some advice needed how to manage an unusual image collection

Post Reply
fbungarz
Posts: 1774
Joined: 08 Dec 06 4:03
Location: Arizona, USA

some advice needed how to manage an unusual image collection

Post by fbungarz »

People who occasionally read my posts here on the forum probably know that I am using PSU to manage my large collection of images of lichens. I managed to convince Hert to have PSU natively support the DarwinCore as XMP. It is great to manage the images of the specimens that I collect.

Recently it occurred to me that there might be another use of PSU, to manage my collections of images of thin-layer chromatography plates. Lichens have complex secondary metabolites that can be extracted with acetone. These extracts then can be placed on a silica coated glass plate, a solvent migrating through the silica will drag along the secondary metabolites. Some have a stronger affinity to a particular solvent, these secondary metabolites travel faster than others. The metabolites also have different colors in visible and UV light of different wavelengths, and these colors differ before and after treating them with sulfuric acid and heat.

The image below is one single plate with 13 lichen extracts (the lines numbered 13-25). The image is a composite of four, at the top the plate before it was treated with sulfuric acid, at two different wavelengths of UV light. The upper right image of the plate shows a bright green fluorescence because of a fluorescent indicator embedded into the silica plate, which facilitates viewing how the substances traveled on the plate. The image on the upper left show distinctive colors. After treatment with sulfuric acid the substances develop a distinct color that can also be seen in visible light (on the lower left), and these colors now show a distinctly different color reaction.

All these aspects are useful in determining what secondary metabolites are present in a lichen. And which particular substances are present in turn helps to distinguish/identify the different species. I have a small database program that can be fed with all that information, filtering out possible results. Once the plates have been analyzed I use Photoshop to place small captions on these images, which spots represent what particular substance. Thus, the first two lichens on this particular plate (row #13 and #14) for example contain usnic acid and zeorin:
PSU_example-TLC.jpg
PSU_example-TLC.jpg (210.83 KiB) Viewed 660 times
One challenge identifying the secondary metabolites is the inability to compare plates side-by-side. That's one way I thought PSU's light table might be useful. But I would also like to database what secondary metabolites are present and already identified on a particular plate. That's where I thought PSU labels might be useful.

I have been thinking of tagging the plates with:
  • Date when the plate was run
  • person who ran the plate
  • plate number
  • solvent
All straightforward so far.
But here is where things get complicated:
Each row represents one particular lichen specimen. I could use area-tags for marking these rows.
But on each of the four images in the composite figure there are 13 separate rows. In the example above four times row number row #13, four times row #14, etc. drawing and tagging all those can take quite some time...

Worse: each row contains colorful spots, each spot a different metabolite. Again as an example: usnic acid in rows #13 and #14 is dark (black) in UV light (both before and after treatment with sulfuric acid), but dark green in visible light. Zeorin cannot be seen before applying sulfuric acid, but is pale pinkish violet afterwards, both in visible and UV light (the lower two images of the plate).

Area-tagging all rows on all plates and all the different spots these images might become really cluttered. So, I thought I should just use labels for the 13 specimens analyzed on a plate, not using area tags, and use area tags only for the spots (i.e., the secondary metabolites).
Or perhaps not use area tags at all, but simply continue labeling the spots in Photoshop?

I know this might seem like a rather unusual request, but I thought I still explain this here and perhaps someone has some better ideas...

Thank you!!!
snowman1
Posts: 375
Joined: 01 Jan 07 2:13
Location: UK
Contact:

Re: some advice needed how to manage an unusual image collection

Post by snowman1 »

A few random thoughts, all of which I am sure you have already thought of:
- Do the spots need to be marked up on the image or will labelling be adequate? I'm guessing the former in which case...
- if you will continue to "label" the spots in photoshop then is there any advantage to also making these areas in PSU? PSU labels would be simpler to manage than areas, no? (I could however see the point if you intended using PSU areas instead of the PS text).
- I'm guessing you will need to tie the row numbers to the lichens they represent and I'm not sure how you would achieve this (BTW I would have called these columns from the way they are presented here so I hope I am understanding correctly). Presumably you will want to search by lichen name, but also to know, for any particular plate, what row number represents what lichen? Perhaps having labels of "13 - lichen x", "14 - lichen y" will serve your purposes and searches, even if on another plate lichen x is row #17 ?

The way to approach this would be to list your use cases and if necessary decide which ones are important and which ones you can do without, and this should lead towards a conclusion of whether PSU is suitable and if so which representations will serve your needs best.

