Elham Mahmoudi, MD
@EliMahMD
Followers
952
Following
2K
Media
30
Statuses
383
Postdoc Fellow @MayoClinic | 💻 🔗🫀Developing #Deeplearning models on #CardiacImaging @MayoAIlab |🪴Plant Mom🌱 Enjoy 🎨🖌️📷 and anything science!
Joined December 2021
It is such an honor for me to take part in providing this highly practical and easy-to-learn educational material, @MayoAILab. If you seek to develop DL models on medical imaging, you can find a comprehensive guide, from basic concepts to walkthrough projects, in the MIDeL.
7
8
113
Honored to announce our latest publication: Object-oriented aortic root localization in pre-TAVR CT scan. Thanks to my mentors, Dr Bradley Erickson @Slowvak , Dr Vinayak Nagaraja and Dr Paul Friedman for their invaluable insights and supervision.
🚨New original research presents a fully automated robust pipeline for #aortic root detection in pre-#TAVR #CTA. ✅ F1: 99.0%, mAP50: 99.5%.✅ Tilt prediction error: 7.9° vs. 3.3° (experts). ➡️ @EliMahMD @adnanalkhouli @BrdKhsrv @ShanliVahdati
4
4
28
RT @Ashish_Kumar_MD: Very Long-Term Outcomes of Cardiac Resynchronization Therapy in Patients With Ischemic and Nonischemic Cardiomyopathy….
0
11
0
RT @woojinrad: 🤔 Your chest X-rays have invisible hospital fingerprints. AI can detect what we can't, but could that be misleading?. 🔬 New….
0
6
0
RT @MayoAILab: Big thanks to our lab alum Kenneth Phillbrick for hosting us at Google Rochester!🌟We got a behind-the-scenes look at Google….
0
1
0
RT @Bonakdar_Milad: Excited to share our latest publication!. 🧵 1/4: Struggling with BPH symptoms, especially with a significant median lob….
0
3
0
RT @Thom_Wolf: we've seen nothing yet! hosted a 9-13 yo vibe-coding event w. @robertkeus this w-e (h/t @antonosika @LovableBuild). takeaway….
0
560
0
✨Can’t wait to read these two amazing studies!.
Honored to receive Honorable Mention at #SIIM25 for our work on de-identifying medical reports on edge devices using SLMs 🏆. Congrats to @theodapamede for winning Best Paper with his CV risk prediction from mammograms! 👏. Grateful to my mentors @Slowvak and @judywawira 🙏🏻
0
0
20
2. We reached to a near-optimal accuracy with few steps of optimization and local inference of open-source models as small as 70B. 💡 Interestingly, When the model was very small (Phi3), chain of thought was misleading the model by looking at irrelevant information like the
1
0
4
📝Here I review the highlights of this work: . 1. Since we were looking at VHD severity (absent to severe), we used Mean Squared Error (MSE) 📏to show how far are the predictions to the actual values. In this visualization you can see how smaller models output wrong and
1
0
6
RT @ChiehJuChao1: Read this excellent work that demonstrates proper LLM prompting techniques can achieve high accuracy in identifying clini….
0
2
0
🚀Our latest paper is now published in the Journal of the American Medical Informatics Association (#JAMIA)!.🧠We shared complete prompts and detailed optimization samples!.🔍 Don't miss the supplementary materials for this!.
academic.oup.com
AbstractBackground. Automated data extraction from echocardiography reports could facilitate large-scale registry creation and clinical surveillance of val
1
9
35
🚀In this hands-on notebook, you combine Focal Loss function with Cross Entropy + Dice Loss to improve minor objects segmentation and classification.
🚨Hands-on alert: a new MIDeL chapter is out!. Learn how Focal Loss boosts medical image segmentation - especially for small, hard-to-spot structures. ✍️By @SahikaAi. 🔗 #MedicalImaging #AI4M #DeepLearning #FocalLoss.
0
0
16
RT @AndrewFoongYK: 🚨We're hiring! The AI and Data Analytics (AIDA) team at Mayo Clinic Radiation Oncology is recruiting Research Fellows to….
0
5
0