From 3eed5c4793eadb493eb0f2efa6d3422d85f70fa7 Mon Sep 17 00:00:00 2001
From: Josh Williams <48445365+jvwilliams23@users.noreply.github.com>
Date: Mon, 20 Feb 2023 16:33:56 +0000
Subject: [PATCH 1/4] Update README.md

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 619dbad..38ac9a4 100644
--- a/README.md
+++ b/README.md
@@ -28,7 +28,7 @@ These will then be split into train and validation sets (which are saved as `tra
 python3 train.py
 ```
 
-### Segmenting
+### Using a CNN to segment an image
 
 ## Get Help
 Please submit an issue to the issues panel on this repository.

From a7f6506f8b4704307343db14e2a5ef8ff659dce6 Mon Sep 17 00:00:00 2001
From: Josh Williams <48445365+jvwilliams23@users.noreply.github.com>
Date: Mon, 20 Feb 2023 16:36:06 +0000
Subject: [PATCH 2/4] Update README.md

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 38ac9a4..1a6f6d0 100644
--- a/README.md
+++ b/README.md
@@ -1,4 +1,4 @@
-# Airway and lung segmentations using convolutional neural networks
+# Convolutional neural network toolkit for segmenting airways and lungs
 
 ## Overview
 This project aimed to train a machine learning algorithm capable of producing airway and lung segmentations from chest CT scans. 

From aabe00c22ee1c785a87a37a1a10eb1cb9715088a Mon Sep 17 00:00:00 2001
From: Josh Williams <48445365+jvwilliams23@users.noreply.github.com>
Date: Mon, 20 Feb 2023 17:07:44 +0000
Subject: [PATCH 3/4] Update README.md

---
 README.md | 25 ++++++++++++++++++++++++-
 1 file changed, 24 insertions(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 1a6f6d0..183c48b 100644
--- a/README.md
+++ b/README.md
@@ -14,7 +14,7 @@ First, install the required dependencies (best practice is to use a virtual envi
 ```bash
 conda create --name pycnn python=3.10
 conda activate pycnn
-pip install hjson numpy pandas torch torchio SimpleITK sklearn vedo
+pip install hjson numpy pandas torch torchio SimpleITK sklearn vedo lungmask
 ```
 
 ### Retraining
@@ -29,6 +29,15 @@ python3 train.py
 ```
 
 ### Using a CNN to segment an image
+To lower the memory consumption and inference time for segmenting the airways, we can first segment the lobes and use this as a bounding box for cropping the image
+```bash
+python segment_lunglobes.py -i /path/to/ct/scan.mhd -id example
+```
+The airways can then be segmented by 
+```bash
+python segment.py -i /path/to/ct/scan.mhd -o example.stl
+```
+which will create three files in the current directory (`example.mhd`, `example.zraw` and `example.stl`).
 
 ## Get Help
 Please submit an issue to the issues panel on this repository.
@@ -39,5 +48,19 @@ If you use the code or models in this repository, please cite our paper
 @article{TODO}
 ```
 
+Also, if you use the `segment_lunglobes.py` script, make sure to read and cite the following great paper:
+```
+@article{hofmanninger2020automatic,
+  title={Automatic lung segmentation in routine imaging is primarily a data diversity problem, not a methodology problem},
+  author={Hofmanninger, Johannes and Prayer, Forian and Pan, Jeanny and R{\"o}hrich, Sebastian and Prosch, Helmut and Langs, Georg},
+  journal={European Radiology Experimental},
+  volume={4},
+  number={1},
+  pages={1--13},
+  year={2020},
+  publisher={SpringerOpen}
+}
+```
+
 ## Acknowledgements
 The model architecture scripts [`unet.py`](https://github.com/Thvnvtos/Lung_Segmentation/blob/unet3d/model.py) and [`enet.py`](https://github.com/davidtvs/PyTorch-ENet/blob/master/models/enet.py) were taken from other repositories (see links).

From a7996b7d5ea14d62457ae035a47ec09cf1e53fc0 Mon Sep 17 00:00:00 2001
From: Josh Williams <48445365+jvwilliams23@users.noreply.github.com>
Date: Mon, 20 Feb 2023 17:08:29 +0000
Subject: [PATCH 4/4] Update typo in README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 183c48b..6e83fb4 100644
--- a/README.md
+++ b/README.md
@@ -25,7 +25,7 @@ To retrain a CNN on some data you have available, add the paths to the images an
 ```
 These will then be split into train and validation sets (which are saved as `train_cases.txt` and `val_cases.txt`). Also you can edit any hyperparameters defined in `trainconfig.json` and run
 ```bash
-python3 train.py
+python train.py
 ```
 
 ### Using a CNN to segment an image