are generated in columns, where the first column comes first in the file, followed by subsequent files\&. Columns are painted left\-to\-right\&.
.PP
At offset 24 is another 32\-bit integer which gives the offset in the file to the start of the image data\&. The image data we have encountered is in 16\-bit little endian format\&.
.SS "Associated Images"
.PP
macro
.RS 4
the image file given by the
MacroImage
value in the VMS/VMU file, or
SourceLens
of \-1 in NDPI
.RE
.SS "Known Properties"
.PP
All key\-value data stored in the VMS/VMU file, and known tags from the NDPI file, are encoded as properties prefixed with \(lqhamamatsu\&.\(rq\&.
.PP
openslide\&.mpp\-x
.RS 4
For VMS, calculated as
hamamatsu\&.PhysicalWidth/(1000*openslide\&.level[0]\&.width)\&. For NDPI, calculated as
10000/tiff\&.XResolution, if
tiff\&.ResolutionUnit
is
centimeter\&.
.RE
.PP
openslide\&.mpp\-y
.RS 4
For VMS, calculated as
hamamatsu\&.PhysicalHeight/(1000*openslide\&.level[0]\&.height)\&. For NDPI, calculated as
10000/tiff\&.YResolution, if
tiff\&.ResolutionUnit
is
centimeter\&.
.RE
.PP
openslide\&.objective\-power
.RS 4
normalized
hamamatsu\&.SourceLens
.RE
.SS "Test Data"
.PP
NDPI format
.RS 4
\m[blue]\fBhttp://openslide\&.cs\&.cmu\&.edu/download/openslide\-testdata/Hamamatsu/\fR\m[]
.RE
.PP
VMS format
.RS 4
\m[blue]\fBhttp://openslide\&.cs\&.cmu\&.edu/download/openslide\-testdata/Hamamatsu\-vms/\fR\m[]
.RE
.SH "LEICA FORMAT"
.PP
Format
.RS 4
single\-file pyramidal tiled BigTIFF with non\-standard metadata
.RE
.PP
File extensions
.RS 4
\&.scn
.RE
.PP
OpenSlide vendor backend
.RS 4
leica
.RE
.SS "Detection"
.PP
Leica slides are stored in single\-file BigTIFF format\&. OpenSlide will detect a file as Leica if:
.sp
.RS 4
.ie n \{\
\h'-04' 1.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  1." 4.2
.\}
The file is TIFF\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 2.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  2." 4.2
.\}
The initial image is tiled\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 3.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  3." 4.2
.\}
The
ImageDescription
tag contains valid XML in either of these namespaces:
<listitem>http://www\&.leica\-microsystems\&.com/scn/2010/03/10 </listitem>
<listitem>http://www\&.leica\-microsystems\&.com/scn/2010/10/01 </listitem>
.RE
.PP
To open Leica files, OpenSlide must be built with libtiff 4 or above\&.
.SS "Relevant TIFF tags"
.TS
allbox tab(:);
lB lB.
T{
Tag
T}:T{
Description
T}
.T&
l l.
T{
ImageDescription
T}:T{
Stores an XML document containing various metadata
T}
.TE
.sp 1
.SS "File Organization"
.PP
The
ImageDescription
tag of the first TIFF directory contains an XML document that defines the structure of the slide\&.
.PP
Leica slides are structured as a collection of images, each of which has multiple dimensions (pyramid levels)\&. The collection has a size, and images have a size and position, measured in nanometers\&. Each dimension has a size in pixels, an optional focal plane number, and a TIFF directory containing the image data\&. Fluorescence images have different dimensions (and thus different TIFF directories) for each channel\&. OpenSlide currently rejects fluorescence images and ignores focal planes other than plane 0\&.
.PP
Brightfield slides have at least two images: a low\-resolution macro image and one or more main images corresponding to regions of the macro image\&. The macro image has a position of (0, 0) and a size matching the size of the collection\&. Fluorescence slides can have two macro images: one brightfield and one fluorescence\&.
.PP
The slide provides enough information to composite the various images, including the macro image, into a single pyramid\&. However, there are some complications:
<listitem>The resolution of the macro image is generally not related to the resolution of the main images by a power of two\&.
</listitem><listitem>Downsampled dimensions are generally downsampled from the next larger dimension by a factor of 4, but main images can be scanned with distinct objectives that may differ by only a factor of 2\&.
</listitem>.PP
Thus, in general, the images in a collection cannot be rendered into a unified pyramid without scaling the original pixel data\&. OpenSlide does not attempt to do this\&. Instead, OpenSlide omits the macro image from the pyramid and refuses to open slides whose main images have inconsistent resolutions\&.
