biomedisa 2024.5.14__py3-none-any.whl

biomedisa/biomedisa_features/crop_helper.py

@@ -0,0 +1,586 @@
+##########################################################################
+##                                                                      ##
+##  Copyright (c) 2024 Philipp Lösel. All rights reserved.              ##
+##                                                                      ##
+##  This file is part of the open source project biomedisa.             ##
+##                                                                      ##
+##  Licensed under the European Union Public Licence (EUPL)             ##
+##  v1.2, or - as soon as they will be approved by the                  ##
+##  European Commission - subsequent versions of the EUPL;              ##
+##                                                                      ##
+##  You may redistribute it and/or modify it under the terms            ##
+##  of the EUPL v1.2. You may not use this work except in               ##
+##  compliance with this Licence.                                       ##
+##                                                                      ##
+##  You can obtain a copy of the Licence at:                            ##
+##                                                                      ##
+##  https://joinup.ec.europa.eu/page/eupl-text-11-12                    ##
+##                                                                      ##
+##  Unless required by applicable law or agreed to in                   ##
+##  writing, software distributed under the Licence is                  ##
+##  distributed on an "AS IS" basis, WITHOUT WARRANTIES                 ##
+##  OR CONDITIONS OF ANY KIND, either express or implied.               ##
+##                                                                      ##
+##  See the Licence for the specific language governing                 ##
+##  permissions and limitations under the Licence.                      ##
+##                                                                      ##
+##########################################################################
+
+import os
+from biomedisa_features.keras_helper import read_img_list, remove_extracted_data
+from biomedisa_features.biomedisa_helper import img_resize, load_data, save_data, set_labels_to_zero
+from tensorflow.python.framework.errors_impl import ResourceExhaustedError
+from tensorflow.keras.applications import DenseNet121, densenet
+from tensorflow.keras.optimizers import Adam
+from tensorflow.keras.models import Model, load_model
+from tensorflow.keras.layers import Input, GlobalAveragePooling2D, Dropout, Dense
+from tensorflow.keras.callbacks import Callback, ModelCheckpoint
+from biomedisa_features.DataGeneratorCrop import DataGeneratorCrop
+from biomedisa_features.PredictDataGeneratorCrop import PredictDataGeneratorCrop
+import tensorflow as tf
+import numpy as np
+from glob import glob
+import h5py
+import tarfile
+import matplotlib.pyplot as plt
+
+class InputError(Exception):
+    def __init__(self, message=None, img_names=[], label_names=[]):
+        self.message = message
+        self.img_names = img_names
+        self.label_names = label_names
+
+def save_history(history, path_to_model):
+    # summarize history for accuracy
+    plt.plot(history['accuracy'])
+    plt.plot(history['val_accuracy'])
+    if 'val_loss' in history:
+        plt.legend(['train', 'test'], loc='upper left')
+    else:
+        plt.legend(['train', 'test (Dice)'], loc='upper left')
+    plt.title('model accuracy')
+    plt.ylabel('accuracy')
+    plt.xlabel('epoch')
+    plt.tight_layout()  # To prevent overlapping of subplots
+    plt.savefig(path_to_model.replace(".h5","_acc.png"), dpi=300, bbox_inches='tight')
+    plt.clf()
+    # summarize history for loss
+    plt.plot(history['loss'])
+    if 'val_loss' in history:
+        plt.plot(history['val_loss'])
+    plt.title('model loss')
