Added YAML metadata to README

README.md

@@ -1,5 +1,18 @@
+---
+license: mit
+task_categories:
+- image-segmentation
+language:
+- en
+tags:
+- medical
+pretty_name: AeroPath
+size_categories:
+- 1B<n<10B
+---
 This README file was generated on 03-11-2023 by David Bouget.
-Last updated: 03-11-2023.
+It was then updated on 07-11-2023 to add Hugging Face metadata by André Pedersen.
+Last updated: 07-11-2023.
 
 --------------------
 GENERAL INFORMATION
@@ -60,4 +73,4 @@ Dataset statistics
 Overall, the CT scan dimensions are covering [487 : 512] × [441 : 512] × [241 : 829] voxels, and the trans-axial voxel size ranges [0.68 : 0.76] × [0.68 : 0.75] mm2 with a reconstructed slice thickness of [0.5 : 1.25] mm.
 
 Annotation procedures
-The annotation process for generating the ground truth was performed in three steps. First, the largest components (i.e., trachea and the first branches) were extracted based on a region growing, or a grow-cut method. Due to leakage, the region growing method did not yield satisfactory results in all cases. Therefore, for certain cases, the grow-cut method in 3D Slicer was used instead. In the second step, BronchiNet was employed to segment the smaller peripheral airways. In the third and final step, the segmentations were refined manually. Bronchial fragments and missed segments were connected, before false positives and fragments that could not be connected based on visual inspection were removed. All manual corrections were performed using the default segment editor in 3D Slicer. The manual correction was performed by a trained engineer, supervised by a pulmonologist. Finally, all annotations were verified on a case-by-case basis by a pulmonologist. The final annotations from the AeroPath segmentation included on average 128 ± 56 branches per CT scan.
+The annotation process for generating the ground truth was performed in three steps. First, the largest components (i.e., trachea and the first branches) were extracted based on a region growing, or a grow-cut method. Due to leakage, the region growing method did not yield satisfactory results in all cases. Therefore, for certain cases, the grow-cut method in 3D Slicer was used instead. In the second step, BronchiNet was employed to segment the smaller peripheral airways. In the third and final step, the segmentations were refined manually. Bronchial fragments and missed segments were connected, before false positives and fragments that could not be connected based on visual inspection were removed. All manual corrections were performed using the default segment editor in 3D Slicer. The manual correction was performed by a trained engineer, supervised by a pulmonologist. Finally, all annotations were verified on a case-by-case basis by a pulmonologist. The final annotations from the AeroPath segmentation included on average 128 ± 56 branches per CT scan.