Datasets:
Fix the Dataset Viewer with full browse tables
#4
by tsmithv11 - opened
- README.md +9 -3
- viewer/default/test.parquet +3 -0
- viewer/default/train.parquet +3 -0
- viewer/default/validation.parquet +3 -0
README.md
CHANGED
|
@@ -17,17 +17,23 @@ configs:
|
|
| 17 |
- config_name: default
|
| 18 |
data_files:
|
| 19 |
- split: train
|
| 20 |
-
path: "
|
| 21 |
- split: validation
|
| 22 |
-
path: "
|
| 23 |
- split: test
|
| 24 |
-
path: "
|
| 25 |
---
|
| 26 |
|
| 27 |
# PTB-XL v1.0.3 in Lance
|
| 28 |
|
| 29 |
Search the full PTB-XL corpus for ECGs with particular annotated findings, then retrieve the matching 12-lead waveforms. Each result represents one 10-second ECG. Its row contains the original report, the joined SCP-ECG statements and likelihoods, and the original 100 Hz and 500 Hz WFDB records.
|
| 30 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 31 |
## Key features
|
| 32 |
|
| 33 |
- The original `ecg_id`, pseudonymous `patient_id`, recommended fold, and source paths are retained.
|
|
|
|
| 17 |
- config_name: default
|
| 18 |
data_files:
|
| 19 |
- split: train
|
| 20 |
+
path: "viewer/default/train.parquet"
|
| 21 |
- split: validation
|
| 22 |
+
path: "viewer/default/validation.parquet"
|
| 23 |
- split: test
|
| 24 |
+
path: "viewer/default/test.parquet"
|
| 25 |
---
|
| 26 |
|
| 27 |
# PTB-XL v1.0.3 in Lance
|
| 28 |
|
| 29 |
Search the full PTB-XL corpus for ECGs with particular annotated findings, then retrieve the matching 12-lead waveforms. Each result represents one 10-second ECG. Its row contains the original report, the joined SCP-ECG statements and likelihoods, and the original 100 Hz and 500 Hz WFDB records.
|
| 30 |
|
| 31 |
+
## Browse on Hugging Face
|
| 32 |
+
|
| 33 |
+
Browse every ECG by original ID, fold, report, and SCP-ECG annotation. Each row links to the original 100 Hz and 500 Hz WFDB headers. Use the LanceDB example below for vector search and both waveform files. [Open the Dataset Viewer](https://huggingface.co/datasets/lance-format/ptb-xl-lance/viewer/default/test)
|
| 34 |
+
|
| 35 |
+
The Dataset Viewer reads the Parquet browse tables in `viewer/`. LanceDB opens the indexed tables in `data/`.
|
| 36 |
+
|
| 37 |
## Key features
|
| 38 |
|
| 39 |
- The original `ecg_id`, pseudonymous `patient_id`, recommended fold, and source paths are retained.
|
viewer/default/test.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:45b80ec18890123179fb345427f00406a8f8c041ff80110061769759f50bb9cc
|
| 3 |
+
size 188773
|
viewer/default/train.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d59c1da280c6b42c6d9e02a56856646d3e45372db245191dc2ab74eb5fee6546
|
| 3 |
+
size 1290945
|
viewer/default/validation.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b469d6a379a054d4979ce6f74a98b11bd7dc7cbbd015feebf56b810cbe6285d7
|
| 3 |
+
size 186605
|