.SS "Associated Images"
.PP
macro
.RS 4
the highest\-resolution dimension of the macro image
.RE
.SS "Known Properties"
.PP
leica\&.aperture
.RS 4
the
numericalAperture
of the main image
.RE
.PP
leica\&.barcode
.RS 4
the
barcode
text\&. (For slides in the
2010/10/01
namespace, OpenSlide 3\&.4\&.0 and earlier report this property as a Base64\-encoded string; OpenSlide 3\&.4\&.1 and later report it in plain text\&. For slides in the
2010/03/10
namespace, OpenSlide reports the barcode as it is stored in the XML, since we do not know whether those barcodes are Base64\-encoded\&. If you have a
2010/03/10
slide with a bar code, please comment in
\m[blue]\fBthis bug\fR\m[]\&\s-2\u[2]\d\s+2
or contact the OpenSlide mailing list\&.)
.RE
.PP
leica\&.creation\-date
.RS 4
the
creationDate
of the main image
.RE
.PP
leica\&.device\-model
.RS 4
the
device
model
of the main image
.RE
.PP
leica\&.device\-version
.RS 4
the
device
version
of the main image
.RE
.PP
leica\&.illumination\-source
.RS 4
the
illuminationSource
of the main image
.RE
.PP
leica\&.objective
.RS 4
the
objective
of the main image
.RE
.PP
openslide\&.mpp\-x
.RS 4
calculated as
10000/tiff\&.XResolution, if
tiff\&.ResolutionUnit
is
centimeter
.RE
.PP
openslide\&.mpp\-y
.RS 4
calculated as
10000/tiff\&.YResolution, if
tiff\&.ResolutionUnit
is
centimeter
.RE
.PP
openslide\&.objective\-power
.RS 4
normalized
leica\&.objective
.RE
.SS "Test Data"
.PP
\m[blue]\fBhttp://openslide\&.cs\&.cmu\&.edu/download/openslide\-testdata/Leica/\fR\m[]
.SH "MIRAX FORMAT"
.PP
Format
.RS 4
multi\-file with very complicated proprietary metadata and indexes
.RE
.PP
File extensions
.RS 4
\&.mrxs
.RE
.PP
OpenSlide vendor backend
.RS 4
mirax
.RE
.SS "Detection"
.PP
OpenSlide will detect a file as MIRAX if:
.sp
.RS 4
.ie n \{\
\h'-04' 1.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  1." 4.2
.\}
The file is not a TIFF\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 2.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  2." 4.2
.\}
The filename ends with
\&.mrxs\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 3.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  3." 4.2
.\}
A directory exists in the same location as the file, with the same name as the file minus the extension\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 4.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  4." 4.2
.\}
A file named
Slidedat\&.ini
exists in the directory\&.
.RE
.SS "Overview"
.PP
MIRAX can store slides in JPEG, PNG, or BMP formats\&. Because JPEG does not allow for large images, and JPEG and PNG provide very poor support for random\-access decoding of part of an image, multiple images are needed to encode a slide\&. To avoid having many individual files, MIRAX packs these images into a small number of data files\&. The index file provides offsets into the data files for each required piece of data\&.
.PP
The camera on MIRAX scanners takes overlapping photos and records the position of each one\&. Each photo is then split into multiple images which do not overlap\&. Overlaps only occur between images that come from different photos\&.
.PP
To generate level
n + 1, each image from level
n
is downsampled by 2 and then concatenated into a new image, 4 old images per new image (2 x 2)\&. This process is repeated for each level, irrespective of image overlaps\&. Therefore, at sufficiently high levels, a single image can contain one or more embedded overlaps of non\-integral width\&.
.SS "Index File"
.PP
The index file starts with a five\-character ASCII version string, followed by the
SLIDE_ID
from the slidedat file\&. The rest of the file consists of 32\-bit little\-endian integers (unaligned), which can be data values or pointers to byte offsets within the index file\&.
.PP
The first two integers point to offset tables for the hierarchical and nonhierarchical roots, respectively\&. These tables contain one record for each
VAL
in the
HIERARCHICAL
slidedat section\&. For example, the record for
NONHIER_1_VAL_2
would be stored at
nonhier_root + 4 * (NONHIER_0_COUNT + 2)\&.
.PP
Each record is a pointer to a linked list of data pages\&. The first two values in a data page are the number of data items in the page and a pointer to the next page\&. The first page always has 0 data items, and the last page has a 0 next pointer\&.
.PP
There is one hierarchical record for each zoom level\&. The record contains data items consisting of an image index, offset and length within a file, and a file number\&. The file number can be converted to a data file name via the
DATAFILE
slidedat section\&. The image index is equal to
image_y * GENERAL\&.IMAGENUMBER_X + image_x\&. Image coordinates which are not multiples of the zoom level\(cqs downsample factor are omitted\&.
