search

kbressem / prostate158

MIT License
26 stars 7 forks source link

Preprocess for Segmentation #13

Open Asagami-Fujino opened 4 days ago

Asagami-Fujino commented 4 days ago

Thank you for providing the data! But I've found mismatch in different modalities. Did you register images of different modalities before segmentation?

Asagami-Fujino commented 3 days ago

And another quesition: during training process which label is used as the GT, adc_tumor_reader1, adc_tumor_reader2, or t2_tumor_reader1?

kbressem commented 2 days ago

They were resampled using DICOM header information, but not registered using more advanced methods. I believe we used reader 1 as reference during training