Lightnews — Scholar-powered news

Reposted by Alex Kalinin

Niranj Chandrasekaran @niranjchandrasekaran.com · Aug 7

Morph Map is now published in Nature Methods. Excited to see what the community discovers with this resource mapping ~15,000 human genes!

rdcu.be/ezGre

1 19 51

Reposted by Alex Kalinin

Eric Topol @erictopol.bsky.social · 28d

Today @nature.com
www.nature.com/articles/d41...

3 65 210

Alex Kalinin @alxndrkalinin.bsky.social · Jul 12

In applications, there are going to be some fun CV presentations in the GenBio workshop - come check it out!

Disclaimer: I have one of those and would love deeper critique from a CV standpoint:
bsky.app/profile/alxn...

Alex Kalinin @alxndrkalinin.bsky.social · Jul 10

🔬🤖 Introducing Spotlight: virtual staining (VS) improved by focusing on cells

VS models often learn to predict both cells and noisy background, because training treats all pixels equally. We address this by explicitly training models to prioritize foreground.

1/5

Figure 1:A schematic overview of Spotlight. a: Typical virtual staining models are trained using a pixel-wise loss such as mean squared error (MSE) on the whole image. b: Spotlight uses foreground estimation obtained by histogram thresholding to restrict pixel-wise loss to foreground areas and also employs soft-thresholding of the prediction to compute segmentation loss.

1 2

Alex Kalinin @alxndrkalinin.bsky.social · Jul 11

Workshop webpage: genbio-workshop.github.io/2025/
#ICML2025 #ICML #GenAI #GenBio #BioML #BioimageAnalysis

GenBio ICML Workshop 2025

GenBio focuses on solving fundamental problems in biology through generative AI.

genbio-workshop.github.io

Alex Kalinin @alxndrkalinin.bsky.social · Jul 10

📄 Dive into the details in our preprint → arxiv.org/abs/2507.05383

I’ll be presenting this work at the GenBio workshop at ICML on Friday, July 18 — come say hi and chat about virtual staining!

Big cheers to our collaborators at @imbavienna.bsky.social & @umich.edu Medical School

5/5

Foreground-aware Virtual Staining for Accurate 3D Cell Morphological Profiling

Microscopy enables direct observation of cellular morphology in 3D, with transmitted-light methods offering low-cost, minimally invasive imaging and fluorescence microscopy providing specificity and c...

arxiv.org

1 1

Alex Kalinin @alxndrkalinin.bsky.social · Jul 10

The visual difference is clear: compared to a baseline (F-net), Spotlight sharply reduces artifacts, resulting in clearer nuclear boundaries and less segmentation artifacts, while preserving foreground textures.

4/5

1

Alex Kalinin @alxndrkalinin.bsky.social · Jul 10

💡Spotlight uses the fact that even simple histogram thresholding (e.g., Otsu) is often sufficient to approximate informative FG regions. We use this to (1) mask MSE loss to focus learning on FG intensities, and (2) add a FG/BG segmentation loss to preserve cell morphology.

3/5

1

Alex Kalinin @alxndrkalinin.bsky.social · Jul 10

Most VS models are trained with pixel-wise losses like MSE, treating background (BG) and foreground (FG) equally. Unlike natural images, BG in cell imaging isn't informative–so models learn to reproduce noise. E.g., in 3D, predictions show axial blur and elongation.

2/5

1

Alex Kalinin @alxndrkalinin.bsky.social · Jul 10

🔬🤖 Introducing Spotlight: virtual staining (VS) improved by focusing on cells

VS models often learn to predict both cells and noisy background, because training treats all pixels equally. We address this by explicitly training models to prioritize foreground.

1/5

2 2 1

Alex Kalinin @alxndrkalinin.bsky.social · Jul 9

Looks neat! How does the number/variety of features compare to Cellprofiler? Does it have Python bindings?

1

Alex Kalinin @alxndrkalinin.bsky.social · Jul 8

Big shout out to the whole team: Alán F. Muñoz, Tim Treis, @shatavishadg.bsky.social , @fabiantheis.bsky.social ntheis.bsky.social, @drannecarpenter.bsky.social, @shantanu-singh.cc

6/6

Bluesky

ntheis.bsky.social

1 1 3

Alex Kalinin @alxndrkalinin.bsky.social · Jul 8

We will be discussing this work at #ICML2025 CODEML workshop - come check it out and chat with us!

📖 Arxiv → arxiv.org/abs/2507.01163
💻 Github → github.com/afermg/cp_me...

5/6

cp_measure: API-first feature extraction for image-based profiling workflows

Biological image analysis has traditionally focused on measuring specific visual properties of interest for cells or other entities. A complementary paradigm gaining increasing traction is image-based...

arxiv.org

2 4

Alex Kalinin @alxndrkalinin.bsky.social · Jul 8

Key benefits:
- Reproducibility: replaces GUI workflows with code
- General: agnostic to data types (3D images, spatial transcriptomics)
- Few dependencies: easy to integrate into existing image analysis frameworks
- Backwards-support: largely matches CellProfiler features

4/6

1

Alex Kalinin @alxndrkalinin.bsky.social · Jul 8

With that in mind, we developed cp_measure, a Python library that extracts morphological features from segmented images from within your pipeline, bridging the gap between the BioAI/ML community and the existing GUI-based tool that populates bioimaging workflows.