+    plt.ylabel('loss')
+    plt.xlabel('epoch')
+    plt.legend(['train', 'test'], loc='upper left')
+    plt.tight_layout()  # To prevent overlapping of subplots
+    plt.savefig(path_to_model.replace(".h5","_loss.png"), dpi=300, bbox_inches='tight')
+    plt.clf()
+
+def make_densenet(inputshape):
+    base_model = DenseNet121(
+        input_tensor=Input(inputshape),
+        include_top=False,)
+    
+    base_model.trainable= False
+    
+    inputs = Input(inputshape)
+    x = densenet.preprocess_input(inputs)
+
+    x = base_model(x, training=False)
+    x = GlobalAveragePooling2D()(x)
+    x = Dropout(0.3)(x)
+
+    outputs = Dense(1, activation='sigmoid')(x)
+
+    model = Model(inputs, outputs)
+    return model
+
+def load_cropping_training_data(normalize, img_list, label_list, x_scale, y_scale, z_scale,
+    labels_to_compute, labels_to_remove, img_in, label_in, normalization_parameters=None, channels=None):
+
+    # read image lists
+    if any(img_list):
+        img_names, label_names = read_img_list(img_list, label_list)
+        InputError.img_names = img_names
+        InputError.label_names = label_names
+
+    # load first label
+    if any(img_list):
+        a, _, _ = load_data(label_names[0], 'first_queue', True)
+        if a is None:
+            InputError.message = f'Invalid label data "{os.path.basename(label_names[0])}"'
+            raise InputError()
+    elif type(label_in) is list:
+        a = label_in[0]
+    else:
+        a = label_in
+    a = a.astype(np.uint8)
+    a = set_labels_to_zero(a, labels_to_compute, labels_to_remove)
+    label_z = np.any(a,axis=(1,2))
+    label_y = np.any(a,axis=(0,2))
+    label_x = np.any(a,axis=(0,1))
+    label = np.append(label_z,label_y,axis=0)
+    label = np.append(label,label_x,axis=0)
+
+    # load first img
+    if any(img_list):
+        img, _ = load_data(img_names[0], 'first_queue')
+        if img is None:
+            InputError.message = f'Invalid image data "{os.path.basename(img_names[0])}"'
+            raise InputError()
+    elif type(img_in) is list:
+        img = img_in[0]
+    else:
+        img = img_in
+    # handle all images having channels >=1
+    if len(img.shape)==3:
+        z_shape, y_shape, x_shape = img.shape
+        img = img.reshape(z_shape, y_shape, x_shape, 1)
+    if channels is None:
+        channels = img.shape[3]
+    if img.shape[3] != channels:
+        InputError.message = f'Number of channels must be {channels} for "{os.path.basename(img_names[0])}"'
+        raise InputError()
+    img = img.astype(np.float32)
+    img_z = img_resize(img, a.shape[0], y_scale, x_scale)
+    img_y = np.swapaxes(img_resize(img, z_scale, a.shape[1], x_scale),0,1)
+    img_x = np.swapaxes(img_resize(img, z_scale, y_scale, a.shape[2]),0,2)
+    img = np.append(img_z,img_y,axis=0)
+    img = np.append(img,img_x,axis=0)
+
+    # normalize image data
+    for c in range(channels):
+        img[:,:,:,c] -= np.amin(img[:,:,:,c])
+        img[:,:,:,c] /= np.amax(img[:,:,:,c])
+        if normalization_parameters is None:
+            normalization_parameters = np.zeros((2,channels))
+            normalization_parameters[0,c] = np.mean(img[:,:,:,c])
+            normalization_parameters[1,c] = np.std(img[:,:,:,c])
+        elif normalize:
+            mean, std = np.mean(img[:,:,:,c]), np.std(img[:,:,:,c])
+            img[:,:,:,c] = (img[:,:,:,c] - mean) / std
+            img[:,:,:,c] = img[:,:,:,c] * normalization_parameters[1,c] + normalization_parameters[0,c]
+
+    # loop over list of images
+    if any(img_list) or type(img_in) is list:
+        number_of_images = len(img_names) if any(img_list) else len(img_in)
+
+        for k in range(1, number_of_images):
+
+            # append label
+            if any(label_list):
+                a, _ = load_data(label_names[k], 'first_queue')
+                if a is None:
+                    InputError.message = f'Invalid label data "{os.path.basename(label_names[k])}"'