I'm conscious that in writing this I am ignoring (through ignorance) two potentially important ways that PSU could help, (1) the use of XMP structures in PSU, which you will be far more familiar with than I given your experience the Darwin scheme, and (2) the potential of macros to help achieve your ends.

I doubt this answer really helps much, but good luck with it! One to try in an experimental catalog at first I think, perhaps even to keep in a separate catalog permanently.
Snowman1
http://www.flickr.com/photos/snowman-1/
--------------------------------------
gcorbin
Posts: 95
Joined: 21 Aug 06 11:31
Location: Brisbane

Re: some advice needed how to manage an unusual image collection

Post by gcorbin »

Wow. You identify lichen. I thought identifying fungi was difficult but lichen is even worse. You are a sucker for punishment.

To the point, my thought on your issue is that you have a plate of 13 lichen samples treated multiple ways creating 4 plate photos per 13 samples. Your issue is that you need to catalogue 13 different objects in your plate photos in an intelligent efficient manner. I don’t know that PSU or any other cataloguing program could handle this specialist requirement.

However, I think the key here is that each plate photograph is actually 13 different samples. My thought is why not separate each photo into 13 different slices, that is a tall thin photo. (This slicing could be automated so a simple task.) Now you have 13 photos, one lichen per photo plus another 3 sets of 13 photos for the different treatments of the sample. This could be recorded in PSU as 13 stacks of 4 photos, one stack per lichen sample, again automated using PSUs powerful image import functionality. Now you have one lichen per photo stack so the samples could be catalogued using standard Darwin core functionality and the different treatment photos catalogued as required with the substances, etc in the standard PSU manner.

The only thing I don’t know is how PSU would handle tall thin photos in light table, etc. I will leave you to test this if you think creating sample slices is a possible solution.
Mke
Posts: 651
Joined: 15 Jun 14 14:39

Re: some advice needed how to manage an unusual image collection

Post by Mke »

Is it normal that the green slide isn't labelled? If so, I'd use that one (only) to identify the specimens, using areas, and identify metabolites (only) on the other 3 slides. Obviously you turn off areas if you want to view the green slide uncluttered.

Based on the way that PSU currently displays area text, I'd turn the sides through 90° to make space for them horizontally - presumably orientation isn't important, just what you're used to. Or talk nicely to Hert to make the label text orientation configurable?

Thanks to a script Hert kindly wrote for me, you can now cut and paste areas (including areas that aren't connected to a label) between images. So you could create a master image with 13 disconnected areas, paste the areas to other slides, add specimen labels, then connect each area to the appropriate label. At the moment that script only pastes to one image at a time, but I guess that could be changed.

Finally, 'flatter' horizontal areas would be useful, to squeeze everything in (unless your image is scaled down). No doubt the areas on your master image could be manually flattened by changing their coordinates in ExifTool.

Lichen.png
Lichen.png (1.25 MiB) Viewed 610 times
fbungarz
Posts: 1774
Joined: 08 Dec 06 4:03
Location: Arizona, USA

Re: some advice needed how to manage an unusual image collection

Post by fbungarz »

Sorry for not responding earlier:
Thanks for all the advice. Very useful. I'll definitely need to think about this a bit more...
I guess, I still haven't quite figured out how to bet do this ;-)
fbungarz
Posts: 1774
Joined: 08 Dec 06 4:03
Location: Arizona, USA

Re: some advice needed how to manage an unusual image collection

Post by fbungarz »

I just realized that i have no clue how area tagging actually works.
My idea was to tag the spots on a plate using area tags.
I imported a huge vocabulary of secondary metabolite names (such as atranorin, or usnic acid). Via "right-click - add area" I can select a particular area in the image manually, but it seems I can only link these areas to a label in the category people???
At least, if I click on the little chain symbol to link the area tag with a label, only one choice pops up (from people). Is there a manual somewhere that better explains how area-tagging works?
Thanks,
Frank
G8DHE
Posts: 508
Joined: 21 Aug 17 12:58

Re: some advice needed how to manage an unusual image collection

Post by G8DHE »

Add the Area you want marked, then type in the Label name you want if its a new Label then select the "Create" option and that will pop up the window below to place the Label in any existing structure;
Create_Area.JPG
Create_Area.JPG (209.77 KiB) Viewed 449 times
However it does seem to have problems, like I can't add an additional area to an existing label, and typing the same Label name again creates a new Label so I ended up with several "Fan"s in Misc! However it did give me a choice of several location names to choose from under the Places category!
Geoff Mather (G8DHE)
Hert
Posts: 7412
Joined: 13 Sep 03 6:24

