search

ymli39 / DeepSEED-3D-ConvNets-for-Pulmonary-Nodule-Detection

DeepSEED: 3D Squeeze-and-Excitation Encoder-Decoder ConvNets for Pulmonary Nodule Detection
MIT License
109 stars 33 forks source link

Need some clarification on outputs #16

Closed shakjm closed 4 years ago

shakjm commented 4 years ago

Hi there, I'm trying to replicate your model and I was wondering how are your outputs being used? They're lbb and pbb. I'm guessing lbb is the labels, where as pbb is a feature vector of [x,5]. What are they representing?

Also, is there any way I can generate an image to see where the bounding boxes land? How do I do that and how do the coordinates match with the images that were preprocessed?

ymli39 commented 4 years ago

The postprocessing code is located in /LIDC_detector/FROCeval.py. [x,5] represents x,y,z,d, and probability. You need to convert x,y,z,d to bounding boxes then draw boxes on the images. Each images are preprocessed based on its own spacing, origin, to convert it back to original coordinate, check /LIDC_detector/FROCeval.py.

shakjm commented 4 years ago

Thank you Mr. Li for the clarification.