.PP
Nonhierarchical records refer to associated images and additional metadata\&. Nonhierarchical data items consist of three zero values followed by an offset, length, and file number as in hierarchical records\&.
.SS "Data Files"
.PP
A data file begins with a header containing a five\-character ASCII version string, the
SLIDE_ID
from the slidedat file, the file number encoded into three ASCII characters, and 256 bytes of padding\&. (In newer slides, the
SLIDE_ID
and file number are encoded as UTF\-16LE, so the second half of each value is truncated away\&.) The remainder of the file contains packed data referenced by the index file\&.
.SS "Slide Position File"
.PP
The slide position file is referenced by the
VIMSLIDE_POSITION_BUFFER\&.default
nonhierarchical section\&. It contains one entry for each camera position (not each image position) in row\-major order\&. Each entry is nine bytes: a flag byte, the
X
pixel coordinate of the photo (4 bytes, little\-endian, may be negative), and the
Y
coordinate (4 bytes, little\-endian, may be negative)\&. In slides with
CURRENT_SLIDE_VERSION
≥ 1\&.9, the flag byte is 1 if the slide file contains images for this camera position, 0 otherwise\&. In older slides, the flag byte is always 0\&.
.PP
In slides with
CURRENT_SLIDE_VERSION
≥ 2\&.2, the slide position file is compressed with DEFLATE and referenced by the
StitchingIntensityLayer\&.StitchingIntensityLevel
nonhierarchical section\&.
.SS "Associated Images"
.PP
thumbnail
.RS 4
the image named \(lqScanDataLayer_SlidePreview\(rq in
Slidedat\&.ini
(optional)
.RE
.PP
label
.RS 4
the image named \(lqScanDataLayer_SlideBarcode\(rq in
Slidedat\&.ini
(optional)
.RE
.PP
macro
.RS 4
the image named \(lqScanDataLayer_SlideThumbnail\(rq in
Slidedat\&.ini
(optional)
.RE
.SS "Known Properties"
.PP
All key\-value data stored in the
Slidedat\&.ini
file are encoded as properties prefixed with \(lqmirax\&.\(rq\&.
.PP
openslide\&.mpp\-x
.RS 4
normalized
MICROMETER_PER_PIXEL_X
from the Slidedat section corresponding to level 0 (typically
mirax\&.LAYER_0_LEVEL_0_SECTION\&.MICROMETER_PER_PIXEL_X)
.RE
.PP
openslide\&.mpp\-y
.RS 4
normalized
MICROMETER_PER_PIXEL_Y
from the Slidedat section corresponding to level 0 (typically
mirax\&.LAYER_0_LEVEL_0_SECTION\&.MICROMETER_PER_PIXEL_Y)
.RE
.PP
openslide\&.objective\-power
.RS 4
normalized
mirax\&.GENERAL\&.OBJECTIVE_MAGNIFICATION
.RE
.SS "See Also"
.PP
\m[blue]\fBIntroduction to MIRAX/MRXS\fR\m[]\&\s-2\u[3]\d\s+2\&. Note that our terminology has changed since that document was written; where it says \(lqtile\(rq, substitute \(lqimage\(rq, and where it says \(lqsubtile\(rq, substitute \(lqtile\(rq\&.
.SS "Test Data"
.PP
\m[blue]\fBhttp://openslide\&.cs\&.cmu\&.edu/download/openslide\-testdata/Mirax/\fR\m[]
.SH "PHILIPS FORMAT"
.PP
Format
.RS 4
single\-file pyramidal tiled TIFF or BigTIFF with non\-standard metadata
.RE
.PP
File extensions
.RS 4
\&.tiff
.RE
.PP
OpenSlide vendor backend
.RS 4
philips
.RE
.SS "Detection"
.PP
Philips TIFF files are stored in single\-file TIFF or BigTIFF format\&. OpenSlide will detect a file as Philips if:
.sp
.RS 4
.ie n \{\
\h'-04' 1.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  1." 4.2
.\}
The file is TIFF\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 2.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  2." 4.2
.\}
The TIFF
Software
tag starts with
Philips\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 3.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  3." 4.2
.\}
The
ImageDescription
tag contains valid XML\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 4.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  4." 4.2
.\}
The root element of the XML is
DataObject
and has an
ObjectType
attribute with a value of
DPUfsImport\&.
.RE
.PP
To open BigTIFF files, OpenSlide must be built with libtiff 4 or above\&.
.SS "File Organization"
.PP
Philips TIFF is an export format\&. The native Philips format, iSyntax, is a custom multi\-file format not currently supported by OpenSlide\&.
.PP
The
ImageDescription
tag of the first TIFF directory contains an XML document with a hierarchical structure containing key\-value pairs\&. The keys are based on DICOM tags\&.
