EntCollNet / BugFlow

Bug flow is a community resource for entomological collection digitization workflows and modules.
Creative Commons Zero v1.0 Universal
22 stars 5 forks source link

Module 4B: Pinned 3D specimen imaging [1st Draft Completion List] #68

Open tmcelrath opened 3 years ago

tmcelrath commented 3 years ago

Module 4B: Pinned 3D specimen imaging

tmcelrath commented 3 years ago

@nlgunter @dryopoiddarling some great stuff here: http://biowikifarm.net/v-mfn/3d-handbook/3d_Imaging_Handbook:Main_Page

tmcelrath commented 3 years ago

See PPT here: https://github.com/EntCollNet/BugFlow/blob/master/modules/module_4/4Brecko%20J.pptx

tmcelrath commented 3 years ago

Split into two sections: photogrammetry and micro-CT scanning. We need a CT scanning person: ideas: Adam Brunke? Phil Barden?

tmcelrath commented 3 years ago

Ed Stanley at UF?

tmcelrath commented 3 years ago

Useful paper/example workflow here: https://doi.org/10.1002/jmor.20938 @nlgunter have you seen?

JCGiron commented 3 years ago

I contacted Thomas van de Kamp (in Twitter: @Thomas_vdKamp), who has several years of experience scanning insects. This is his response:

This is an amazing and very important project!

The procedures for CT scanning depend a lot from the X-ray source you want to use. I’ll give you a quick summary below, but take in mind that there are many different preparation protocols in order to visualize particular organs etc. For digitization of sample collections, however, a more general standard protocol is probably the best.

Laboratory X-ray source:

Scans at most laboratory setups take hours to days. You could scan both dried and alcohol-preserved specimens. However, the latter will provide very low soft tissue contrast using a laboratory setup and therefore require staining (e.g. with Iodine ore phosphoric tungsten acid). The anatomy of air dried insects is usually very badly preserved. Critical point or HMDS-dried insects provide much better results. Metallic needles are generally problematic and result in nasty artifacts. If possible, any needles should be removed before scanning. Because of the cone beam, you’ll be quite flexible in scanning insects in different sizes.

I suggest you check out this paper: https://onlinelibrary.wiley.com/doi/full/10.1002/cne.23741. It compares different protocols for insect CT-scanning using a laboratory source.

Synchrotron:

At KIT light source we routinely scan both dried and alcohol-preserved insects. Scan duration is around 30 seconds, therefore we can scan up to thousands of specimens in one week. Because of the additional phase contrast of a synchrotron, soft tissues can be visualized without any staining. Due to the superior preservation, I prefer alcohol-fixed samples over any form of dried specimens. Sometimes air bubbles occur that can destroy the scan – this happens more frequently in large insects and can be countered by adding filters into the beam. Needles are also a problem.

A synchrotron provides a parallel beam, therefore you are restricted to fixed magnifications and field of views (our beam size is 7 x 7 mm, but mostly I’m using 2 x 2 or 4 x 4 mm - most beamlines have even smaller FoVs). For automated high-throughput scanning – which in my opinion would be the best option for digitizing collections – it is paramount to sort samples in size groups for different magnifications before the actual experiment.

Synchrotron beamlines differ a lot from each other and every experimental station is unique. Therefore, the sample preparation should always reflect the parameters of the particular beamline you aim to use. Unfortunately, access to synchrotrons is limited and competition is high. However, at KIT we’re aiming to build an automated setup for sample digitization. We’re still in an early phase, but I hope we can establish a center for scientific collection digitization over the next years.

I hope I could help you a little bit. :)

Best wishes

Thomas

JCGiron commented 3 years ago

Regarding storage, this is what Thomas said:

File storage for 3D data is indeed an important issue and something which in my opinion is not properly solved yet. At KIT a standard scan is about 54 GB (raw data plus reconstruction) - and we easily scan 2000 samples in one week of experiments. At KIT with its strong focus on computer science we are connected to our own Large Scale Data Facility and we’re currently in the early phase of building a portal for collaborators and the scientific community to access the datasets. But I’m well aware that many institutions – and museums in particular – do simply not have the resources to store petabytes of data. Moreover, the proper curation of such data requires substantial IT knowledge, as everything has to be logged, backuped and archived etc.

Public databases like NFS-funded https://www.morphosource.org/ are a good solution, but I don’t know if they can handle the increasing demand of storing 3D data in the future.

JCGiron commented 3 years ago

Check out this open processing software: https://slicermorph.github.io/. They even have weekly meetings, just in case.