+                    raise InputError()
+            else:
+                a = label_in[k]
+            a = a.astype(np.uint8)
+            a = set_labels_to_zero(a, labels_to_compute, labels_to_remove)
+            next_label_z = np.any(a,axis=(1,2))
+            next_label_y = np.any(a,axis=(0,2))
+            next_label_x = np.any(a,axis=(0,1))
+            label = np.append(label,next_label_z,axis=0)
+            label = np.append(label,next_label_y,axis=0)
+            label = np.append(label,next_label_x,axis=0)
+
+            # append image
+            if any(img_list):
+                a, _ = load_data(img_names[k], 'first_queue')
+                if a is None:
+                    InputError.message = f'Invalid image data "{os.path.basename(img_names[k])}"'
+                    raise InputError()
+            else:
+                a = img_in[k]
+            if len(a.shape)==3:
+                z_shape, y_shape, x_shape = a.shape
+                a = a.reshape(z_shape, y_shape, x_shape, 1)
+            if a.shape[3] != channels:
+                InputError.message = f'Number of channels must be {channels} for "{os.path.basename(img_names[k])}"'
+                raise InputError()
+            a = a.astype(np.float32)
+            img_z = img_resize(a, a.shape[0], y_scale, x_scale)
+            img_y = np.swapaxes(img_resize(a, z_scale, a.shape[1], x_scale),0,1)
+            img_x = np.swapaxes(img_resize(a, z_scale, y_scale, a.shape[2]),0,2)
+            next_img = np.append(img_z,img_y,axis=0)
+            next_img = np.append(next_img,img_x,axis=0)
+            for c in range(channels):
+                next_img[:,:,:,c] -= np.amin(next_img[:,:,:,c])
+                next_img[:,:,:,c] /= np.amax(next_img[:,:,:,c])
+                if normalize:
+                    mean, std = np.mean(next_img[:,:,:,c]), np.std(next_img[:,:,:,c])
+                    next_img[:,:,:,c] = (next_img[:,:,:,c] - mean) / std
+                    next_img[:,:,:,c] = next_img[:,:,:,c] * normalization_parameters[1,c] + normalization_parameters[0,c]
+            img = np.append(img, next_img, axis=0)
+
+    # remove extracted data
+    if any(img_list):
+        remove_extracted_data(img_names, label_names)
+
+    # limit intensity range
+    img[img<0] = 0
+    img[img>1] = 1
+    img = np.uint8(img*255)
+
+    # number of channels must be three (reuse or cut off)
+    min_channels = min(3, channels)
+    img_rgb = np.empty((img.shape[:3] + (3,)), dtype=np.uint8)
+    for i in range(3):
+        img_rgb[...,i] = img[...,i % min_channels]
+
+    return img_rgb, label, normalization_parameters, channels
+
+def train_cropping(img, label, path_to_model, epochs, batch_size,
+                    validation_split, flip_x, flip_y, flip_z, rotate,
+                    img_val, label_val):
+
+    # img shape
+    zsh, ysh, xsh, channels = img.shape
+
+    # list of IDs
+    list_IDs_fg = list(np.where(label)[0])
+    list_IDs_bg = list(np.where(label==False)[0])
+
+    # validation data
+    if np.any(img_val):
+        list_IDs_val_fg = list(np.where(label_val)[0])
+        list_IDs_val_bg = list(np.where(label_val==False)[0])
+    elif validation_split:
+        split_fg = int(len(list_IDs_fg) * validation_split)
+        split_bg = int(len(list_IDs_bg) * validation_split)
+        list_IDs_val_fg = list_IDs_fg[split_fg:]
+        list_IDs_val_bg = list_IDs_bg[split_bg:]
+        list_IDs_fg = list_IDs_fg[:split_fg]
+        list_IDs_bg = list_IDs_bg[:split_bg]
+
+    # upsample IDs
+    max_IDs = max(len(list_IDs_fg), len(list_IDs_bg))
+    tmp_fg = []
+    while len(tmp_fg) < max_IDs:
+        tmp_fg.extend(list_IDs_fg)
+        tmp_fg = tmp_fg[:max_IDs]
+    list_IDs_fg = tmp_fg[:]
+
+    tmp_bg = []
+    while len(tmp_bg) < max_IDs:
+        tmp_bg.extend(list_IDs_bg)
+        tmp_bg = tmp_bg[:max_IDs]
+    list_IDs_bg = tmp_bg[:]
+
+    # validation data
+    if np.any(img_val) or validation_split:
