Question about Data Preprocessing Pipeline
Hello, I encountered some issues while reproducing your work. I found that when directly feeding the preprocessed data from the RadGenome-ChestCT dataset into the SAT model, I cannot obtain the expected segmentation results. Therefore, I'm confused about the complete data preprocessing pipeline and hope to get your clarification.
Regarding the preprocessing pipeline for the CT-Rate dataset, I would like to confirm which of the following is the correct order of operations?
Pipeline A: Preprocess first, then segment
First use preprocess_ctrate_train.py to preprocess the original CT-Rate data
Then feed the preprocessed data into the SAT model for organ segmentation
Pipeline B: Segment first, then preprocess
First directly feed the original CT-Rate data into the SAT model for organ segmentation
SAT outputs its own preprocessed CT data along with the corresponding segmentation masks
Then use the SAT outputs (CT and masks) as input to preprocess_ctrate_train.py for further unified preprocessing
Could you please clarify the complete data processing pipeline? If it's neither of the above two pipelines, could you provide the actual preprocessing pipeline and code used?
Thank you for your help!