A cochlea imaged in light-sheet microscopy (right) with staining for spiral ganglion neurons (red) and inner hair cells (blue). You will develop AI-based methods to analyze these structures, for example via segmentation of the individual cells (right) that will support gene therapy development for hearing loss and a better overall understanding of the anatomy of hearing.

Looking for a PhD position at the intersection of AI, imaging, and gene therapy? Apply for this position in my lab: tinyurl.com/2a2v6tvx

Part of sfb1690.uni-goettingen.de to study hearing, vision, and more.

Plus, you can create pretty pictures as the one below :).

Rendering of a full cochlea (left) with three stains (PV, VGlut3, CTBP2) shown in three different colors (red, blue, cyan). The whole cochlea is a spiral shaped structure, with spiral ganglion neurons (SGNs) in the inner helix, labeled by PV and inner hair cells (IHCs) in the outer helix, labeled by Vglur3. The figure also shows zoom ins. The right hand side shows segmentation results for SGNs, IHCs (represented by colored masks) and synapse detections (represented by colored dots).

Preprint alert! CochleaNet, our framework for analyzing light-sheet data of the cochlea. It consists of three networks to segment spiral ganglion neurons, inner hair cells, and to detect synapses. See rendering of a full cochlea in the image, find the preprint at doi.org/10.1101/2025....

Don’t miss Elena’s (from @ilastik-team.bsky.social lab) brilliant work @ #ICCV2025 with @anwaiarchit.bsky.social & @cppape.bsky.social poster 292 @ 11:15AM 🔬They tackle segmentation of massive 3D microscopy images & show how BatchRenorm removes tiling artifacts boosting transferability and clarity🌺

SynapseNet is a deep learning based software tool that automates the segmentation of vesicles, mitochondria, synaptic compartments, and the active zone. This is visualized in three panels. On the left, a section of an electron tomogram with a synaptic compartment densely filled with vesicles, which appear as round structures with dark boundary and light body in the image, is shwon. The top right panel shows the segmentation result of vesicles, visualized by masks with an individual color per vesicle and outlines for the segmented compartment (red) and active zone (blue). The bottom right panels shows a 3D rendering of the segmentation with vesicles shown as yellow spheres.

Are you studying synapses in electron microscopy? Tired of annotating vesicles? We have the tool for you! SynapseNet implements automatic segmentation and analysis of vesicles and other synaptic structures and has now been published:
www.molbiolcell.org/doi/full/10....

Large images have to be broken into tiles both for training and inference with neural networks. The tile predictions then need to be merged to produce the final volume prediction.

Segment large images without tiling artifacts: sharing our work that should have been presented at ICCV in 2 weeks - the brilliant first author Elena can’t go because of visa issues.
The paper: arxiv.org/abs/2503.19545 1/🧵

Data is the key to AI advances in biology and "Still, when it comes to data, nothing compares to the real thing." An important editorial in Nature methods with a nice little shout out to microSAM:
www.nature.com/articles/s41...

Anwai represented the lab at MIDL very well! Read his thread for details on our two latest papers on foundation models for microscopy, histopathology and medical imaging.

There are exciting opportunities and foreseeable clinical applications of vision foundation models for biomedical image analysis! The MIDL Community did seem excited, and so are we! (some amazing follow-up applications coming soon, keep an eye open 😉)

And both of our efforts and these beautiful presentations at #MIDL2025 have been successful due to immense efforts from our amazing lab members @caroteu.bsky.social and Titus Griebel. And of-course @cppape.bsky.social being the absolute best PI, as always (I can't say this enough tbh)!

And sticking to our @cppape.bsky.social lab's theme of rocking in the open-source world, all the models and code are completely open to use and easy to access. You can take @midl-conference.bsky.social's word on this! 😉

PathoSAM: github.com/computationa...

Late PEFT: github.com/computationa...

Next, we introduce a novel finetuning paradigm for foundation models "Late PEFT", inspired by our striking observations of PEFT's memory requirements, questioning: "Are PEFT methods Resource Efficient?"

We have an answer: PEFT "can be resource efficient"! Check out more at: doi.org/10.48550/arX...

To begin with, here's "PathoSAM", our foundation model for nucleus segmentation in histopathology, presented as an Oral! (and my very first one).

If you want a SOTA model for interactive and automatic segmentation for your histopathology images, go check out the paper now doi.org/10.48550/arX...

We presented our latest work on "PathoSAM" and "Late PEFT" last week at #MIDL2025 (Salt Lake City)!

