Remarkable demonstration of progress in long term dependencies.

"Human radiologists are already much worse than computer radiologists." I have a lot of respect for Sam Altman but he really hasn't done his homework on this one. cnbc.com/2019/02/26/sam…

Why radiologists won’t be replaced by #AI, deep learning American College of Radiology #imaging #radiology #JACR ow.ly/WJUo30ovaH0

Why radiologists won’t be replaced by deep learning @eranrad María J. Díaz Candamio Dania Daye, MD PhD Charles Kahn, MD Curt Langlotz Dr. Emil Lee @DrGMcGinty CancerGeek Just another south FL dude @FalgunChokshiMD Seth M Hardy, MD MBA Tirath Patel, MD Judy Gichoya @DrSoGu Andy K. Moriarity Syed Furqan Zaidi, MD, MBA, FACR @HawkImaging radiologybusiness.com/topics/artific…

"Deep learning won’t be replacing radiologists anytime soon." dooz.us/2PqD9wX Radiology Business

“Some patients also exhibited a ‘crazy-paving pattern,’” ⁦Business Insider⁩ putting chest radiology on the map woop woop! ⁦Society of Thoracic Radiology⁩ businessinsider.com/china-coronavi…

Congratulations to 2020 Best Junior Faculty Scientific Presentation Awardee Alex Bratt for "Left Atrial Volume as a Biomarker of Atrial Fibrillation on Routine Chest CT: A Deep Learning Approach"! #thoracicrad2020

Really heart wrenching washingtonpost.com/health/2020/03…

🇺🇸

CAD-RADST2.0 2022 Coronary Artery Disease – Reporting and Data System An Expert Consensus Document of the Society of Cardiovascular Computed Tomography (SCCT), ACC, ACR and NASCI journalofcardiovascularct.com/article/S1934-… Heart_SCCT

Insisting that authors include a LIME/SHAP/IG/TCAV/GRADCAM saliency map in a paper shd be a disqualifying offense for reviewers. Including any such map without a powerful disclaimer (“this means nothing”) shd be a disqualifying offense for authors Extra true for ML in healthcare

Hey Nuance Communications plz use this in powerscribe

Most fascinating paper I’ve seen in a while

Probably a watershed moment for diagnostic radiology. From a data structure/size standpoint, there is very little difference between a video and a CT/MRI. Buckle up

Would be game changing for #MachineLearning #Radiology research if this works for HIPAA x.com/newsycombinato…

Really nice article by Kevin Seals, MD on Butterfly Network 's hot new ultrasound tech. Are piezoelectric crystals a thing of the past? x.com/kevinsealsmd/s…

Excellent talk by Nathan Silberman from Butterfly Network on using GANs for domain adaptation. Largely solves the problem of heterogeneous data sources (i.e. different hospitals, vendors).

Was a lot of fun discussing the future of #AI in radiology with pioneering SIIM/Mass General Brigham Data Science Office (DSO) leader Katherine Andriole, Tarik Alkasab from Mass General Imaging, and Alex Bratt from @WeillCornellRad. Summarized by AuntMinnie.com here: auntminnie.com/index.aspx?sec…

Turns out it's possible to recreate training data from a NN using only black box api access--no need for params. Upshot for medical researchers and vendors is that if you train on unanonymized patient records, your model is PHI. arxiv.org/abs/1802.08232

Group develops machine learning model to predict future behavior of a chaotic system. Wonder how well it would do modeling other types of chaos (e.g financial markets, human behavior, physiology/pathology). Current DL seems poorly suited for these problems quantamagazine.org/machine-learni…