.PP
The level dimensions given in the TIFF
ImageWidth
and
ImageLength
fields, and also in the
ImageDescription
XML, are merely the TIFF tile size multiplied by the number of tiles in each dimension\&. Thus, they include the size of the padding in the right\-most column and bottom\-most row of tiles\&. Each level typically uses the same tile size but requires a different amount of padding, so the aspect ratios of the levels are inconsistent and the level dimensions are not proportional to the level downsamples\&. Correct downsamples can be calculated from the levels\(cq pixel spacings in the XML metadata\&.
.PP
Slides with multiple regions of interest are structured as a single image pyramid enclosing all regions\&. Slides may omit pixel data for TIFF tiles not in an ROI; this is represented as a
TileOffset
of 0 and a
TileByteCount
of 0\&. When such tiles are downsampled into a tile that does contain pixel data, their contents are rendered as white pixels\&.
.PP
Label and macro images are stored as Base64\-encoded JPEGs in the
ImageDescription
XML\&. Some slides also store these images as stripped TIFF directories whose
ImageDescriptions start with
Label
and
Macro, respectively\&.
.SS "Relevant TIFF tags"
.TS
allbox tab(:);
lB lB.
T{
Tag
T}:T{
Description
T}
.T&
l l
l l.
T{
ImageDescription
T}:T{
Stores an XML document containing various metadata and associated image data
T}
T{
Software
T}:T{
Starts with Philips
T}
.TE
.sp 1
.SS "Associated Images"
.PP
label
.RS 4
the TIFF directory with an
ImageDescription
starting with
Label, or the image data in the
DPScannedImage
with a
PIM_DP_IMAGE_TYPE
of
LABELIMAGE
.RE
.PP
macro
.RS 4
the TIFF directory with an
ImageDescription
starting with
Macro, or the image data in the
DPScannedImage
with a
PIM_DP_IMAGE_TYPE
of
MACROIMAGE
.RE
.SS "Known Properties"
.PP
All key\-value data encoded in the
DPUfsImport
object, in the first
DPScannedImage
object with a
PIM_DP_IMAGE_TYPE
of
WSI, and in that object\(cqs
PixelDataRepresentation
objects is represented as properties prefixed with \(lqphilips\&.\(rq\&.
.PP
openslide\&.mpp\-x
.RS 4
calculated as
1000 * philips\&.DICOM_PIXEL_SPACING[1]
.RE
.PP
openslide\&.mpp\-y
.RS 4
calculated as
1000 * philips\&.DICOM_PIXEL_SPACING[0]
.RE
.SS "Test Data"
.PP
No public data available\&. Contact the
\m[blue]\fBmailing list\fR\m[]\&\s-2\u[4]\d\s+2
if you have some\&.
.SH "SAKURA FORMAT"
.PP
Format
.RS 4
SQLite database containing pyramid tiles and metadata
.RE
.PP
File extensions
.RS 4
\&.svslide
.RE
.PP
OpenSlide vendor backend
.RS 4
sakura
.RE
.SS "Detection"
.PP
OpenSlide will detect a file as Sakura if:
.sp
.RS 4
.ie n \{\
\h'-04' 1.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  1." 4.2
.\}
The file is a SQLite database\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 2.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  2." 4.2
.\}
The
DataManagerSQLiteConfigXPO
table contains exactly one row, and its
TableName
field refers to a unique table\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 3.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  3." 4.2
.\}
The unique table contains a row with
id = "++MagicBytes"
and
data = "SVGigaPixelImage"\&.
.RE
.SS "File Organization"
.PP
Sakura slides are SQLite 3 database files written by the eXpress Persistent Objects ORM\&. Tables contain slide metadata, associated images, and JPEG tiles\&. Tiles are addressed as
(focal plane, downsample, level\-0 X coordinate, level\-0 Y coordinate, color channel), with separate grayscale JPEGs for each color channel\&. Despite the generality of the address format, tiles appear to be organized in a regular grid, with power\-of\-two level downsamples and without overlapping tiles\&. The structure of the file allows scans to be sparse, but it is not clear if this is actually done\&.
.SS "SQL Tables"
.PP
Some irrelevant tables and columns have been omitted from the summary below\&.
DataManagerSQLiteConfigXPO.PP
Useful only to get a reference to the unique table\&. OpenSlide requires this table to contain exactly one row\&.
.TS
allbox tab(:);
lB lB lB.
T{
Column
T}:T{
Type
T}:T{
Description
T}
.T&
l l l.
T{
TableName
T}:T{
text
T}:T{
Name of the unique table, described below
T}
.TE
.sp 1
SVSlideDataXPO.PP
High\-level metadata about a slide\&. OpenSlide assumes this table will contain exactly one row\&.
