BoneJ
Scientific software for bone image analysis.
- 📢 Release alert! This one lets *all* Intel GPUs use BoneJ+ 🤖 If you have an Intel Iris Xe, UHD or Arc, support is more robust. Background: Intel omitted 64-bit floating point precision (double) in their GPUs since c. 2020: this update uses the universally-supported single-precision float type.
- 📢 New release! This one makes BoneJ's GPU-accelerated extensions easier to access from Python by using ImageJ2 / scijava-style wrappers. Just let the Fiji updater run and grab a voucher from bonej.org/shop
- OMG this is a good catch. We wrestled with this a long time ago when setting up Slice Geometry. The correct way to set slice spacing (pixel spacing in z) from a DICOM is to use the slice position info (0020,1041 or 0020,0032) and divide by (stack size - 1). It can be out by not much or quite a bit.
- Analysis of 985(!!) baboon skulls' CT scan data from MorphoSource shows that >18% have slice spacing errors that propagate to downstream analyses-- an important cautionary tale on checking CT metadata vs. errors, and a goliath effort: peerj.com/articles/201...
- A closer read of this paper makes it seem that some of the scans were done in chunks of a few slices each, and that the chunks had different slice spacing from each other. That's a bit of a mess: making a composite assuming equal z-spacing throughout the volume will lead to distortions.
- Could be fixed by determining the total number of slices and start and end slice position, then calculating average slice spacing and interpolating the original image data accordingly. A bit messy. Good on 3D Slicer for detecting, alerting users and sorting it!
- 👀
- Analysis of 985(!!) baboon skulls' CT scan data from MorphoSource shows that >18% have slice spacing errors that propagate to downstream analyses-- an important cautionary tale on checking CT metadata vs. errors, and a goliath effort: peerj.com/articles/201...
- Data Management Plan: * We will keep all data and literature in open browser tabs sychronized between the computers of all project members using the same Google Chrome profile. * By the end of the project we publish the username and password of that profile under a CC-BY 4.0 license.
- 🔥🔥🔥
- 📢 Release announcement! Local Thickness+ brings GPU-accelerated Tb.Th and Tb.Sp, for much faster measurements. Now out in BoneJ. Just let the updater run, and grab a token from bonej.org/shop to make it go. Drop us a PM to get a coupon for a free token.
- Large structures could make Tb.Sp in particular seem to run forever, or least longer than you would reasonably wait. 😴 Try it now. Big structures are still challenging, but should be done in like 1 hour instead of a week.
- The CPU implementation has a problem of thickness spheres overhanging the input mask by a pixel, introducing some inaccuracy. This GPU implementation has a more accurate Euclidean distance transform giving the sphere-fitting higher fidelity to the input geometry and less orthogonal grid bias.
- Small structures are also problematic as ever. But what's the thickness of a 1-pixel-wide line? The answer is much more meaningful for features represented by at least several pixels' thickness (what is several you may ask, well, try it and see...).
- Smaller simpler images now complete in a few seconds or less, depending on image size and feature size. Unconstrained 'outside' is still challenging, but is now handled by treating it as spheres seeded from the image borders with a radius equal to the distance to the nearest feature in the image.
- 🤔
- 🤩
- Signed up! Come along and get an intro on BoneJ's features.
- 🦴 Up next for BoneJ+: GPU-accelerated Local Thickness. The classic version is widely used for measuring trabecular thickness and separation (Tb.Th, Tb.Sp), but suffers from g.o.i.n.g v.e.r.y s.l.ow. on images with large or unbound features. Looking at a release in October: stay tuned! 📻
- So bad it's good Rrrr we love the Ccc
- 📢 GPU-accelerated Anisotropy now in BoneJ+ The fabric tensor expresses the scale and orientation of textures like trabecular bone and other porous materials. Runs on GPUs from Intel, NVIDIA and AMD, 10-100× faster than the classic version 🔥🔥🔥 Get a token from bonej.org/shop to try it out!
- Getting closer... Anisotropy+ working well in testing, even on antique Intel laptop GPUs, going at least 10× faster than the CPU version. Release coming soon!
- 🔥🔥🔥 Upcoming plugin for BoneJ+, Anisotropy+ running on Intel, NVIDIA and AMD GPUs at the same time! This one uses fields of line probes to determine the fabric tensor and calculate the degree of anisotropy, a measure of underlying texture orientation and scale. Looking good for release next week 🤞
- Next up for BoneJ+: GPU accelerated mean intercept length for measuring anisotropy via the fabric tensor, used to find directionality in trabecular bone 🦴 and other porous materials (soil, rock, food,..) 🫧🪨🍦 Kernel drafted, host code integration under way... 🧑💻 Stay tuned!
