Metadata editability

[Jason Swedlow]

Dear All-

Apologies-- this is not a direct confocal question, but it does affect use and analysis of confocal data.  If you don't care about image metadata, then just ignore and delete.

As the OME project moves towards release of OMERO-Beta4 (http://trac.openmicroscopy.org.uk/omero/roadmap), we have a number of issues coming up we'd like feedback on.

The first is metadata editability.  In Beta4, we've gone for something we call "metadata completion".  This means that, for a given image file format, we capture and find a home for all of the metadata in that format which fits into OMERO.  For some formats, that's easy, because there is so little metadata.  But many are quite rich, and this project has been a huge effort by the Bio-Formats (Melissa Linkert) and OMERO (Brian Loranger, Chris Allan, Jean-Marie Burel) teams.

The result is that we will support 5 rich file formats in Beta4 "completely".  Note that we have to make decision about what each piece of metadata means-- we certify that it has been imported into OMERO, although there are a few edge cases where we've had to make decisions about where each piece of metadata goes.

This raises a critical question, that we have debated within OME for years, namely:

What image metadata should be editable? Imagine that some value was either unset or wrongly set on the microscope, a user may want to correct the situation after import. Then, if we allow editing, how much info about that editing should we track?

Some technical points:

-- in OMERO, we store every write to the DB as an Event.  So we know every change, who did it and when.

-- we really do NOT want to store every metadata change made.  Doing this properly means making multiple copies of every database entry every time one thing changes-- so things get bloated very rapidly.

A compromise we are considering seriously-- keep only the last metadata value, but log all the changes.  If a user has changed any metadata that came with an image fille, we can add functionality to the client applications to let the user find this out.  You'll know that the data has changed, by whom, and when, you just wont know what the previous data was.

If necessary, we could implement an audit database, which stores previous metadata versions, or maybe just the first version-- the one that was acquired with the original .  This would be optional-- if needed it could be turned on.  We won't get that done for the first Beta4.0 release, but if it's a priority, it could come later this year.

Thanks for your ideas and comments.

Cheers,

Jason


--
**************************
Wellcome Trust Centre for Gene Regulation & Expression
College of Life Sciences
MSI/WTB/JBC Complex
University of Dundee
Dow Street
Dundee  DD1 5EH
United Kingdom

phone [phone]
Intl phone:  [phone]
FAX   [phone]
email: [hidden email]

Lab Page: http://www.dundee.ac.uk/lifesciences/swedlow/
Open Microscopy Environment: http://openmicroscopy.org
**************************

[Chris Tully]

Jason,

Having been involved in the FDA compliance issues for image processing
software, here are my suggestions:

1) Have a fairly granular set of controls for who can edit which
metadata.  I.e. the lab manager can but the user can't.  Or maybe in
the case of a manual microscope all users need to be able to edit
magnification but only the manager can edit acquisition time.
2) An optional audit database should have the option to only keep the
original metadata or keep it all.
3) Consider how changes are made and log the minimum set of changes to
reproduce the result.  For example, if I change the acquisition times
to reflect a 10 ms interval because the acquisition software failed to
record actual acquisition times, there is no need to record the change
for each frame of a sequence.  Instead just record the change as a
global instruction.

In my experience MOST people do not need the Audit features but for
those who do, they are a non-optional item - they MUST be there and be
granular and robust.

Chris

Chris Tully
Microscopy and Image Analysis Expert
[hidden email]
240-888-1021
http://www.linkedin.com/in/christully



On Sun, Feb 1, 2009 at 3:09 PM, Jason Swedlow <[hidden email]> wrote:

