abhe

ML+Imaging scientist, NIH-F32 Postdoc Fellow @UCSFimaging

[UCSF Center for Intelligent Imaging @UCSF_Ci2]

Get to know Phil Xie, a junior at @UCBerkeley studying electrical engineering, computer science & bioengineering. Phil is a @UCSF_ci2 summer fellow & is working with faculty mentors Dr. Abhejit Rajagopal & Dr. Peder Larson (@pezlarson). https://t.co/M1MzJ2ORjv

[Efrat Shimron @Efrat_Shimron]

Excited to moderate the ML/AI Emerging Algorithms session today! We have a wonderful line of speakers, with many talks on bleeding edge methods - Swin transformers, federated learning, contrastive learning, diffusion models & more. Join us in room 714A/B at 13:45 (Tuesday)

[abhe @absudabsu]

Think this sounds cool?

Come hear us talk about it at 3pm today (Monday) at ISMRM! — session on AI/ML Emerging Algorithms and Methods.

@UCSFimaging @UCSF_Ci2 @RadiologyUcla @duke_dair

We’re looking for more sites and collaborators to join our consortium, so come say hi 🙂👩🏻‍🔬👨🏽‍💻

[Thomas Hope @thomashopemd]

2) Federated learning allows you to pool data across sites without the need to actually transfer the original data (in this case DICOM images). This approach may help enable future academic collaborations...

[Thomas Hope @thomashopemd]

1) Have you ever run into issues sharing data across institutions for training neural networks, issues navigating PHI? Federated learning will change all of that. Check out this work from @absudabsu from @UCSFimaging. Check it out at: https://t.co/AZpajQC3gH

[abhe @absudabsu]

Excited to share results from our multi-disciplinary cross-institutional team! This is just the start… get ready for big things as we open source our toolkit w/ examples in the next couple weeks 👨🏽‍💻🧬👩‍🔬🩻🧪@UCSF_Ci2 @UCSF_BCHSI @UCJointCPH @PCFnews @thomashopemd @felixfengmd

[Radiology: Artificial Intelligence @Radiology_AI]

T - 1 Day: Don't miss the March #RadAIchat this Wed. Mar. 1st at 8 PM ET on "Synthetic Data in Medical Imaging" moderated by @khosravi_bardia and featuring @giemmecci @Dr_ASChaudhari @pezlarson #MachineLearning #DeepLearning #radiomics @cekahn @HElhalawaniMD @RSNA

[UCSF Bakar Computational Health Sciences Institute @UCSF_BCHSI]

Tomorrow, 2/28! Bakar Lab presentations in-person/Zoom, 1pm. Predictive Modeling Leveraging EHR IDs Early Predictors of Alzheimer’s Disease @AliceTang_mstp
@SirotaLab A spectral graph modeling of brain oscillations @parulv1
@RajLab_UCSF Rm WGVCV 208 or https://t.co/Buior3xCMF

[UCSF Center for Intelligent Imaging @UCSF_Ci2]

Applications for our summer & 6-month programs are open until February 28. We encourage high school, undergrad & graduate students who are interested in AI, machine learning, imaging, medicine &/or academic research to apply. Learn more! https://t.co/jjrPSGaZQW

[Aniket Tolpadi @atolpadi7]

See our paper on the K2S challenge we hosted at MICCAI 2022! We curated a 300-patient dataset of raw k-space and tissue segmentations, tasking challenge participants with leveraging it to segment bone and cartilage from 8X accelerated knee MR acquisitions.
https://t.co/9VoGzj4bbc

[abhe @absudabsu]

@deliprao https://t.co/5nqPlnbcoc
Here we compute confidence scores via Gaussian Mixture Models fit to the feature space of a CNN (see appendix for non-Unets). We averaged scores over a full train/test dataset for unsupervised model ranking, but we also have it working for single inputs

[UCSF Center for Intelligent Imaging @UCSF_Ci2]

Our summer fellowship is perfect for high school & undergraduate students who have exposure to applied computer science, are interested in careers in academic research &/or plan to pursue graduate school. Learn more here & RT this post! ➡️ https://t.co/jjrPSGbxGu

[UCSF Center for Intelligent Imaging @UCSF_Ci2]

We're preparing an exciting line-up of lectures, debates, workshops, symposiums & other events to celebrate the third anniversary of @UCSF_ci2. Read more about the significance of this "birthday" here! https://t.co/7h7wYzr0Fd

[abhe @absudabsu]

@ethanepperly @Tim_Roughgarden Quite literally, if you interpret it as a deep rational polynomial network !

[UCSF Imaging @UCSFimaging]

During #NPAW2022, we decided to shine a spotlight on Dr. Abhejit Rajagopal (@absudabsu), an expert in designing robust #MachineLearning algorithms for reconstruction & classification. Get to know him & his work ➡️ https://t.co/g5J6jgHcZ9 #UCSFPostdocs

[abhe @absudabsu]

@ylecun It’s not all self-supervised…
- supervised learning from caretakers
- hot restarts, e.g. when they fall
- data augmentation, e.g. being exposed to new environments and concepts

Mixed supervision makes it better!

[abhe @absudabsu]

@MountainOfMoon

You might consider following our now-local hero at Caltech, Dr. @klbouman !

[abhe @absudabsu]

As with all physics-based computational imaging..

making the image == finding a solution
to an inverse problem.

✨->🌌->📸🔭->🌠

When the problem is constrained enough, it provides evidence that the physics was correctly modeled!

That’s how we make pictures
like 🖼🎇🩻

[Arya Mazumdar @MountainOfMoon]

Why are the blurry images of black holes interesting? Are humans supposed to stare at black holes? What scientific insight this visualization gives? Does any theory get validated by them? What physics do we know now that we did not before? Any good articles about this?

[abhe @absudabsu]

Great to see Dr. Mehmet Kurt of @KurtLabResearch present at @UniofOxford on his research and vision for quantifying brain damage, improving diagnosis, and studying development of the human brain with neuromechanics and multimodal imaging! Fantastic talk Mehmet, exciting vision!🚀

[abhe @absudabsu]

Great way to close out ISMRM with two amazing talks from Dr. Jeremy Gordon and Dr. @MichaelOhliger on HP C-13, rapid acquisition techniques, and where we’re going next! Very insightful for other modalities as well!
@UCSFimaging @UCSF

[Mehmet Kurt @KurtLabResearch]

🚨 Kurtlab @UW is hiring 🚨 We are looking to hire 1-2 postdoctoral scholars in the fields of brain biomechanics and neuroimaging . Please contact me for more details. If you are attending #ISMRM2022, happy to chat in person.