+        max_IDs = max(len(list_IDs_val_fg), len(list_IDs_val_bg))
+        tmp_fg = []
+        while len(tmp_fg) < max_IDs:
+            tmp_fg.extend(list_IDs_val_fg)
+            tmp_fg = tmp_fg[:max_IDs]
+        list_IDs_val_fg = tmp_fg[:]
+        tmp_bg = []
+
+        while len(tmp_bg) < max_IDs:
+            tmp_bg.extend(list_IDs_val_bg)
+            tmp_bg = tmp_bg[:max_IDs]
+        list_IDs_val_bg = tmp_bg[:]
+
+    # input shape
+    input_shape = (ysh, xsh, channels)
+
+    # parameters
+    params = {'dim': (ysh, xsh),
+              'batch_size': batch_size,
+              'n_classes': 2,
+              'n_channels': channels,
+              'shuffle': True}
+
+    # validation parameters
+    params_val = {'dim': (ysh, xsh),
+                  'batch_size': batch_size,
+                  'n_classes': 2,
+                  'n_channels': channels,
+                  'shuffle': False}
+
+    # data generator
+    training_generator = DataGeneratorCrop(img, label, list_IDs_fg, list_IDs_bg, **params)
+    if np.any(img_val):
+        validation_generator = DataGeneratorCrop(img_val, label_val, list_IDs_val_fg, list_IDs_val_bg, **params_val)
+    elif validation_split:
+        validation_generator = DataGeneratorCrop(img, label, list_IDs_val_fg, list_IDs_val_bg, **params_val)
+    else:
+        validation_generator = None
+
+    # create a MirroredStrategy
+    cdo = tf.distribute.ReductionToOneDevice()
+    strategy = tf.distribute.MirroredStrategy(cross_device_ops=cdo)
+    print('Number of devices: {}'.format(strategy.num_replicas_in_sync))
+
+    # create callback
+    if np.any(img_val) or validation_split:
+        save_best_only = True
+    else:
+        save_best_only = False
+    checkpoint_cb = tf.keras.callbacks.ModelCheckpoint(str(path_to_model), save_best_only=save_best_only)
+
+    # compile model
+    with strategy.scope():
+        model = make_densenet(input_shape)
+        model.compile(loss= tf.keras.losses.BinaryCrossentropy(),
+                      optimizer= Adam(learning_rate=0.001),
+                      metrics=['accuracy'])
+    # train model
+    history = model.fit(training_generator,
+              validation_data=validation_generator,
+              epochs=max(1,epochs),
+              callbacks=checkpoint_cb)
+
+    # save results in figure on train end
+    if np.any(img_val) or validation_split:
+        save_history(history.history, path_to_model.replace(".h5","_cropping.h5"))
+
+    # compile model for finetunning
+    with strategy.scope():
+        model = load_model(str(path_to_model))
+        model.trainable = True
+        model.compile(loss= tf.keras.losses.BinaryCrossentropy(),
+                      optimizer= Adam(learning_rate=1e-5),
+                      metrics=['accuracy'])
+
+    # finetune model
+    history = model.fit(training_generator,
+              validation_data=validation_generator,
+              epochs=max(1,epochs),
+              callbacks=checkpoint_cb)
+
+    # save results in figure on train end
+    if np.any(img_val) or validation_split:
+        save_history(history.history, path_to_model.replace(".h5","_cropfine.h5"))
+
+def load_data_to_crop(path_to_img, channels, x_scale, y_scale, z_scale,
+                        normalize, normalization_parameters, img):
+    # read image data
+    if img is None:
+        img, _, _ = load_data(path_to_img, 'first_queue', return_extension=True)
+        InputError.img_names = [path_to_img]
+        InputError.label_names = []
+    img_data = np.copy(img, order='C')
+    if img is None:
+        InputError.message = "Invalid image data %s." %(os.path.basename(path_to_img))
+        raise InputError()
+    # handle all images having channels >=1
+    if len(img.shape)==3:
+        z_shape, y_shape, x_shape = img.shape
+        img = img.reshape(z_shape, y_shape, x_shape, 1)
+    if img.shape[3] != channels:
+        InputError.message = f'Number of channels must be {channels}.'