> Dear All-
>
> Apologies-- this is not a direct confocal question, but it does affect use
> and analysis of confocal data.  If you don't care about image metadata, then
> just ignore and delete.
>
> As the OME project moves towards release of OMERO-Beta4
> (http://trac.openmicroscopy.org.uk/omero/roadmap), we have a number of
> issues coming up we'd like feedback on.
>
> The first is metadata editability.  In Beta4, we've gone for something we
> call "metadata completion".  This means that, for a given image file format,
> we capture and find a home for all of the metadata in that format which fits
> into OMERO.  For some formats, that's easy, because there is so little
> metadata.  But many are quite rich, and this project has been a huge effort
> by the Bio-Formats (Melissa Linkert) and OMERO (Brian Loranger, Chris Allan,
> Jean-Marie Burel) teams.
>
> The result is that we will support 5 rich file formats in Beta4
> "completely".  Note that we have to make decision about what each piece of
> metadata means-- we certify that it has been imported into OMERO, although
> there are a few edge cases where we've had to make decisions about where
> each piece of metadata goes.
>
> This raises a critical question, that we have debated within OME for years,
> namely:
>
> What image metadata should be editable? Imagine that some value was either
> unset or wrongly set on the microscope, a user may want to correct the
> situation after import. Then, if we allow editing, how much info about that
> editing should we track?
>
> Some technical points:
>
> -- in OMERO, we store every write to the DB as an Event.  So we know every
> change, who did it and when.
>
> -- we really do NOT want to store every metadata change made.  Doing this
> properly means making multiple copies of every database entry every time one
> thing changes-- so things get bloated very rapidly.
>
> A compromise we are considering seriously-- keep only the last metadata
> value, but log all the changes.  If a user has changed any metadata that
> came with an image fille, we can add functionality to the client
> applications to let the user find this out.  You'll know that the data has
> changed, by whom, and when, you just wont know what the previous data was.
>
> If necessary, we could implement an audit database, which stores previous
> metadata versions, or maybe just the first version-- the one that was
> acquired with the original .  This would be optional-- if needed it could be
> turned on.  We won't get that done for the first Beta4.0 release, but if
> it's a priority, it could come later this year.
>
> Thanks for your ideas and comments.
>
> Cheers,
>
> Jason
>
>
> --
> **************************
> Wellcome Trust Centre for Gene Regulation & Expression
> College of Life Sciences
> MSI/WTB/JBC Complex
> University of Dundee
> Dow Street
> Dundee  DD1 5EH
> United Kingdom
>
> phone [phone]
> Intl phone:  [phone]
> FAX   [phone]
> email: [hidden email]
>
> Lab Page: http://www.dundee.ac.uk/lifesciences/swedlow/
> Open Microscopy Environment: http://openmicroscopy.org
> **************************
>

[mahogny]

>
> What image metadata should be editable? Imagine that some value was either
> unset or wrongly set on the microscope, a user may want to correct the
> situation after import. Then, if we allow editing, how much info about that
> editing should we track?
>  
I suggest a rich pluggable system - chance is sites anyway need to do
local adaptions if they have a policy. I suspect if you try to get too
much into details yourself it will quickly get out of hand.

for those who don't have a policy, they take whatever they get. and if
they want to change something, they will - don't attempt to hardcode any
locking down policies.

/Johan

--
--
------------------------------------------------
Johan Henriksson
MSc Engineering
PhD student, Karolinska Institutet
http://mahogny.areta.org http://www.endrov.net

[Arne Seitz-2]

Dear Jason,

 

personally I don’t feel very comfortable when I think of the possibility of editing meta data. Although I totally agree that there might be cases where it is needed.

Maybe it might be a good idea to divide parameters in two categories namely changeable and unchangeable. E.g. I don’t see a reason why the photomultiplier gain of a confocal microscope needed to be changed by the user afterwards. And this is probably true for a lot of other parameters as well.

 

For parameters where it makes sense to change every change should be clearly traceable. I don’t see the need to make a copy of the entire dataset but the changed values should be stored. By saying this if have to admit that I’m technically too ignorant to foresee whether this can be done easily.

 

 

Cheers Arne

 

____________________________________________________

 

Dr. Arne Seitz, Scientific Officer

Advanced Light Microscopy Facility

EMBL Heidelberg

 

[phone]

---

Von: Confocal Microscopy List [mailto:[hidden email]] Im Auftrag von Jason Swedlow
Gesendet: Sonntag, 1. Februar 2009 21:10
An: [hidden email]
Betreff: Metadata editability

 

Dear All-

Apologies-- this is not a direct confocal question, but it does affect use and analysis of confocal data.  If you don't care about image metadata, then just ignore and delete.

