Raphael Prevost

Head of Machine Learning / Senior Scientist @ImFusionGmbH. PhD in #MedicalImaging

[Raphael Prevost @raphael_prevost]

📢 @ImFusionGmbH continues to grow! 🚀
Our CT team is also hiring a full-time senior scientist. 🩻
If you want to work on advanced medical imaging problems, this is a great opportunity! 👇
https://t.co/j0q8WxnQTZ
@MICCAI_Society @MiccaiStudents @midl_conference

[Raphael Prevost @raphael_prevost]

📣We are hiring Software Engineers in multiple teams at @ImFusionGmbH : Machine Learning, Anatomy, Computer Vision, Interventional Imaging.

This is the best time to join the team, and help us drive innovation in medical imaging and surgery!
To apply ➡️https://t.co/Xx9PuCz4ZM

[Raphael Prevost @raphael_prevost]

@ImFusionGmbH x @nvidia Holoscan 🚀
https://t.co/b7B8eTiQUo

[Raphael Prevost @raphael_prevost]

The speed at which people repurpose and study these models is just ridiculous 🤯
Very interesting work, we'll definitely try this out!

[Bo Wang @BoWang87]

🚀 The Segment Anything Model (SAM) has been upgraded to SAM2, featuring an efficient image encoder for segmenting images and videos. But does SAM2 outperform SAM1 in medical image and video segmentation?

We're thrilled to present our paper "Segment Anything in Medical Images and Videos: Benchmark and Deployment"! We comprehensively benchmark SAM2 across 11 medical image modalities and videos.

📄 Paper: https://t.co/NSymKcOJ8q
💻 Code: https://t.co/9B7CG8J655

**Highlights:**
1. SAM2 doesn’t always outperform SAM1 in 2D medical images, but excels in video segmentation, making it more accurate and efficient for 3D images, such as CT and MR scans.

2. MedSAM still outperforms SAM2 on most 2D modalities, but SAM2 surpasses MedSAM for 3D image segmentation in a slice-by-slice approach.

3. Segmentation performance varies with model size; sometimes the smallest model outperforms larger ones.

4. Fine-tuning SAM2 significantly boosts its performance for medical image segmentation.

While SAM2 may struggle with challenging objects that have unclear boundaries or low contrast, it excels in generating good initial segmentation masks for common medical images and videos. However, the official interface doesn’t support medical data formats and has limitations on video length. To address this, we've developed a 3D Slicer Plugin and Gradio API for efficient 3D medical image and video segmentation. We invite you to try them out and provide feedback!

🔧 Deployment:
- 3D Slicer Plugin: https://t.co/j83JChav2r
- Gradio API: https://t.co/4zJUuPFR12

(Note: Due to GPU limitations, the online API is available for only 12 hours and may be slow. We highly recommend deploying the Gradio API with your own computing resources: https://t.co/q5UydWs6Xd

A big shoutout to Jun Ma (@JunMa_11) who recently joined our UHN AI hub (@UHNAIHUB) as Machine Learning Lead, and kudos to all co-authors: Sumin Kim, Feifei Li, Mohammed Baharoon (@BaharoonMS), Reza Asakereh, and Hongwei Lyu! This is true teamwork!

Looking forward to collaborating with the community to advance 3D medical image and video segmentation foundation models!

@UHN @UofTCompSci @UofT_LMP @UofT_TCAIREM @VectorInst

#MedTech #AIinHealthcare #DeepLearning #MedicalImaging #SAM2 #MedSAM #AIResearch

[Raphael Prevost @raphael_prevost]

📢 We have two openings for research students in my team at @ImFusionGmbH!
If you are interested in working on model training for various medical imaging applications, apply here: https://t.co/oauumeHDhg

[Raphael Prevost @raphael_prevost]

Last day of the @midl_conference
It's still time to see the demos at the @ImFusionGmbH booth!

[Raphael Prevost @raphael_prevost]

@ImFusionGmbH is happy to sponsor @midl_conference !
I am in Paris with several colleagues. Feel free to reach out at our booth for more infos on our research, software, job offers, etc.

[Mert Karaoglu @mertkaraoglu]

📢 Exciting news! We are organizing the STIR point tracking challenge this year at #MICCAI2024 @MICCAI_Society as a part of @endo_vis.

🔗https://t.co/SNOf1jivTX

[Raphael Prevost @raphael_prevost]

Create your custom applications using TotalSegmentator with the ImFusion SDK! 🙌
Github repo: https://t.co/6Y8KkprrHf

[ImFusion @ImFusionGmbH]

🛠️ Start building your own ImFusion plugin for specific anatomy👉https://t.co/7tE11nENmF

Learn to integrate Total Segmentator for automatic labeling and using Anatomy Module's data structures. Continue with further anatomy specific processing - all within ImFusion Suite.

[ImFusion @ImFusionGmbH]

In a world where regulatory requirements for Software as a Medical Device (SaMD) are complex and ever-evolving, staying ahead is crucial. That's why we release our new white paper, "Getting the Green Light - Regulatory Framework for SaMD".
🔗Download today https://t.co/oFRI5ngYNT

[Raphael Prevost @raphael_prevost]

@CEA_Officiel Félicitations! Avez-vous prévu de partager ces images à des fins de recherche?

[Raphael Prevost @raphael_prevost]

🧠First in-vivo brain image with a 11.7 Tesla MR scanner! 🤯
https://t.co/wBoUkEmeek

[Raphael Prevost @raphael_prevost]

If you have any idea related to medical imaging, we can make it happen!😉

[ImFusion @ImFusionGmbH]

🔍 Case study: Shoulder Instability Treatment Transformed by ImFusion's Expertise!

Shoulder instability diagnosis is complex. Dr. Moroder & team aimed to revolutionize care by maximizing diagnostic value from existing imaging techniques.

🧵1/2

[ImFusion @ImFusionGmbH]

Looking for a Senior IT System Engineer / DevOps Engineer

🤝 Join us if you want to work in an agile company with extensive feedback culture, collaborate on innovative projects, while enjoying complete flexibility and a comprehensive benefits package

https://t.co/AlTuqWIh7L

[Raphael Prevost @raphael_prevost]

📢Our team is still growing: another opportunity to join @ImFusionGmbH !
All our open job offers are available here: https://t.co/1VIAkK1tXW

[Raphael Prevost @raphael_prevost]

Advanced visualization is low-key one of the most important factors for understanding your data and therefore conducting efficient R&D!

[ImFusion @ImFusionGmbH]

Multi-modal visualization with ImFusion became even better: We improved our real-time global illumination volume renderer with support for on-the-fly deformation fields. See how we combine this feature with cut planes and masking for deformable US-CT fusion.

[Raphael Prevost @raphael_prevost]

@ja_schnabel @MICCAI_Society Looking forward to it!
Let us know when you have more info on the sponsorship packages 😉

[Raphael Prevost @raphael_prevost]

This is actually true for a number of very important papers, including the famous UNet!
Simplicity (what we focus on when choosing a baseline) turns out to regularly have merit.

[Raphael Prevost @raphael_prevost]

The lesion is not in the right hemisphere, but in the left one!🔀
The worst part is that it actually does mention that slices are typically "flipped"... but somehow does not follow its own warning!🤦
This is another proof that there is no logical reasoning behind current LLMs..

[Raphael Prevost @raphael_prevost]

⚠ Don't be fooled by the fluency of ChatGPT! ⚠

You might see conversations like this one which might look reasonable and impressive for a layman, but please don't use those models for medical images. 🙅

First those images are obviously not T2 scans, but it gets worse... 👇