.TS
allbox tab(:);
lB lB lB.
T{
Column
T}:T{
Type
T}:T{
Description
T}
.T&
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l.
T{
OID
T}:T{
integer
T}:T{
Primary key
T}
T{
m_labelScan
T}:T{
integer
T}:T{
Foreign key to label associated image in SVScannedImageDataXPO
T}
T{
m_overviewScan
T}:T{
integer
T}:T{
Foreign key to macro associated image in SVScannedImageDataXPO
T}
T{
SlideId
T}:T{
text
T}:T{
UUID
T}
T{
Date
T}:T{
text
T}:T{
File creation date?
T}
T{
Description
T}:T{
text
T}:T{
Descriptive text?
T}
T{
Creator
T}:T{
text
T}:T{
Author?
T}
T{
DiagnosisCode
T}:T{
text
T}:T{
Unknown, have seen \(lq0\(rq
T}
T{
HRScanCount
T}:T{
integer
T}:T{
Presumably the number of corresponding rows in SVHRScanDataXPO
T}
T{
Keywords
T}:T{
text
T}:T{
Descriptive text?
T}
T{
TotalDataSizeBytes
T}:T{
integer
T}:T{
Presumably the sum of TotalDataSizeBytes in corresponding SVHRScanDataXPO rows
T}
.TE
.sp 1
SVHRScanDataXPO.PP
A single high\-resolution scan of a slide from
SVSlideDataXPO\&. OpenSlide assumes this table will contain exactly one row\&.
.TS
allbox tab(:);
lB lB lB.
T{
Column
T}:T{
Type
T}:T{
Description
T}
.T&
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l.
T{
OID
T}:T{
integer
T}:T{
Primary key
T}
T{
ParentSlide
T}:T{
integer
T}:T{
Foreign key to SVSlideDataXPO
T}
T{
ScanId
T}:T{
text
T}:T{
UUID
T}
T{
Date
T}:T{
text
T}:T{
Scan date?
T}
T{
Description
T}:T{
text
T}:T{
Descriptive text?
T}
T{
Name
T}:T{
text
T}:T{
Scan name?
T}
T{
PosOnSlideMm
T}:T{
blob
T}:T{
16 bytes of binary
T}
T{
ResolutionMmPerPix
T}:T{
real
T}:T{
Millimeters per pixel
T}
T{
NominalLensMagnification
T}:T{
real
T}:T{
Objective power
T}
T{
ThumbnailImage
T}:T{
blob
T}:T{
thumbnail associated image data
T}
T{
TotalDataSizeBytes
T}:T{
integer
T}:T{
Same as TOTAL_SIZE blob in unique table
T}
T{
FocussingMethod
T}:T{
integer
T}:T{
Unknown; have seen \(lq1\(rq
T}
T{
FocusStack
T}:T{
blob
T}:T{
8 bytes of binary per focal plane; the center focal plane apparently has all zeroes
T}
.TE
.sp 1
SVScannedImageDataXPO.PP
Contains associated images other than the
thumbnail\&.
.TS
allbox tab(:);
lB lB lB.
T{
Column
T}:T{
Type
T}:T{
Description
T}
.T&
l l l
l l l
l l l
l l l
l l l
l l l
l l l.
T{
OID
T}:T{
integer
T}:T{
Primary key
T}
T{
Id
T}:T{
text
T}:T{
UUID
T}
T{
PosOnSlideMm
T}:T{
blob
T}:T{
16 bytes of binary
T}
T{
ScanCenterPosMm
T}:T{
blob
T}:T{
16 bytes of binary
T}
T{
ResolutionMmPerPix
T}:T{
real
T}:T{
Millimeters per pixel
T}
T{
Image
T}:T{
blob
T}:T{
JPEG image data
T}
T{
ThumbnailImage
T}:T{
blob
T}:T{
Low\-resolution JPEG thumbnail
T}
.TE
.sp 1
tile.PP
This table is most naturally used to map tile coordinates to tile IDs, but is not suitable for individual lookups because it has no useful indexes\&. In addition, some Sakura slides don\(cqt have it\&. OpenSlide ignores the table and constructs tile IDs directly from tile coordinates\&.
.TS
allbox tab(:);
lB lB lB.
T{
Column
T}:T{
Type
T}:T{
Description
T}
.T&
l l l
l l l
l l l
l l l
l l l.