Re: some advice needed how to manage an unusual image collection

Post by Hert »

Tagging Areas with people is optimized.
1. Place the mouse over a face
2. Press E to a dd a new area
3. If a face is detected there (no online detection takes place) then the area for the face is created

The link icon for an Area offers suggestions from either the People, Objects, or Places category. If you want to link a label from another Category to an area then either drag the catalog label from the LAP to the area or drag a catalog label from the Category Tree to the area
This is a user-to-user forum. If you need product support then please send a message
Mke
Posts: 651
Joined: 15 Jun 14 14:39

Re: some advice needed how to manage an unusual image collection

Post by Mke »

Hert wrote: 04 Oct 22 10:00If you want to link a label from another Category to an area then either drag the catalog label from the LAP to the area or drag a catalog label from the Category Tree to the area
FWIW, this is generally what I do.
fbungarz
Posts: 1774
Joined: 08 Dec 06 4:03
Location: Arizona, USA

Re: some advice needed how to manage an unusual image collection

Post by fbungarz »

If you want to link a label from another Category to an area then either drag the catalog label from the LAP to the area or drag a catalog label from the Category Tree to the area
Agggrhh...
OK, I did not realize that area tagging was so restrictive. In my case, I am definitely not looking for face-tagging. There are no faces in these images.
And that the lookup of existing labels is restricted to People, Objects, and Places also explains why the automatic linking doesn't work. I imported a long list of lichen secondary metabolites to which these are matched and created a separate category called "Secondary Metabolites". Its a huge tree and drag-drop for matching the labels would be quite a pain.
I guess, I could try nesting these under "Objects", but I'd prefer to have these terms in a separate controlled vocabulary.

Well, it seems I am regularly trying to do things that PSU wasn't necessarily designed for...
:?

It would have been nice if area tagging worked with linking the tags to just any label in the catalog. I have been thinking of testing it for other purposes in my main catalog, but didn't try it out so far. There I am also using more categories than just the default ones...
fbungarz
Posts: 1774
Joined: 08 Dec 06 4:03
Location: Arizona, USA

Re: some advice needed how to manage an unusual image collection

Post by fbungarz »

I just realized there is a much easier way to do this.
I simply assign the labels first, then, with the image open, I drag an assigned label onto the image and a linked area tag is created.
Works like a charm, even for labels not included in any of the three categories.
HURRAY !!!
Mke
Posts: 651
Joined: 15 Jun 14 14:39

Re: some advice needed how to manage an unusual image collection

Post by Mke »

fbungarz wrote: 04 Oct 22 21:57 I simply assign the labels first, then, with the image open, I drag an assigned label onto the image and a linked area tag is created.
Yes! LAP = Label Assignment Panel:
Mke wrote: 04 Oct 22 15:38
Hert wrote: 04 Oct 22 10:00...drag the catalog label from the LAP to the area...
FWIW, this is generally what I do.
simato73
Posts: 56
Joined: 27 Nov 16 22:10

Re: some advice needed how to manage an unusual image collection

Post by simato73 »

Hi fbungarz,

This is quite exciting! Speaking more as a scientist than a photographer using PSu :wink:
I guess I don't have much to suggest towards the approach you suggest. Area tagging boggles my mind.
From a scientific point of view, I am quite curious as to why you still use TLC while you could use other techniques that are faster, more accurate and already digitalised, the obvious one would be HPLC/UPLC. Of course you would have to develop a suitable analytical method first, but without being an analytical chemist I'd say it should be far from impossible.

On the other hand, I realise you must already have a decent body of images. This I think wold be a nice image recognition ML project - at my workplace we host students from Mathematics or IT courses and they can do wonders.
Simone
fbungarz
Posts: 1774
Joined: 08 Dec 06 4:03
Location: Arizona, USA

Re: some advice needed how to manage an unusual image collection

Post by fbungarz »

Hi Simone,
From a scientific point of view, I am quite curious as to why you still use TLC while you could use other techniques that are faster, more accurate and already digitalised...
TLC is simply MUCH cheaper and therefore still very widely and routinely used by lichenologists. And I don't need a quantitative, just a qualitative method. When comparing different species absence/presence is far more important than the amount of a secondary metabolite being produced.
Cheers,
Frank
Post Reply