Introduction:

This project is done under Deep Learning Course 2021 at Ariel University.

Pneumonia is a life threatening disease, which occurs in the lungs caused by either bacterial or viral infection. It can be life endangering if not acted upon in the right time and thus early diagnosis of pneumonia is vital. The aim of this paper is to automatically detect bacterial and viral pneumonia using the digital x-ray images.

Because we are limited in our lost power and the amount of dataset we have we had to overcome it by using the principles of the different architectures of different networks, and allow with a small dataset a wide enough range to achieve good results with these limitations (Data Augmentation).

Data:

Using the data from the site: https://www.kaggle.com/paultimothymooney/chest-xray-pneumonia
The data set is organized into three folders (train, test, selection) and contains subfolders for each image category (pneumonia / normal). There are 5,856 X-ray (JPEG) images and 2 categories (Pneumonia / Normal).

After we downloaded the data set we changed the amount of images in each folder so that we got a new structure.

Since we have relatively few images and in order for our model to learn as much as possible from the training data,
we used data augmentation as shown in the code below:

 # Data generation objects
    train_datagen = ImageDataGenerator(rescale=1. / 255, zoom_range=0.3, shear_range=0.2, vertical_flip=True)
    val_datagen = ImageDataGenerator(rescale=1. / 255)

    train_gen = train_datagen.flow_from_directory(
        directory=input_path + 'train',
        target_size=(img_dims, img_dims),
        batch_size=batch_size,
        class_mode='binary',
        shuffle=True)

    val_gen = val_datagen.flow_from_directory(
        directory=input_path + 'val',
        target_size=(img_dims, img_dims),
        batch_size=batch_size,
        class_mode='binary',
        shuffle=True)

And for the prediction we created the test manually without data augmentation:

 test_data = []
    test_labels = []
    label = 0
    for cond in ['/NORMAL/', '/PNEUMONIA/']:
        for img in tqdm(os.listdir(input_path + 'test' + cond)):
            img = plt.imread(input_path + 'test' + cond + img)
            img = cv2.resize(img, (img_dims, img_dims))
            if len(img.shape) == 1:
                img = np.dstack([img, img, img])
            img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
            img = img.astype(np.float32) / 255
            if cond == '/NORMAL/':
                label = 0
            elif cond == '/PNEUMONIA/':
                label = 1
            test_data.append(img)
            test_labels.append(label)

    test_data = np.array(test_data)
    test_labels = np.array(test_labels)

Model:


model = models.Sequential()

model.add(layers.Conv2D(16,(3,3),activation = 'relu', padding='same',input_shape=(img_dims,img_dims,3)))
model.add(layers.MaxPooling2D((2,2)))

model.add(layers.Conv2D(32,(3,3),activation = 'relu', padding='same'))
model.add(layers.BatchNormalization())
model.add(layers.MaxPooling2D((2,2)))

model.add(layers.Conv2D(64,(3,3),activation = 'relu', padding='same'))
model.add(layers.BatchNormalization())
model.add(layers.MaxPooling2D((2,2)))

model.add(layers.Conv2D(128,(3,3),activation = 'relu', padding='same'))
model.add(layers.BatchNormalization())
model.add(layers.MaxPooling2D((2,2)))

model.add(layers.Conv2D(256,(3,3),activation = 'relu', padding='same'))
model.add(layers.BatchNormalization())
model.add(layers.MaxPooling2D((2,2)))


model.add(layers.Flatten())
model.add(layers.Dense(units=128,activation='relu'))
model.add(layers.Dropout(0.2))
model.add(layers.Dense(units=1, activation='sigmoid'))

result:

Given that our processing power was very limited,
in addition we were limited in the amount of our data, the results were quite good..

We wanted to predict the test on our model and these are the results:

What you see below each picture is how close he was able to predict whether the photograph is with or without pneumonia.
And what's inside the parentheses are the real labels of the image that is,
if the image with (0) actually means that the percentages next to it tell us how close it is that it detected that the image is without pneumonia.

davidhct / pneumonia-detection-with-cnn Goto Github PK

pneumonia-detection-with-cnn's Introduction

Introduction:

This project is done under Deep Learning Course 2021 at Ariel University.

Data:

Since we have relatively few images and in order for our model to learn as much as possible from the training data,
we used data augmentation as shown in the code below:

And for the prediction we created the test manually without data augmentation:

Model:

result:

We wanted to predict the test on our model and these are the results:

Feature Maps and Filters:

Feature Maps:

Filters:

pneumonia-detection-with-cnn's People

Contributors

Watchers

Recommend Projects

React

Vue.js

Typescript

TensorFlow

Django

Laravel

D3

Recommend Topics

javascript

web

server

Machine learning

Visualization

Game

Recommend Org

Facebook

Microsoft

Google

Alibaba

D3

Tencent

davidhct / pneumonia-detection-with-cnn Goto Github PK

pneumonia-detection-with-cnn's Introduction

Introduction:

This project is done under Deep Learning Course 2021 at Ariel University.

Data:

Since we have relatively few images and in order for our model to learn as much as possible from the training data, we used data augmentation as shown in the code below:

And for the prediction we created the test manually without data augmentation:

Model:

result:

We wanted to predict the test on our model and these are the results:

Feature Maps and Filters:

Feature Maps:

Filters:

pneumonia-detection-with-cnn's People

Contributors

Watchers

Recommend Projects

Recommend Topics

Recommend Org

Since we have relatively few images and in order for our model to learn as much as possible from the training data,
we used data augmentation as shown in the code below: