Data processing

The data processing is a important part of any AUCMEDI pipeline and represented by two core pillars.

• The input_interface function provides important information on the dataset.
• The DataGenerator class is a powerful interface handling all data processing in AUCMEDI.

Pillars of AUCMEDI

• aucmedi.data_processing.io_data.input_interface
• aucmedi.data_processing.data_generator.DataGenerator
• aucmedi.neural_network.model.NeuralNetwork

With an initialized Neural Network instance, it is possible to run training and predictions.

Example

# Import
from aucmedi import *

# Initialize input data reader
ds = input_interface(interface="csv",
                     path_imagedir="dataset/images/",
                     path_data="dataset/annotations.csv",
                     ohe=False, col_sample="ID", col_class="diagnosis")
(samples, class_ohe, nclasses, class_names, image_format) = ds

# Initialize model
model = NeuralNetwork(n_labels=nclasses, channels=3, architecture="2D.ResNet50")

# Do some training
datagen_train = DataGenerator(samples[:100], "images_dir/", labels=class_ohe[:100], image_format=image_format,
                              resize=model.meta_input, standardize_mode=model.meta_standardize)
model.train(datagen_train, epochs=50)

# Do some predictions
datagen_test = DataGenerator(samples[100:150], "images_dir/", labels=None, image_format=image_format,
                             resize=model.meta_input, standardize_mode=model.meta_standardize)
preds = model.predict(datagen_test)