You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: PW45_2026_Boston/Projects/AiModelDevelopmentForLungUltrasoundAnalysis/README.md
+74Lines changed: 74 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -77,6 +77,8 @@ Therefore, we have started to develop detection models to automatically find ple
77
77
78
78
1. We have trained some preliminary YOLO models for bounding box detection. We trained one oriented bounding box YOLO model for pleural lines, and another axis-aligned box YOLO model for B-lines.
79
79
2. We calculated the percent pleura and compared it to the experts.
80
+
3. Implemented and validated a data preprocessing script: 1) parses data and formats into COCO-styled repository, 2) enables sector-to-scanline conversions, 3) enables filtering by metadata (_site, annotator, lung zone, patients, before/after diuretic, transducer type, transducer manufacturer_)
81
+
4. Implemented a dataloader that allocates equal proportions of different metadata tags to train/vali/test datasets. Implemented k-fold validation and inclusion of data by various metadata tags.
80
82
81
83
82
84
@@ -95,6 +97,78 @@ YOLO results after performing pleural line detection and B-line detection, and c
95
97
Percent pleura for one test lung zone - x = frame #, and y = percentage. We see that the AI predicted
0 commit comments