As the OME project moves towards release of OMERO-Beta4 (http://trac.openmicroscopy.org.uk/omero/roadmap), we have a number of issues coming up we'd like feedback on.

The first is metadata editability.  In Beta4, we've gone for something we call "metadata completion".  This means that, for a given image file format, we capture and find a home for all of the metadata in that format which fits into OMERO.  For some formats, that's easy, because there is so little metadata.  But many are quite rich, and this project has been a huge effort by the Bio-Formats (Melissa Linkert) and OMERO (Brian Loranger, Chris Allan, Jean-Marie Burel) teams.

The result is that we will support 5 rich file formats in Beta4 "completely".  Note that we have to make decision about what each piece of metadata means-- we certify that it has been imported into OMERO, although there are a few edge cases where we've had to make decisions about where each piece of metadata goes.

This raises a critical question, that we have debated within OME for years, namely:

What image metadata should be editable? Imagine that some value was either unset or wrongly set on the microscope, a user may want to correct the situation after import. Then, if we allow editing, how much info about that editing should we track?

Some technical points:

-- in OMERO, we store every write to the DB as an Event.  So we know every change, who did it and when.

-- we really do NOT want to store every metadata change made.  Doing this properly means making multiple copies of every database entry every time one thing changes-- so things get bloated very rapidly.

A compromise we are considering seriously-- keep only the last metadata value, but log all the changes.  If a user has changed any metadata that came with an image fille, we can add functionality to the client applications to let the user find this out.  You'll know that the data has changed, by whom, and when, you just wont know what the previous data was.

If necessary, we could implement an audit database, which stores previous metadata versions, or maybe just the first version-- the one that was acquired with the original .  This would be optional-- if needed it could be turned on.  We won't get that done for the first Beta4.0 release, but if it's a priority, it could come later this year.

Thanks for your ideas and comments.

Cheers,

Jason


--
**************************
Wellcome Trust Centre for Gene Regulation & Expression
College of Life Sciences
MSI/WTB/JBC Complex
University of Dundee
Dow Street
Dundee  DD1 5EH
United Kingdom

phone [phone]
Intl phone:  [phone]
FAX   [phone]
email: [hidden email]

Lab Page: http://www.dundee.ac.uk/lifesciences/swedlow/
Open Microscopy Environment: http://openmicroscopy.org
**************************

[Rietdorf, Jens]

Dear Jason,

 

I’d prefer different access levels to meta data, pretty much like the access rights to files. Some fields, like comments, need access by all, user, scientific supervisor, instrument supervisor, unit head, others will need the possibility to restrict access. All meta data need to be accessible at some level. All changes should be logged in a history including info about who applied changes and when changes were applied.

I would also fancy some access needs, for fields that should not stay empty, for example if an image has no value without the information which labeling method was used, this input should be required by the scientist (and not be allowed to any other level) before the dataset is saved. Some changes may require a validation by the supervisor(s).

 

Just my 2cents, jens

---
dr. jens rietdorf[hidden email]
head microscopy
novartis research foundation
friedrich-miescher-institute, wro1066.2.16
maulbeerstr.66, 4058 basel, switzerland
couriel:rietdorf(at)fmi(dot)ch
fon: +41.61.69.75172
fax: [phone]

 

jens

 

From: Confocal Microscopy List [mailto:[hidden email]] On Behalf Of Jason Swedlow
Sent: Sunday, February 01, 2009 21:10 PM
To: [hidden email]
Subject: Metadata editability

 

Dear All-

Apologies-- this is not a direct confocal question, but it does affect use and analysis of confocal data.  If you don't care about image metadata, then just ignore and delete.

As the OME project moves towards release of OMERO-Beta4 (http://trac.openmicroscopy.org.uk/omero/roadmap), we have a number of issues coming up we'd like feedback on.

