Hello MedSAM2 team,
Thank you for your great work! While reviewing the dataset list, I noticed that the number of images for the SegThy dataset(US, MRI) is listed as 32 and 14.
Since the SegThy dataset consists of ultrasound (US) volume data, I have a few questions regarding how the data was processed for training:
Were all frames within a given sweep used for model training, or was only a specific subset of frames extracted and used?
Given that ultrasound data is provided as a volume, did the training use slices from all axes (e.g., axial, sagittal, coronal), or was it restricted to a single specific axis?
Thank you for your time and clarification!
Hello MedSAM2 team,
Thank you for your great work! While reviewing the dataset list, I noticed that the number of images for the SegThy dataset(US, MRI) is listed as 32 and 14.
Since the SegThy dataset consists of ultrasound (US) volume data, I have a few questions regarding how the data was processed for training:
Were all frames within a given sweep used for model training, or was only a specific subset of frames extracted and used?
Given that ultrasound data is provided as a volume, did the training use slices from all axes (e.g., axial, sagittal, coronal), or was it restricted to a single specific axis?
Thank you for your time and clarification!