update readme

Author: taigw

examples/JSRT/README.md

@@ -38,7 +38,7 @@ mkdir result
 pymic_net_run test config/train_test.cfg
 ```
 
-2. Then edit `config/evaluation.cfg` by setting `ground_truth_folder_list` as your `JSRT_root/label`, and run the following command to obtain quantitative evaluation results in terms of dice. 
+2. Then edit `config/evaluation.cfg` by setting `ground_truth_folder_root` as your `JSRT_root`, and run the following command to obtain quantitative evaluation results in terms of dice. 
 
 ```bash
 pymic_evaluate config/evaluation.cfg

examples/JSRT2/README.md

@@ -3,11 +3,11 @@
 ![image_example](../JSRT/picture/JPCLN001.png)
 ![label_example](../JSRT/picture/JPCLN001_seg.png)
 
-In this example, we show how to use a customized CNN to segment the heart from X-Ray images. The configurations are the same as those in the `JSRT` example except the network structure. 
+In this example, we show how to use a customized CNN and a customized loss function to segment the heart from X-Ray images. The configurations are the same as those in the `JSRT` example except the network structure and loss function. 
 
-The customized CNN is detailed in `my_net2d.py`, which is a modification of the 2D UNet. In this new network, we use a residual connection in each block. 
+The customized CNN is detailed in `my_net2d.py`, which is a modification of the 2D UNet. In this new network, we use a residual connection in each block. The customized loss is detailed in `my_loss.py`, where we combine Dice loss and MAE loss as our new loss function.
 
-To use the customized CNN, we also write a customized main function in `jsrt_train_infer.py`, where we import a TrainInferAgent from PyMIC and set the network as our customized CNN.
+We also write a customized main function in `jsrt_train_infer.py` so that we can combine TrainInferAgent from PyMIC with our customized CNN and loss function.
 
 ## Data and preprocessing
 1. Data preprocessing is the same as that in the the `JSRT` example. Please follow that example for details.
@@ -31,7 +31,7 @@ python jsrt_net_run.py train config/train_test.cfg
 python jsrt_net_run.py test config/train_test.cfg
 ```
 
-2. Edit `config/evaluation.cfg` by setting `ground_truth_folder_list` as your `JSRT_root/label`, and run the following command to obtain quantitative evaluation results in terms of dice.
+2. Edit `config/evaluation.cfg` by setting `ground_truth_folder_root` as your `JSRT_root`, and run the following command to obtain quantitative evaluation results in terms of dice.
 
 ```
 pymic_evaluate config/evaluation.cfg

examples/JSRT2/config/evaluation.cfg

@@ -3,13 +3,11 @@ metric = dice
 label_list = [255]
 organ_name = heart
 
-segmentation_folder_list  = [result]
-segmentation_postfix = 
-segmentation_format  = png
-
-ground_truth_folder_list  = [/home/disk2t/data/JSRT/label]
-ground_truth_postfix = 
-ground_truth_format  = png
-
-patient_file_names  = ../JSRT/config/jsrt_test_names.txt
-
+ground_truth_folder_root  = /home/disk2t/data/JSRT
+segmentation_folder_root  = result
+evaluation_image_pair     = ../JSRT/config/jsrt_test_gt_seg.csv
+
+ground_truth_label_convert_source = None
+ground_truth_label_convert_target = None
+segmentation_label_convert_source = None
+segmentation_label_convert_target = None

examples/JSRT2/config/train_test.cfg

@@ -7,6 +7,7 @@ train_csv = ../JSRT/config/jsrt_train.csv
 valid_csv = ../JSRT/config/jsrt_valid.csv
 test_csv  = ../JSRT/config/jsrt_test.csv
 
+load_pixelwise_weight = False
 # modality number
 modal_num = 1
 
@@ -37,12 +38,22 @@ LabelToProbability_inverse   = False
 # this section gives parameters for network
 # In this example, a customized network is used.
 
+# type of network
+net_type = MyUNet2D
+
+# number of class, required for segmentation task
+class_num     = 2
+in_chns       = 1
+feature_chns  = [4, 16, 24, 32, 48]
+dropout       = [0.0, 0.0, 0.3, 0.3, 0.4, 0.5]
+bilinear     = True
+
 [training]
 # device name" cuda:n or cpu
 device_name = cuda:0
 
 batch_size    = 4
-loss_function = dice_loss
+loss_function = my_loss
 
 # for optimizers
 optimizer     = Adam

examples/JSRT2/jsrt_net_run.py

@@ -3,8 +3,24 @@
 
 import sys
 from pymic.net_run.net_run import  TrainInferAgent
+from pymic.net_run.net_factory import net_dict
 from pymic.util.parse_config import parse_config
 from my_net2d import MyUNet2D 
+from my_loss  import MySegmentationLossCalculator
+
+my_net_dict = {
+    "MyUNet2D": MyUNet2D
+}
+
+def get_network(params):
+    net_type = params["net_type"]
+    if(net_type in my_net_dict):
+        net = my_net_dict[net_type](params)
+    elif(net_type in net_dict):
+        net = net_dict[net_type](params)
+    else:
+        raise ValueError("Undefined network: {0:}".format(net_type))
+    return net 
 
 def main():
     if(len(sys.argv) < 3):
@@ -15,17 +31,12 @@ def main():
     cfg_file = str(sys.argv[2])
     config   = parse_config(cfg_file)
 
-    # use custormized CNN
-    net_param = {'in_chns':1,
-              'feature_chns':[4, 16, 32, 64, 128],
-              'dropout':  [0, 0, 0.3, 0.4, 0.5],
-              'class_num': 2,
-              'bilinear': True}
-    config['network'] = net_param
-    net    = MyUNet2D(net_param)
-
+    # use custormized CNN and loss function
+    net      = get_network(config['network'])
+    loss_cal = MySegmentationLossCalculator(config['training'])
     agent  = TrainInferAgent(config, stage)
     agent.set_network(net)
+    agent.set_loss_calculater(loss_cal)
     agent.run()
 
 if __name__ == "__main__":

examples/fetal_hc/README.md

@@ -35,7 +35,7 @@ mkdir result
 pymic_net_run test config/train_test.cfg
 ```
 
-2. Then edit `config/evaluation.cfg` by setting `ground_truth_folder_list` as your `HC_root/training_set_label`, and run the following command to obtain quantitative evaluation results in terms of dice. 
+2. Then edit `config/evaluation.cfg` by setting `ground_truth_folder_root` as your `HC_root`, and run the following command to obtain quantitative evaluation results in terms of dice. 
 
 ```bash
 pymic_evaluate config/evaluation.cfg

examples/prostate/README.md

@@ -33,7 +33,7 @@ mkdir result
 pymic_net_run test config/train_test.cfg
 ```
 
-2. Then edit `config/evaluation.cfg` by setting `ground_truth_folder_list` as your `data/promise12/preprocess/label`, and run the following command to obtain quantitative evaluation results in terms of dice. 
+2. Then edit `config/evaluation.cfg` by setting `ground_truth_folder_root` as your `data/promise12/preprocess`, and run the following command to obtain quantitative evaluation results in terms of dice. 
 
 ```bash
 pymic_evaluate config/evaluation.cfg