The first is metadata editability.  In Beta4, we've gone for something we call "metadata completion".  This means that, for a given image file format, we capture and find a home for all of the metadata in that format which fits into OMERO.  For some formats, that's easy, because there is so little metadata.  But many are quite rich, and this project has been a huge effort by the Bio-Formats (Melissa Linkert) and OMERO (Brian Loranger, Chris Allan, Jean-Marie Burel) teams.

The result is that we will support 5 rich file formats in Beta4 "completely".  Note that we have to make decision about what each piece of metadata means-- we certify that it has been imported into OMERO, although there are a few edge cases where we've had to make decisions about where each piece of metadata goes.

This raises a critical question, that we have debated within OME for years, namely:

What image metadata should be editable? Imagine that some value was either unset or wrongly set on the microscope, a user may want to correct the situation after import. Then, if we allow editing, how much info about that editing should we track?

Some technical points:

-- in OMERO, we store every write to the DB as an Event.  So we know every change, who did it and when.

-- we really do NOT want to store every metadata change made.  Doing this properly means making multiple copies of every database entry every time one thing changes-- so things get bloated very rapidly.

A compromise we are considering seriously-- keep only the last metadata value, but log all the changes.  If a user has changed any metadata that came with an image fille, we can add functionality to the client applications to let the user find this out.  You'll know that the data has changed, by whom, and when, you just wont know what the previous data was.

If necessary, we could implement an audit database, which stores previous metadata versions, or maybe just the first version-- the one that was acquired with the original .  This would be optional-- if needed it could be turned on.  We won't get that done for the first Beta4.0 release, but if it's a priority, it could come later this year.

Thanks for your ideas and comments.

Cheers,

Jason


--
**************************
Wellcome Trust Centre for Gene Regulation & Expression
College of Life Sciences
MSI/WTB/JBC Complex
University of Dundee
Dow Street
Dundee  DD1 5EH
United Kingdom

phone [phone]
Intl phone:  [phone]
FAX   [phone]
email: [hidden email]

Lab Page: http://www.dundee.ac.uk/lifesciences/swedlow/
Open Microscopy Environment: http://openmicroscopy.org
**************************

[Glen MacDonald-2]

Dear Jason,
The most flexible approach may be something like storing the original  
metadata written by the instrument with limited access for alteration,  
but allow a working copy that the user can edit.  If someone is going  
to fabricate metadata or muck it up out of inexperience, they will do  
it regardless of the database.  There is too much variability to  
create a foolproof audit trail for metadata and still have a useful  
database.  There are many imaging systems in use that do not have full  
metadata capability, like the older confocals with manual filters and  
lens turrets,  and many homebrew devices.  Systems can get  
reconfigured with lenses, filters, etc., but the metadata may not get  
updated, especially for short term changes. I've encountered  
commercial acquisition software with bugs that required editing the  
metadata.  And, as previously mentioned, most annotations for the  
experimental conditions are done off-line.
My 2 cents is to keep it simple.

Regards and thanks for pushing on this project,
Glen

> Dear All-
>
> Apologies-- this is not a direct confocal question, but it does  
> affect use and analysis of confocal data.  If you don't care about  
> image metadata, then just ignore and delete.
>
> As the OME project moves towards release of OMERO-Beta4 (http://trac.openmicroscopy.org.uk/omero/roadmap 
> ), we have a number of issues coming up we'd like feedback on.
>
> The first is metadata editability.  In Beta4, we've gone for  
> something we call "metadata completion".  This means that, for a  
> given image file format, we capture and find a home for all of the  
> metadata in that format which fits into OMERO.  For some formats,  
> that's easy, because there is so little metadata.  But many are  
> quite rich, and this project has been a huge effort by the Bio-
> Formats (Melissa Linkert) and OMERO (Brian Loranger, Chris Allan,  
> Jean-Marie Burel) teams.
>
> The result is that we will support 5 rich file formats in Beta4  
> "completely".  Note that we have to make decision about what each  
> piece of metadata means-- we certify that it has been imported into  
> OMERO, although there are a few edge cases where we've had to make  
> decisions about where each piece of metadata goes.
>
> This raises a critical question, that we have debated within OME for  
> years, namely:
>
> What image metadata should be editable? Imagine that some value was  
> either unset or wrongly set on the microscope, a user may want to  
> correct the situation after import. Then, if we allow editing, how  
> much info about that editing should we track?