- 📢 Intel GPUs now supported by BoneJ+ 🤖 Today's new release lets BoneJ+ run on Intel GPUs, which are very common in laptops, in addition to NVIDIA and AMD GPUs. Even antique, 10-year-old ones are working. Ellipsoid Factor+ is also going 1.5-3× faster. Just let the ImageJ/Fiji updater run!
- Note though that GPU performance is very different among models. There's a ~10× difference between the fastest and slowest on the daily test system (Dell 7370 laptop with external AMD and NVIDIA cards), which chimes with crowdsourced benchmark data: www.videocardbenchmark.net/compare/3341...
- Today's Intel laptop GPUs (like the A770M or Arc Pro A30M) are 10-20× faster than the 10-year-old HD 515 in our testing machine. www.videocardbenchmark.net/compare/3341...
- Looking good for a release early this week (CW36) BoneJ+ will start to work on Intel GPUs and go faster on NVIDIA and AMD too.
- 😮 BoneJ+ running Ellipsoid Factor on NVIDIA, AMD and Intel GPUs simultaneously! Look out for this super-helpful update coming soon: BoneJ will start to use GPU acceleration on many (most?) laptops, which have integrated Intel graphics.
- 📢Release alert! BoneJ 7.1.4 is out now, bringing 💥100x faster💥 GPU-accelerated 3D Skeletonization, part of the new BoneJ+ extensions. You'll need a token to try it out, from bonej.org/shop Drop us a line bonej@bonej.org to get a coupon for a 100% discount on your first token.
- Just finished a new GPU-accelerated 3D skeletonization plugin for BoneJ+ that goes 🔥🔥50x faster🔥🔥 than the CPU version, even on an external gaming GPU. Final testing this afternoon, release next week. Drop us a line on bonej@bonej.org if you'd like a free coupon to try it out!
- Scratch that: More like 💥💥100x faster💥💥 !! 🤯 25 seconds vs. 45 minutes (782x811x549 image, on the laptop+thunderbolt GPU system in the pic above) Speed improvement is better on bigger images.
- [Not loaded yet]
- Second!
- 🤩
- Finally made it to the @mnhn.fr! The vertebrate skeleton is just so stunning in all its variety, this place showcases it like no other.
- Busy month coming up! Annual Meetings and workshops at: SBMS: 13 June, Interlaken docs.google.com/forms/d/e/1F... SwissBIAS: 18-19 June, Bern www.eventbrite.ch/e/swissbias-... BRS: 25-27 June, Edinburgh boneresearchsociety.org/meeting/brs2... See you there!
- 2009!
- 😍
- 🚨Important service release announcement A new BoneJ (7.0.20) is available that keeps up with changes in ImageJ (specifically, the move to scijava 40.0.0). Just let the updater run.
- [Not loaded yet]
- Should be plenty of time 😂
- [Not loaded yet]
- What's "Threads"?
- i don't want to hear your most boomer complaint. what's your most millennial complaint?
- I don't want a video explaining how to do the thing. I want step-by-step instructions in a plain text document, on a single page, ideally that I can see all of without scrolling while using a reasonably big font.
- Let me guess, you would like instructions for similar things concentrated together in a sensible order, one or two pages per thing, with a list of keywords and where they appear at the back and a guide to its organisation at the front. Sounds like a radical technology!
- 😂
- Getting closer to releasing first versions of BoneJ+, which brings GPU acceleration and improved precision. Ellipsoid Factor is first up, and nearly ready! Give us a shout if you'd like to be an alpha tester 🤓
- 👀
- Stay tuned for GPU-accelerated and more accurate Ellipsoid Factor in my upcoming Plus release! Ellipsoid Factor was a key metric to define the unusual morphology of the helmeted hornbill's casque trabecular bone, which is much more rod-like than expected.
- 🚨 New paper on helmeted hornbill trabecular bone with Hong Kong colleagues. The bone in the casque is remarkable - my eye was first caught by a specimen in the @nhm-london.bsky.social. Trabeculae are arranged unusually like a bundle of long, thick sticks melted together. doi.org/10.1111/nyas...
- [Not loaded yet]
- Time to enlist some elfin safety experts.
- I like mvn and JUnit because they keep me in one piece...