+        raise InputError()
+    z_shape, y_shape, x_shape, _ = img.shape
+    img = img.astype(np.float32)
+    img_z = img_resize(img, z_shape, y_scale, x_scale)
+    img_y = np.swapaxes(img_resize(img,z_scale,y_shape,x_scale),0,1)
+    img_x = np.swapaxes(img_resize(img,z_scale,y_scale,x_shape),0,2)
+    img = np.append(img_z,img_y,axis=0)
+    img = np.append(img,img_x,axis=0)
+    for c in range(channels):
+        img[:,:,:,c] -= np.amin(img[:,:,:,c])
+        img[:,:,:,c] /= np.amax(img[:,:,:,c])
+        if normalize:
+            mean, std = np.mean(img[:,:,:,c]), np.std(img[:,:,:,c])
+            img[:,:,:,c] = (img[:,:,:,c] - mean) / std
+            img[:,:,:,c] = img[:,:,:,c] * normalization_parameters[1,c] + normalization_parameters[0,c]
+    img[img<0] = 0
+    img[img>1] = 1
+    img = np.uint8(img*255)
+
+    # number of channels must be three (reuse or cut off)
+    channels = min(3, channels)
+    img_rgb = np.empty((img.shape[:3] + (3,)), dtype=np.uint8)
+    for i in range(3):
+        img_rgb[...,i] = img[...,i % channels]
+    return img_rgb, z_shape, y_shape, x_shape, img_data
+
+def crop_volume(img, path_to_model, path_to_final, z_shape, y_shape, x_shape, batch_size,
+        debug_cropping, save_cropped, img_data, x_range, y_range, z_range, x_puffer=25, y_puffer=25, z_puffer=25):
+
+    # img shape
+    zsh, ysh, xsh, channels = img.shape
+
+    # list of IDs
+    list_IDs = [x for x in range(zsh)]
+
+    # make length of list divisible by batch size
+    rest = batch_size - (len(list_IDs) % batch_size)
+    list_IDs = list_IDs + list_IDs[:rest]
+
+    # parameters
+    params = {'dim': (ysh,xsh),
+              'dim_img': (zsh, ysh, xsh),
+              'batch_size': batch_size,
+              'n_channels': channels}
+
+    # data generator
+    predict_generator = PredictDataGeneratorCrop(img, list_IDs, **params)
+
+    # input shape
+    input_shape = (ysh, xsh, channels)
+
+    # load model
+    model = make_densenet(input_shape)
+
+    # load weights
+    hf = h5py.File(path_to_model, 'r')
+    cropping_weights = hf.get('cropping_weights')
+    iterator = 0
+    for layer in model.layers:
+        if layer.get_weights() != []:
+            new_weights = []
+            for arr in layer.get_weights():
+                new_weights.append(cropping_weights.get(str(iterator)))
+                iterator += 1
+            layer.set_weights(new_weights)
+    hf.close()
+
+    # predict
+    probabilities = model.predict(predict_generator, verbose=0, steps=None)
+    probabilities = probabilities[:zsh]
+    probabilities = np.ravel(probabilities)
+
+    # plot prediction
+    if debug_cropping and path_to_final:
+        import matplotlib.pyplot as plt
+        import matplotlib
+        x = range(len(probabilities))
+        y = list(probabilities)
+        plt.plot(x, y)
+
+    # create mask
+    probabilities[probabilities > 0.5] = 1
+    probabilities[probabilities <= 0.5] = 0
+
+    # remove outliers
+    for k in range(4,zsh-4):
+        if np.all(probabilities[k-1:k+2] == np.array([0,1,0])):
+            probabilities[k-1:k+2] = 0
+        elif np.all(probabilities[k-2:k+2] == np.array([0,1,1,0])):
+            probabilities[k-2:k+2] = 0
+        elif np.all(probabilities[k-2:k+3] == np.array([0,1,1,1,0])):
+            probabilities[k-2:k+3] = 0
+        elif np.all(probabilities[k-3:k+3] == np.array([0,1,1,1,1,0])):
+            probabilities[k-3:k+3] = 0
+        elif np.all(probabilities[k-3:k+4] == np.array([0,1,1,1,1,1,0])):
+            probabilities[k-3:k+4] = 0
+        elif np.all(probabilities[k-4:k+4] == np.array([0,1,1,1,1,1,1,0])):
+            probabilities[k-4:k+4] = 0
+        elif np.all(probabilities[k-4:k+5] == np.array([0,1,1,1,1,1,1,1,0])):
+            probabilities[k-4:k+5] = 0
+
+    # create final
+    if z_range is not None:
+        z_lower, z_upper = z_range
+    else:
+        z_lower = max(0,np.argmax(probabilities[:z_shape]) - z_puffer)
+        z_upper = min(z_shape,z_shape - np.argmax(np.flip(probabilities[:z_shape])) + z_puffer +1)
+
+    if y_range is not None:
+        y_lower, y_upper = y_range
+    else:
+        y_lower = max(0,np.argmax(probabilities[z_shape:z_shape+y_shape]) - y_puffer)
+        y_upper = min(y_shape,y_shape - np.argmax(np.flip(probabilities[z_shape:z_shape+y_shape])) + y_puffer +1)
