search

frankkramer-lab / MIScnn

A framework for Medical Image Segmentation with Convolutional Neural Networks and Deep Learning
GNU General Public License v3.0
402 stars 116 forks source link

PNG / NPY IO #13

Closed asmagen closed 4 years ago

asmagen commented 4 years ago

Hello @muellerdo,

How do I perform multi class semantic (instance) segmentation over my set of 2D PNG/TIF grayscale images and their corresponding masks file (NPY numpy array, TIF or any other) matrix representation with # channels corresponding to the classes? Do I need to use NIFTI_interface? If so, how do I convert my PNG/TIF/NPY files to this format?

Thank you

muellerdo commented 4 years ago

Hey @asmagen,

thanks for your interest in MIScnn.

Yesterday, I implemented an image IO interface for more intuitive and easier integration of 2D image formats into MIScnn.

You have always the possibility to implement a custom data IO interface for your own file structure, but now you can also use the Image_interface class with a given fixed file structure.

I added an documented example of using the Image_interface on the ISBI cell tracking challenge 2015. https://github.com/frankkramer-lab/MIScnn/blob/master/examples/CellTracking.ipynb

matrix representation with # channels corresponding to the classes

The number of channels is defined by your imaging data. Grayscale has 1 channel whereas RGB has 3 channels.

I hope that I was able to help you. If you have further questions don't hesitate to ask.

Also I'm always interested in projects using MIScnn, so if you want to share your project/code as an usage example, you are more than welcome :)

Cheers, Dominik

PS: Copy & Paste? :P https://github.com/black0017/MedicalZooPytorch/issues/5 https://github.com/Project-MONAI/MONAI/issues/471

asmagen commented 4 years ago

Thank you. Regarding more project info and questions, I thought you might have an insight to suggest for my post here.

muellerdo commented 4 years ago

I'm sorry to say that due to time limiting issues currently, I am unable to provide support for non-related MIScnn questions.

But segmentation_models is a remarkable and popular package, so I'm optimistic that you will find good support there! :)