Thread by @AChoiHeart

Great session on Artificial Intelligence and Machine learning with session chairs @imagingmedsci, Dr. Bratt, @michael_t_lu, @ivanaisgum, Dr. Al'Aref and @DeCeccoCN at #SCCT2021!

#ICYMI - Here's a brief tweetorial from the #YesCCT session (1/7)
@mirvatalasnag @HeartOTXHeartMD

#AI & ML in #YesCCT: #SCCT2021 #Tweetorial (2/7)
1⃣: The big picture & black box by Dr. Michael Lu
-Multiple “black box” definitions
-Explainability
-Predictability
-More accurate than SOC, visualizes output (& modifiable), communicates uncertainty (eg grey-zone FFR-CT)

3⃣#AI & ML in #YesCCT #SCCT2021: New frontiers in atherosclerosis @AChoiHeart

-Use of #CNN for whole heart stenosis & atherosclerosis
-Improved reproducibility of stenosis to enhance guideline adherence
-Individualized risk prediction
-New Rx paradigm!

2⃣#AI & ML in #YesCCT: #SCCT2021 (2/7)
Image reconstruction & processing by
@ivanaisgum

-Improving image quality via advanced algorithms
-Noise reduction via generative adversarial networks
-Motion estimation & correction
-Image synthesis in #CAC

4⃣ Multiparametric risk prediction from big data by Dr. Al’Aref
-Proliferation of big data!
-Proliferation of imaging modalities
-Multiple challenges in high dimensionality, heterogeneity and irregularities
-Goal is multiparametric multimodal predictions

@mirvatalasnag

5⃣ #AI and #ML #SCCT2021 (6/7)
Enhancing Functional Evaluation: @carlodececco
-Cardiac functional analysis, FFR-CT & CTP
-AI-based FFR-CT may improve specificity in chest pain for obstructive CAD in @RadiologyCTI
-Clinical availability & results reproducibility-main challenges

6⃣ #AI and ML #SCCT2021: Avoiding hype and ensuring validity by Dr. Bratt (7/7)
-Look at the images in your dataset (Easy to trick the model)
-Be cognizant of limitations
-Have systems in place for human intervention

@mirvatalasnag