+
+    if x_range is not None:
+        x_lower, x_upper = x_range
+    else:
+        x_lower = max(0,np.argmax(probabilities[z_shape+y_shape:]) - x_puffer)
+        x_upper = min(x_shape,x_shape - np.argmax(np.flip(probabilities[z_shape+y_shape:])) + x_puffer +1)
+
+    # plot result
+    if debug_cropping and path_to_final:
+        y = np.zeros_like(probabilities)
+        y[z_lower:z_upper] = 1
+        y[z_shape+y_lower:z_shape+y_upper] = 1
+        y[z_shape+y_shape+x_lower:z_shape+y_shape+x_upper] = 1
+        plt.plot(x, y, '--')
+        plt.tight_layout()  # To prevent overlapping of subplots
+        #matplotlib.use("GTK3Agg")
+        plt.savefig(path_to_final.replace('.tif','.png'), dpi=300)
+
+    # crop image data
+    cropped_volume = img_data[z_lower:z_upper, y_lower:y_upper, x_lower:x_upper]
+    if save_cropped and path_to_final:
+        save_data(path_to_final, cropped_volume, compress=False)
+
+    return z_lower, z_upper, y_lower, y_upper, x_lower, x_upper, cropped_volume
+
+#=====================
+# main functions
+#=====================
+
+def load_and_train(normalize,path_to_img,path_to_labels,path_to_model,
+                epochs,batch_size,validation_split,
+                flip_x,flip_y,flip_z,rotate,labels_to_compute,labels_to_remove,
+                path_val_img=[None],path_val_labels=[None],
+                img=None, label=None, img_val=None, label_val=None,
+                x_scale=256, y_scale=256, z_scale=256):
+
+    # load training data
+    img, label, normalization_parameters, channels = load_cropping_training_data(normalize,
+                        path_to_img, path_to_labels, x_scale, y_scale, z_scale, labels_to_compute, labels_to_remove,
+                        img, label)
+
+    # load validation data
+    if any(path_val_img) or img_val is not None:
+        img_val, label_val, _, _ = load_cropping_training_data(normalize,
+                            path_val_img, path_val_labels, x_scale, y_scale, z_scale,
+                            labels_to_compute, labels_to_remove,
+                            img_val, label_val, normalization_parameters, channels)
+
+    # train cropping
+    train_cropping(img, label, path_to_model, epochs,
+                    batch_size, validation_split,
+                    flip_x, flip_y, flip_z, rotate,
+                    img_val, label_val)
+
+    # load weights
+    model = load_model(str(path_to_model))
+    cropping_weights = []
+    for layer in model.layers:
+        if layer.get_weights() != []:
+            for arr in layer.get_weights():
+                cropping_weights.append(arr)
+
+    # configuration data
+    cropping_config = np.array([channels, x_scale, y_scale, z_scale, normalize, 0, 1])
+
+    return cropping_weights, cropping_config, normalization_parameters
+
+def crop_data(path_to_data, path_to_model, path_to_cropped_image, batch_size,
+    debug_cropping=False, save_cropped=True, img_data=None,
+    x_range=None, y_range=None, z_range=None):
+
+    # get meta data
+    hf = h5py.File(path_to_model, 'r')
+    meta = hf.get('cropping_meta')
+    configuration = meta.get('configuration')
+    channels, x_scale, y_scale, z_scale, normalize, mu, sig = np.array(configuration)[:]
+    channels, x_scale, y_scale, z_scale, normalize, mu, sig = int(channels), int(x_scale), \
+                            int(y_scale), int(z_scale), int(normalize), float(mu), float(sig)
+    if '/cropping_meta/normalization' in hf:
+        normalization_parameters = np.array(meta.get('normalization'), dtype=float)
+    else:
+        # old configuration
+        normalization_parameters = np.array([[mu],[sig]])
+        channels = 1
+    hf.close()
+
+    # load data
+    img, z_shape, y_shape, x_shape, img_data = load_data_to_crop(path_to_data, channels,
+                    x_scale, y_scale, z_scale, normalize, normalization_parameters, img_data)
+
+    # make prediction
+    z_lower, z_upper, y_lower, y_upper, x_lower, x_upper, cropped_volume = crop_volume(img, path_to_model,
+        path_to_cropped_image, z_shape, y_shape, x_shape, batch_size, debug_cropping, save_cropped, img_data,
+        x_range, y_range, z_range)
+
+    # region of interest
+    region_of_interest = np.array([z_lower, z_upper, y_lower, y_upper, x_lower, x_upper])
+
+    return region_of_interest, cropped_volume
+