Methods to clasify and to segment echo data using deep learning

[mxochicale]

🚀 Feature

The pathway for the use deep learning for echo data seems to be a bit wide but with the experience of the team such paht can be narrowed down. That said, this issue can be used to track works that might be helpful for our work.

The following are few relevant repos shared by @huynhatd13. Thanks Nhat for sharing.

• https://github.com/raventan95/echo-plax-segmentation
• https://github.com/raventan95/echo-view-classifier
• https://bitbucket.org/rahuldeo/echocv/src/master/
• https://data.unityimaging.net/
• https://github.com/echonet/dynamic

[mxochicale]

"Is the application of SELF-SUPERVISED learning technique to echo data a good avenue to explore?"

• One potential direction of this work is the use of self supervised learning to which Hamideh has shared:

  • https://www.fast.ai/2020/01/13/self_supervised/ that leads to
  • https://lilianweng.github.io/lil-log/2019/11/10/self-supervised-learning.html
  • Just found this practical implementations: https://www.analyticsvidhya.com/blog/2020/09/a-detailed-study-of-self-supervised-contrastive-loss-and-supervised-contrastive-loss/

• On Friday 29th October, Hamideh share the following links that might also offer options for self-supervising techniques https://github.com/Sara-Ahmed/SiT https://colab.research.google.com/github/keras-team/keras-io/blob/master/examples/vision/ipynb/simsiam.ipynb#scrollTo=bX12ZAu4BTIU

• On Wed 15 Dec 14:51:21 GMT 2021, Hamideh mentioned that a simplest SSL algorithms is https://arxiv.org/pdf/2002.05709.pdf ; https://github.com/sthalles/SimCLR ; https://github.com/google-research/simclr

[mxochicale]

"Are we going in the direction of Manifold Learning?"

I am just aware of two potential view (4CB, 2CB) and others views might not be well captured in the echo-datasstes. In any case, just bumped into this tutorial that will be helpful to sketch our results: https://jakevdp.github.io/PythonDataScienceHandbook/05.10-manifold-learning.html

[mxochicale]

Spacio-temporal encoding

TASED-Net: https://arxiv.org/pdf/1908.05786.pdf; cites: https://scholar.google.com/scholar?cites=18144771163557392256&as_sdt=2005&sciodt=0,5&hl=en ; https://github.com/MichiganCOG/TASED-Net

Transformers

Literature

• "Automatic Segmentation of Head and Neck Tumor: How Powerful Transformers Are?" Proceedings of Machine Learning Research – Under Review:1–14, 2022 https://openreview.net/forum?id=reIO5WfgbLd

Spatio-Temporal Features transformers

• google-scholar search for Spatio-Temporal Transformer added Thu 27 Jan 11:08:26 GMT 2022

• ICCV2021 paper Learning Spatio-Temporal Transformer for Visual Tracking: https://github.com/researchmm/Stark added Thu 27 Jan 11:18:26 GMT 2022

• ViViT: A Video Vision Transformer: https://github.com/google-research/scenic/tree/main/scenic/projects/vivit added Thu 27 Jan 11:28:26 GMT 2022

[mxochicale]

Classification networks were addressed here: https://github.com/vital-ultrasound/echocardiography/tree/main/source/models and leave segmentations for future work!