The community is growing and MIDL is becoming the venue-to-go for high quality research discussion!🧵

Are you looking for an exciting position at the intersection of super-resolution microscopy and AI? Then check out the PhD and PostDoc position we offer for a joint project with the Group of Stephan Hell at MPI Göttingen. Please share with anyone interested, read on for links and details.

We released version 1.6 of micro_sam:
- Improvements for automatic tracking.
- A new experimental mode for object classification.
- **New versions of the LM and EM models**
The models fix artifacts in automatic segmentation, see old vs. new prediction and better 3D segmentation results due to it.

Spotiflow: accurate and efficient spot detection for fluorescence microscopy with deep stereographic flow regression Nature Methods - Spotiflow uses deep learning for subpixel-accurate spot detection in diverse 2D and 3D images. The improved accuracy offered by Spotiflow enables improved biological insights in...

Spotiflow, our deep learning based spot detection method for microscopy, is now published in @natmethods.nature.com!
Since the pre-print, we have added many features, notably native 3D detection!
@maweigert.bsky.social @gioelelamanno.bsky.social @epfl-brainmind.bsky.social
Paper: rdcu.be/epIB7
(1/N)

Congratulations @maweigert.bsky.social and @albertdm.bsky.social 🥳

Spotiflow: accurate and efficient spot detection for fluorescence microscopy with deep stereographic flow regression - Nature Methods Spotiflow uses deep learning for subpixel-accurate spot detection in diverse 2D and 3D images. The improved accuracy offered by Spotiflow enables improved biological insights in both iST and live imag...

A nice advance for imaging-based spatially resolved transcriptomics from the Weigert and La Manno labs. Spotiflow uses deep learning for subpixel-accurate spot detection in diverse 2D and 3D images. www.nature.com/articles/s41...

Announcing the new release v1.4.0 of microSAM. The main changes are:
1. Simplified installation on windows.
2. Preliminary support for automatic tracking.
3. Improved interface for model selection.
Read on for a quick summary of the changes.

@anwaiarchit.bsky.social & @cppape.bsky.social demonstrating the power of micro-sam, the napari plugin for the microscopy segment anything model, in their awesome workshop at the #TiM2025 conference in Münsingen. 🔬🦠💻

Our March issue is now live! 🥳
nature.com/nmeth/volume...

The cover represents the process of cell and organelle segmentation by Segment Anything for Microscopy.
Paper here: nature.com/articles/s41...

Cover by Sebastian von Haaren.

Another feather for Segment Anything for Microscopy. We made it to the cover for @naturemethods.bsky.social!

And all thanks to our amazing @haarensv.bsky.social for this! <3

Hi @psobolewskiphd.bsky.social,
It's under the same name in our GUI model list. With the latest version installed, this will be stored in cache automatically!

Pflanzenzellen, die mit einem Fluoreszenzmikroskop aufgenommen und mit dem Modell automatisch segmentiert wurden. Die zugrunde liegenden Daten sind dreidimensional und das Bild zeigt eine Darstellung der segmentierten Zellen, die jeweils durch eine andere Farbe repräsentiert werden. Foto: Nature Methods: 10.1038/s41592-024-02580-4

Segmentierung von Zellen in der Lichtmikroskopie mit μSAM. Das Bild zeigt, wie Zellen in der Phasenkontrastmikroskopie mit μSAM segmentiert werden können. Grüne Punkte und Kästchen zeigen die Benutzereingabe und farbige Masken die entsprechende Vorhersage des Modells. Foto: Erstellt von Anwai Archit mit dem μSAM-Tool, verfügbar in Nature Methods: 10.1038/s41592-024-02580-4

Automatische Zellanalyse mit #KI: Forschende trainierten eine bestehende, KI-basierte Software neu. Das Modell „Segment Anything for Microscopy“ kann Bilder von Geweben, Zellen und anderen Strukturen genau segmentieren: s.gwdg.de/HqkMz2; s.gwdg.de/iTejKW

Forschungsteam mit bsky.app/profile/cppa...

It was soooo much fun to brainstorm solutions with everyone, together!❤️

#EMBLDeepLearning

Thank you @unigoettingen.bsky.social for the feature!😍

μsam got some super cool feature updates last week. Don't wait for the next release, go check us out now!

github.com/computationa...

Because we have seen these improvements and due to popular demand, cc @ritastrack.bsky.social @jianxuchen.bsky.social, we have decided to start a call for community data submission to further improve our models: computational-cell-analytics.github.io/micro-sam/mi... . Looking forward to any feedback

Our next micro_sam release is here! We have a new model for light microscopy, that massively improves for automatic segmentation! See the qualitative and quantitative comparison in the images, v2 is our previous version, v3 is the new one.

Help improve MicroSAM!

Look at all of those #OMEZarrs! 🤩