T{
TILEID
T}:T{
text
T}:T{
Foreign key to unique table
T}
T{
PYRAMIDLEVEL
T}:T{
integer
T}:T{
Downsample of the pyramid level
T}
T{
COLUMNINDEX
T}:T{
integer
T}:T{
Level\-0 X coordinate of the top\-left corner of the tile
T}
T{
ROWINDEX
T}:T{
integer
T}:T{
Level\-0 Y coordinate of the top\-left corner of the tile
T}
T{
COLORINDEX
T}:T{
integer
T}:T{
0 for red, 1 for green, 2 for blue
T}
.TE
.sp 1


Unique table
.PP
This is the table named by
DataManagerSQLiteConfigXPO\&.TableName\&. It contains named blobs including the JPEG tile data\&.
.TS
allbox tab(:);
lB lB lB.
T{
Column
T}:T{
Type
T}:T{
Description
T}
.T&
l l l
l l l
l l l.
T{
id
T}:T{
text
T}:T{
Primary key
T}
T{
size
T}:T{
integer
T}:T{
Length of data field
T}
T{
data
T}:T{
blob
T}:T{
Data item
T}
.TE
.sp 1
.PP
This table stores a variety of blob types\&.
.TS
allbox tab(:);
lB lB.
T{
id
T}:T{
Description
T}
.T&
l l
l l
l l
l l
l l
l l.
T{
++MagicBytes
T}:T{
SVGigaPixelImage
T}
T{
++VersionBytes
T}:T{
Format version, e\&.g\&. 1\&.0\&.0
T}
T{
Header
T}:T{
See below
T}
T{
TOTAL_SIZE
T}:T{
The data field is empty\&.  The size field is the sum of all other size fields except ++MagicBytes and ++VersionBytes\&.
T}
T{
T;2048|4096;4;2;0
T}:T{
Image tile with downsample 4, X coordinate 2048, Y coordinate 4096, channel 2 (blue), focal plane 0
T}
T{
T;2048|4096;4;2;0#
T}:T{
MD5 hash of the T;2048|4096;4;2;0 image tile
T}
.TE
.sp 1
Header blob
.PP
The
Header
blob is a small binary structure containing little\-endian integers as follows:
.TS
allbox tab(:);
lB lB lB.
T{
Offset
T}:T{
Size
T}:T{
Description
T}
.T&
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l
l l l.
T{
0
T}:T{
4
T}:T{
Tile size in pixels
T}
T{
4
T}:T{
4
T}:T{
Image width in pixels
T}
T{
8
T}:T{
4
T}:T{
Image height in pixels
T}
T{
12
T}:T{
4
T}:T{
Unknown; have seen \(lq8\(rq (bits per channel?)
T}
T{
16
T}:T{
4
T}:T{
Number of focal planes
T}
T{
20
T}:T{
4
T}:T{
Unknown; have seen \(lq3\(rq (number of channels?)
T}
T{
24
T}:T{
4
T}:T{
Unknown; have seen \(lq1\(rq
T}
T{
28
T}:T{
2
T}:T{
Unknown; have seen \(lq256\(rq
T}
T{
30
T}:T{
4
T}:T{
Unknown; have seen \(lq1\(rq
T}
T{
34
T}:T{
4
T}:T{
Unknown; have seen \(lq2\(rq
T}
T{
38
T}:T{
4
T}:T{
Unknown; have seen \(lq3\(rq
T}
T{
42
T}:T{
4
T}:T{
Unknown; have seen \(lq4\(rq
T}
T{
46
T}:T{
4
T}:T{
Unknown; have seen \(lq5\(rq
T}
T{
50
T}:T{
4
T}:T{
Unknown; have seen \(lq6\(rq
T}
.TE
.sp 1
.SS "Associated Images"
.PP
label
.RS 4
SVScannedImageDataXPO\&.Image
corresponding to
SVSlideDataXPO\&.m_labelScan
.RE
.PP
macro
.RS 4
SVScannedImageDataXPO\&.Image
corresponding to
SVSlideDataXPO\&.m_overviewScan
.RE
.PP
thumbnail
.RS 4
SVHRScanDataXPO\&.ThumbnailImage
.RE
.SS "Known Properties"
.PP
sakura\&.Creator
.RS 4
SVSlideDataXPO\&.Creator
.RE
.PP
sakura\&.Date
.RS 4
SVSlideDataXPO\&.Date
.RE
.PP
sakura\&.Description
.RS 4
SVSlideDataXPO\&.Description
.RE
.PP
sakura\&.DiagnosisCode
.RS 4
SVSlideDataXPO\&.DiagnosisCode
.RE
.PP
sakura\&.FocussingMethod
.RS 4
SVHRScanDataXPO\&.FocussingMethod
.RE
.PP
sakura\&.Keywords
.RS 4
SVSlideDataXPO\&.Keywords
.RE
.PP
sakura\&.NominalLensMagnification
.RS 4
SVHRScanDataXPO\&.NominalLensMagnification
.RE
.PP
sakura\&.ResolutionMmPerPix
.RS 4
SVHRScanDataXPO\&.ResolutionMmPerPix
.RE
.PP
sakura\&.ScanId
.RS 4
SVHRScanDataXPO\&.ScanId
.RE
.PP
sakura\&.SlideId
.RS 4
SVSlideDataXPO\&.SlideId
.RE
.PP
openslide\&.mpp\-x
.RS 4
calculated as
1000 * sakura\&.ResolutionMmPerPix
.RE
.PP
openslide\&.mpp\-y
.RS 4
calculated as
1000 * sakura\&.ResolutionMmPerPix
.RE
.PP
openslide\&.objective\-power
.RS 4
normalized
sakura\&.NominalLensMagnification
.RE
.SS "Test Data"
.PP
No public data available\&. Contact the
\m[blue]\fBmailing list\fR\m[]\&\s-2\u[4]\d\s+2
if you have some\&.
