Adding TMED-2 Echocardiogram Dataset

[sudham123]

Name:

Riya Verma, Sudha Gattu

NetID:

rverm2, sudhamg2

Type

Dataset Contribution

Paper:

Detecting Heart Disease from Multi-View Ultrasound Images
via Supervised Attention Multiple Instance Learning

Citation to Original Paper:

@Article{huang2023detecting, title={Detecting Heart Disease from Multi-View Ultrasound Images via Supervised Attention Multiple Instance Learning}, author={Huang, Zhe and Wessler, Benjamin S and Hughes, Michael C}, journal={arXiv preprint arXiv:2306.00003}, year={2023} }

Link:

https://arxiv.org/pdf/2306.00003

Description:

We added the EchoBagDataset class to process echocardiogram datasets from TMED-2 for multiple instance learning tasks. Imported one table from a CSV file: echo_images, which contains study IDs and diagnosis labels. The class supports dynamically loading image files from a directory based on patient-study identifiers listed in the table. It also has functions to retrieve image bags and their corresponding labels for each patient-study instance.

Files to look at to test the implementation:

pyhealth/datasets/echo.py → new EchoBagDataset class

pyhealth/datasets/configs/echo.yaml → config file specifying table schema for the dataset

pyhealth/unittests/test_datasets/test_echo.py → unit tests for dataset loading and item retrieval