3/6

1 2

Alex Kalinin @alxndrkalinin.bsky.social · Jul 8

We felt there were a limited number of programmatic tools for featurizing segmented cell images, and CellProfiler is the de-facto standard for interpretable features.

2/6

1

Alex Kalinin @alxndrkalinin.bsky.social · Jul 8

🔬API-first feature extraction for image-based profiling workflows

If you need to obtain interpretable features from your segmented microscopy images, but want to do it in a fully automated way, we know the struggle.

1/6

1 20 46

Reposted by Alex Kalinin

Anne Carpenter @drannecarpenter.bsky.social · Jul 2

If you’re interested in single cell data analysis, come give Image based profiles a try! Huge dataset being made available for exploration at this hackathon (+ symposium):

Berlin, November cytodata25.eu-openscreen.eu

1 7

Reposted by Alex Kalinin

AI4Life @ai4life.bsky.social · Jun 27

The 2nd AI4Life Challenge is live!
Calling the AI & bioimaging community to tackle a key microscopy challenge: removing noise while preserving detail.

📦 Paired noisy/clean datasets
📈 Ground-truth evaluation
🧠 DL focus

Build, test, compete 👉 ai4life.eurobioimaging.eu/challenge-2/

1 15 27

Alex Kalinin @alxndrkalinin.bsky.social · Jun 23

There is still time to submit an abstract to CytoData 2025!

cytodata.bsky.social @cytodata.bsky.social · Jun 20

🌟 𝗥𝗲𝗺𝗶𝗻𝗱𝗲𝗿 🌟
The deadline for abstract submissions for oral presentations at Cytodata 2025 in Berlin is approaching!

👉 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗮𝗯𝘀𝘁𝗿𝗮𝗰𝘁 𝗯𝘆 𝗝𝘂𝗻𝗲 25!
cytodata25.eu-openscreen.eu/registration/

#BerlinConference #Imageanalysis #Microscopy

1

Reposted by Alex Kalinin

Alexander Krull @alex-krull.bsky.social · Jun 10

The submission for 2025's BioImage Computing @iccv.bsky.social is now open. If things go like the last years, I am looking forward to all you brilliant submissions.
www.bioimagecomputing.com

BioImage Computing

a truly interdisciplinary workshop

www.bioimagecomputing.com

2 5

Alex Kalinin @alxndrkalinin.bsky.social · Jun 10

Huge thanks to the whole team for making this work possible! 👏 @johnarevalo.bsky.social, Erik Serrano, @lvulliard.bsky.social,‬ Hillary Tsang, Michael Bornholdt, Alán Muñoz, @sugansivaguru.bsky.social, Bartek Rajwa, @drannecarpenter.bsky.social, Gregory P. Way, @shantanu-singh.cc
7/7

1

Alex Kalinin @alxndrkalinin.bsky.social · Jun 10

mAP performs well across diverse data types (imaging, proteomics, transcriptomics), perturbation types (gene edits, small molecules), and resolutions (bulk, single-cell). Our fast, open-source Python package copairs makes application easy and scalable:
github.com/cytomining/c...
6/7

GitHub - cytomining/copairs: Find profile pairs and compute metrics between them.

Find profile pairs and compute metrics between them. - cytomining/copairs

github.com

1

Alex Kalinin @alxndrkalinin.bsky.social · Jun 10

Our method is non-parametric and robust: similarities are converted into ranks, with statistical significance determined by permutation testing. No need for assumptions about distribution, linearity, or sample size—just choose a sensible distance measure for your data! 🚀
5/7

1

Alex Kalinin @alxndrkalinin.bsky.social · Jun 10

Similarly, mAP can quantify phenotypic consistency *within* groups—like compounds with the same mechanism or genes with similar function—by checking how highly related perturbations within the group rank when retrieved against others.
4/7

1

Alex Kalinin @alxndrkalinin.bsky.social · Jun 10

We approach this as an information retrieval problem: replicates of a phenotypically active perturbation should rank higher in similarity when retrieved against control replicates. We quantify this with a single, data-driven metric—mean average precision (mAP).
3/7

Schematic overview of the mAP framework for profiling data analysis. (A) Shows multiple replicate profiles per perturbation and controls. (B) Selects one replicate as a query and measures distances to other replicates and controls. (C) Profiles ranked by decreasing similarity, converted into binary labels for precision and recall calculation at each rank. (D) Average precision calculated by averaging precision at ranks containing perturbation replicates, shown graphically as an area under a precision-recall curve. (E) Average precision scores for all replicates averaged into a single mean average precision (mAP) score per perturbation. (F) The approach is also extended to retrieve groups of perturbations sharing biological annotations, such as mechanism of action (MoA).

1