.SH "TRESTLE FORMAT"
.PP
Format
.RS 4
single\-file pyramidal tiled TIFF, with non\-standard metadata and overlaps; additional files contain more metadata and detailed overlap info
.RE
.PP
File extensions
.RS 4
\&.tif
.RE
.PP
OpenSlide vendor backend
.RS 4
trestle
.RE
.SS "Detection"
.PP
Trestle slides are stored in single\-file TIFF format\&. OpenSlide will detect a file as Trestle if:
.sp
.RS 4
.ie n \{\
\h'-04' 1.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  1." 4.2
.\}
The file is TIFF\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 2.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  2." 4.2
.\}
The TIFF
Software
tag starts with
MedScan\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 3.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  3." 4.2
.\}
The
ImageDescription
tag is present\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 4.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  4." 4.2
.\}
All images are tiled\&.
.RE
.SS "Relevant TIFF tags"
.TS
allbox tab(:);
lB lB.
T{
Tag
T}:T{
Description
T}
.T&
l l
l l
l l.
T{
ImageDescription
T}:T{
Stores some important key\-value pairs, see below
T}
T{
Software
T}:T{
Starts with \(lqMedScan\(rq
T}
T{
XResolution, YResolution
T}:T{
Seems to store microns\-per\-pixel (MPP), which may or may not take into account the correct objective power\&. Note that this is inverted from standard TIFF, which stores pixels\-per\-unit, not units\-per\-pixel\&.
T}
.TE
.sp 1
.SS "Extra data stored in ImageDescription"
.PP
The
ImageDescription
tag contains semicolon\-delimited key\-value pairs\&. A key\-value pair is equals\-delimited\&. We use the
OverlapsXY
and
Background Color
keys from the
ImageDescription, and ignore the rest\&. All of these values are stored as properties starting with \(lqtrestle\&.\(rq\&.
.TS
allbox tab(:);
lB lB.
T{
Key
T}:T{
Description
T}
.T&
l l
l l
l l
l l
l l.
T{
Background Color
T}:T{
Hex\-encoded background color info, assumed to be in the format RRGGBB\&.
T}
T{
White Balance
T}:T{
Hex\-encoded white balance
T}
T{
Objective Power
T}:T{
Reported objective power, often incorrect\&.
T}
T{
JPEG Quality
T}:T{
The JPEG quality value\&.
T}
T{
OverlapsXY
T}:T{
Overlaps, see below\&.
T}
.TE
.sp 1
.SS "TIFF Image Directory Organization"
.PP
The first image in the TIFF file is the full\-resolution image\&. The subsequent images are assumed to be decreasingly sized reduced\-resolution images\&.
.SS "Overlaps"
.PP
The
OverlapsXY
pseudo\-field encodes a list of tile overlap values as ASCII\&.
.PP
Example: \(lq64 64 32 32 16 16\(rq (note the initial space)\&.
.PP
These values represent the standard overlaps between adjacent tiles in X and Y, in pixels\&. This example encodes 3 levels worth of overlaps\&. Further overlaps are assumed to have the value 0\&.
.PP
Individual tile overlaps may differ from the standard overlaps\&. These individual overlaps are recorded in
\&.tif\-Nb
files adjacent to the
\&.tif
file, where
N
is the level number\&. OpenSlide does not read these files, though they have been partially decoded; see
\m[blue]\fBissue 21\fR\m[]\&\s-2\u[5]\d\s+2
for details\&.
.SS "Associated Images"
.PP
macro
.RS 4
the image with a filename extension of \(lq\&.Full\(rq (optional)
.RE
.SS "Known Properties"
.PP
All data encoded in the
ImageDescription
TIFF field is represented as properties prefixed with \(lqtrestle\&.\(rq\&.
