MendeleyCXR: Montgomery + Shenzhen subsets (704 samples)

MendeleyCXR/LICENSE

@@ -0,0 +1,5 @@
+CC BY 4.0
+https://creativecommons.org/licenses/by/4.0/
+
+Derived from: https://data.mendeley.com/datasets/8gf9vpkhgy/
+Attribution: Danilov, V., Proutski, A., Kirpich, A., Litmanovich, D. & Gankin, Y. Chest X-ray dataset for lung segmentation, v2. Mendeley Data (2022). doi:10.17632/8gf9vpkhgy.2. The collection aggregates the V7 Darwin COVID-19 chest X-ray set (images from Kermany et al., Mendeley Data doi:10.17632/rscbjbr9sj.2, and from the ieee8023 covid-chestxray-dataset, Cohen et al. 2020), the Montgomery County set, and the Shenzhen set (Jaeger et al. 2014, NIH/NLM); please credit those sources as well.

MendeleyCXR/README.md

@@ -0,0 +1,31 @@
+# MendeleyCXR (chest radiographs with lung masks; Montgomery and Shenzhen subsets)
+
+Part of the FleXray data release. Repackaged from the original distribution at
+https://data.mendeley.com/datasets/8gf9vpkhgy/ under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).
+
+**Citation:** Danilov, V., Proutski, A., Kirpich, A., Litmanovich, D. & Gankin, Y. Chest X-ray dataset for lung segmentation, v2. Mendeley Data (2022). doi:10.17632/8gf9vpkhgy.2. The collection aggregates the V7 Darwin COVID-19 chest X-ray set (images from Kermany et al., Mendeley Data doi:10.17632/rscbjbr9sj.2, and from the ieee8023 covid-chestxray-dataset, Cohen et al. 2020), the Montgomery County set, and the Shenzhen set (Jaeger et al. 2014, NIH/NLM); please credit those sources as well.
+
+## Contents
+
+- 704 samples (test: 105, train: 493, val: 106); one image and one label per sample.
+- `images/*.png`: 16-bit grayscale PNG, intensities in [0, 65535] map linearly to [0, 1].
+- `labels/`: dense index map (PNG). Label ids/channels:
+- `0`: background
+- `1`: lungs
+- `dataset.yml`: `fxr-dataset` manifest (schema version 1). Splits are the
+  ones used in the FleXray paper; `metadata.original_shape` is the source image shape.
+
+## Preprocessing
+
+Every image was min-max normalized to [0, 1] per image, padded to a square with zeros
+(centered) and resized to 256 x 256 (area interpolation for images, nearest for labels). Labels were remapped from the
+source annotation to the dataset-native names listed above; the FleXray package maps
+those names into its label protocol at training time.
+
+## Usage
+
+```bash
+pip install flexray
+fxr-dataset validate dataset.yml
+fxr-dataset pack dataset.yml /data/flexray/MendeleyCXR
+```