Glen MacDonald
Core for Communication Research
Virginia Merrill Bloedel Hearing Research Center
Box 357923
University of Washington
Seattle, WA 98195-7923  USA
[phone]
[hidden email]

******************************************************************************
The box said "Requires WindowsXP or better", so I bought a Macintosh.
******************************************************************************

[Jason Swedlow]

Dear All-

As Guy said, "...there is no single answer".

Thanks much for all the feedback-- this is very helpful.  We will take this all on board and release functionality that satisfies most of what you've specified over 2009.  In our first release, Beta4.0, you won't see very much-- we've shut down editing in our clients, for the moment, but are building the various structures we need on the server.  We'll let you know as these become available-- look for a release in the next 2 weeks, and then another May/June.

Thanks again, listers.  A truly great community!

Cheers,

Jason

On Tue, Feb 3, 2009 at 6:17 PM, Glen MacDonald <[hidden email]> wrote:

Dear Jason,
The most flexible approach may be something like storing the original metadata written by the instrument with limited access for alteration, but allow a working copy that the user can edit.  If someone is going to fabricate metadata or muck it up out of inexperience, they will do it regardless of the database.  There is too much variability to create a foolproof audit trail for metadata and still have a useful database.  There are many imaging systems in use that do not have full metadata capability, like the older confocals with manual filters and lens turrets,  and many homebrew devices.  Systems can get reconfigured with lenses, filters, etc., but the metadata may not get updated, especially for short term changes. I've encountered commercial acquisition software with bugs that required editing the metadata.  And, as previously mentioned, most annotations for the experimental conditions are done off-line.
My 2 cents is to keep it simple.

Regards and thanks for pushing on this project,
Glen

Dear All-

Apologies-- this is not a direct confocal question, but it does affect use and analysis of confocal data.  If you don't care about image metadata, then just ignore and delete.

As the OME project moves towards release of OMERO-Beta4 (http://trac.openmicroscopy.org.uk/omero/roadmap), we have a number of issues coming up we'd like feedback on.

The first is metadata editability.  In Beta4, we've gone for something we call "metadata completion".  This means that, for a given image file format, we capture and find a home for all of the metadata in that format which fits into OMERO.  For some formats, that's easy, because there is so little metadata.  But many are quite rich, and this project has been a huge effort by the Bio-Formats (Melissa Linkert) and OMERO (Brian Loranger, Chris Allan, Jean-Marie Burel) teams.

The result is that we will support 5 rich file formats in Beta4 "completely".  Note that we have to make decision about what each piece of metadata means-- we certify that it has been imported into OMERO, although there are a few edge cases where we've had to make decisions about where each piece of metadata goes.

This raises a critical question, that we have debated within OME for years, namely:

What image metadata should be editable? Imagine that some value was either unset or wrongly set on the microscope, a user may want to correct the situation after import. Then, if we allow editing, how much info about that editing should we track?

Glen MacDonald
Core for Communication Research
Virginia Merrill Bloedel Hearing Research Center
Box 357923
University of Washington
Seattle, WA 98195-7923  USA
[phone]
[hidden email]

******************************************************************************
The box said "Requires WindowsXP or better", so I bought a Macintosh.
******************************************************************************

--
**************************
Wellcome Trust Centre for Gene Regulation & Expression
College of Life Sciences
MSI/WTB/JBC Complex
University of Dundee
Dow Street
Dundee  DD1 5EH
United Kingdom

phone [phone]
Intl phone:  [phone]
FAX   [phone]
email: [hidden email]

Lab Page: http://www.dundee.ac.uk/lifesciences/swedlow/
Open Microscopy Environment: http://openmicroscopy.org
**************************