m1dsolo/echocardmae

How m1dsolo/echocardmae works

EchoCardMAE is a research workflow for echocardiography video representation learning, ejection-fraction (EF) prediction, and cardiac segmentation. Researchers prepare dataset-specific arrays, optionally pretrain a masked video autoencoder, then fine-tune or evaluate task models. The graph retains the separate data-preparation and task branches, plus the shared transformer internals. Some preprocessing and model files were not excerpted; their workflow placement is based on the README, and unsampled implementation details are omitted.

Main entry points: utils.py Array dataset utilities · pretrain.py MAE pretraining · train_seg.py Segmentation training · train_ef.py EF fine-tuning

Components 15

Data preparation

EF preprocessing echonet/avi2npy.py
Segmentation preprocessing echonet/avi2edes_npy.py
Array dataset utilities utils.py

Self-supervised pretraining

Masked video autoencoder model/echocardmae_vit.py
MAE pretraining pretrain.py
Contrastive loss loss/infonce_loss.py
Transformer building blocks model/utils.py

EF prediction

Video transformer model/vit.py
EF fine-tuning echonet/train_ef.py
EF evaluation echonet/val_ef.py

Segmentation

Segmentation transformer model/vit_seg.py
Segmentation training echonet/train_seg.py
Dice loss loss/dice_loss.py

Other components

Researcher
Echo datasets

Connections 24

Researcher → Echo datasets: obtains data README.md
Echo datasets → EF preprocessing: provides videos README.md
Echo datasets → Segmentation preprocessing: provides videos README.md
Researcher → EF preprocessing: runs preprocessing README.md
Researcher → Segmentation preprocessing: runs preprocessing README.md
EF preprocessing → Array dataset utilities: uses utilities echonet/avi2npy.py
Segmentation preprocessing → Array dataset utilities: uses utilities echonet/avi2edes_npy.py
Researcher → MAE pretraining: starts training README.md
MAE pretraining → Array dataset utilities: loads dataset pretrain.py
MAE pretraining → Masked video autoencoder: trains model pretrain.py
MAE pretraining → Contrastive loss: computes alignment pretrain.py
Masked video autoencoder → Transformer building blocks: uses blocks model/echocardmae_vit.py
Researcher → EF fine-tuning: starts fine-tuning README.md
EF fine-tuning → Array dataset utilities: loads dataset echonet/train_ef.py
EF fine-tuning → Video transformer: trains model echonet/train_ef.py
Video transformer → Transformer building blocks: uses blocks model/vit.py
Researcher → EF evaluation: runs evaluation README.md
EF evaluation → Array dataset utilities: loads dataset echonet/val_ef.py
EF evaluation → Video transformer: evaluates model echonet/val_ef.py
Researcher → Segmentation training: starts training README.md
Segmentation training → Array dataset utilities: loads dataset echonet/train_seg.py
Segmentation training → Segmentation transformer: trains model echonet/train_seg.py
Segmentation training → Dice loss: computes loss echonet/train_seg.py
Segmentation transformer → Transformer building blocks: uses blocks model/vit_seg.py

Drawn by GitDiagram on November 11, 2026 from the repository's file tree, README and source files. Read as Markdown ·