.PP
openslide\&.mpp\-x
.RS 4
copy of
tiff\&.XResolution
(note that this is a totally non\-standard use of this TIFF tag)
.RE
.PP
openslide\&.mpp\-y
.RS 4
copy of
tiff\&.YResolution
(note that this is a totally non\-standard use of this TIFF tag)
.RE
.PP
openslide\&.objective\-power
.RS 4
normalized
trestle\&.Objective Power
.RE
.SS "Test Data"
.PP
\m[blue]\fBhttp://openslide\&.cs\&.cmu\&.edu/download/openslide\-testdata/Trestle/\fR\m[]
.SH "VENTANA FORMAT"
.PP
Format
.RS 4
single\-file pyramidal tiled BigTIFF with non\-standard metadata and overlaps
.RE
.PP
File extensions
.RS 4
\&.bif,
\&.tif
.RE
.PP
OpenSlide vendor backend
.RS 4
ventana
.RE
.SS "Detection"
.PP
Ventana slides are stored in single\-file BigTIFF format\&. OpenSlide will detect a file as Ventana if:
.sp
.RS 4
.ie n \{\
\h'-04' 1.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  1." 4.2
.\}
The file is TIFF\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 2.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  2." 4.2
.\}
The
XMP
tag contains valid XML\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 3.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  3." 4.2
.\}
The XML contains an
iScan
element, either as the root element or as a child of a
Metadata
root element\&.
.RE
.PP
To open Ventana files, OpenSlide must be built with libtiff 4 or above\&.
.SS "Associated Images"
.PP
macro
.RS 4
the TIFF directory whose
ImageDescription
is
Label Image
or
Label_Image
.RE
.PP
thumbnail
.RS 4
the TIFF directory whose
ImageDescription
is
Thumbnail
.RE
.SS "Known Properties"
.PP
All XML attributes in the
iScan
element are represented as properties prefixed with \(lqventana\&.\(rq\&.
.PP
openslide\&.mpp\-x
.RS 4
normalized
ventana\&.ScanRes
.RE
.PP
openslide\&.mpp\-y
.RS 4
normalized
ventana\&.ScanRes
.RE
.PP
openslide\&.objective\-power
.RS 4
normalized
ventana\&.Magnification
.RE
.SS "Test Data"
.PP
\m[blue]\fBhttp://openslide\&.cs\&.cmu\&.edu/download/openslide\-testdata/Ventana/\fR\m[]
.SH "GENERIC TILED TIFF FORMAT"
.PP
Format
.RS 4
single\-file pyramidal tiled TIFF
.RE
.PP
File extensions
.RS 4
\&.tif
.RE
.PP
OpenSlide vendor backend
.RS 4
generic\-tiff
.RE
.SS "Detection"
.PP
OpenSlide will detect a file as generic TIFF if:
.sp
.RS 4
.ie n \{\
\h'-04' 1.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  1." 4.2
.\}
No other detections succeed\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 2.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  2." 4.2
.\}
The file is TIFF\&.
.RE
.sp
.RS 4
.ie n \{\
\h'-04' 3.\h'+01'\c
.\}
.el \{\
.sp -1
.IP "  3." 4.2
.\}
The initial image is tiled\&.
.RE
.SS "TIFF Image Directory Organization"
.PP
The first image in the TIFF file is the full\-resolution image\&. Any other tiled images in the file with the \(lqreduced resolution\(rq bit set are assumed to be reduced\-resolution versions of the original\&.
.SS "Associated Images"
.PP
None\&.
.SS "Known Properties"
.PP
Many TIFF tags are encoded as properties starting with \(lqtiff\&.\(rq\&.
.SS "Test Data"
.PP
\m[blue]\fBhttp://openslide\&.cs\&.cmu\&.edu/download/openslide\-testdata/Generic\-TIFF/\fR\m[]
.SH "AUTHORS"
.PP
The Carnegie Mellon School of Computer Science\&.
.PP
This manual page was written by Mathieu Malaterre <malat@debian\&.org> for the Debian GNU/Linux system (but may be used by others)\&.
.SH "NOTES"
.IP " 1." 4
pages for each vendor format
.RS 4
\%http://openslide.org/formats/
.RE
.IP " 2." 4
this bug
.RS 4
\%http://openslide.orghttps://github.com/openslide/openslide/issues/155
.RE
.IP " 3." 4
Introduction to MIRAX/MRXS
.RS 4
\%http://openslide.orghttps://lists.andrew.cmu.edu/pipermail/openslide-users/2012-July/000373.html
.RE
.IP " 4." 4
mailing list
.RS 4
\%http://openslide.orghttps://lists.andrew.cmu.edu/mailman/listinfo/openslide-users/
.RE
.IP " 5." 4
issue 21
.RS 4
\%http://openslide.orghttps://github.com/openslide/openslide/issues/21#issuecomment-23615583
.RE