Use load_ decathlon_ Datalist (Monai) fast loading JSON data

reflection ： When you have a data that contains all the information JSON When you file , How to load into MONAI Inside the frame ？

As shown in the figure , From this json You can know the data in the file label Information , The true meaning of each category . And its mode is CT, Image is 3D And put the training set 、 The address of the image included in the verification set and the test connection is written .

We can feed the data to the model through this file .

stay MONAI It does provide loading json How to format data . Very convenient .

Empathy , We can write our own data as json Format loading .

This tutorial will cover these two contents , If you are interested, let's have a look

---

load_decathlon_datalist Load data

Where to call ？

from monai.data import load_decathlon_datalist

Function parameter
load_decathlon_datalist(data_list_file_path: PathLike,is_segmentation: bool = True, data_list_key: str = “training”, base_dir: Optional[PathLike] = None,)

Args:

• data_list_file_path: the path to the json file of datalist. Yours json File address
• is_segmentation: whether the datalist is for segmentation task, default is True. Whether it is a split task
• data_list_key: the key to get a list of dictionary to be used, default is “training”. Which data set do you want to load （traning, validation, test）, there key It's worth it json The name of the corresponding dataset in the file （ Look at the picture above ）.
• base_dir: the base directory of the dataset, if None, use the datalist directory. Home directory of data . You can know from the picture , The address of the data is from imagesTs/imagesTr/labelsTr At the beginning . And the upper level address of these addresses needs to be provided . If you don't fill in , Default and json The files are in the same directory .

Demo sample ：

I am here tested.py Load data in the file

from monai.data import load_decathlon_datalist
data_dir = "dataset/dataset.json"
datalist = load_decathlon_datalist(data_dir, True, "training 
# datalist = load_decathlon_datalist(data_dir, True, "training", 'dataset')  add base_dir

such , accord with MONAI data Your dictionary will be created .

We can see , To have this json file , We can easily create data .

Next , Let's see how to create this json file

---

Create data json file

from collections import OrderedDict
import json

json_dict = OrderedDict()
json_dict['name'] = "your task"
json_dict['description'] = "btcv yucheng"
json_dict['tensorImageSize'] = "3D"
json_dict['reference'] = "see challenge website"
json_dict['licence'] = "see challenge website"
json_dict['release'] = "0.0"
json_dict['modality'] = {
    
        "0": "CT",
    }
json_dict['test'] = [
    "imagesTs/img0061.nii.gz",
    "imagesTs/img0062.nii.gz",
    "imagesTs/img0063.nii.gz",
    "imagesTs/img0064.nii.gz",
    "imagesTs/img0065.nii.gz",
    "imagesTs/img0066.nii.gz"]  #  Write in the list containing the data .
    
#  What information do you want to save , stay json_dict Add a dictionary data inside 

#  preservation json
with open(os.path.join(out_base, "dataset.json"), 'w') as f:
    json.dump(json_dict, f, indent=4, separators=(',', ': '))

Articles are constantly updated , You can pay attention to the official account of WeChat 【 Medical image AI combat camp 】 Get the latest , The official account of the frontier technology in the field of medical image processing . Stick to the practice , Take you hand in hand to do the project , Play the game , Write a paper . All original articles provide theoretical explanation , Experimental code , experimental data . Only practice can grow faster , Pay attention to our , Learn together ~

I am a Tina, I'll see you on our next blog ~

Working during the day and writing at night , cough

If you think it's well written, finally , Please thumb up , Comment on , Collection . Or three times with one click