Cardiac_Condition_Classifier.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Cardiac Condition Classifier.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "2EYP7mn7dRaq"
      },
      "source": [
        "#download PTB diagnostic ECG Database\r\n",
        "#479\r\n",
        "#https://www.physionet.org/content/ptbdb/1.0.0/\r\n",
        "!wget -r -N -c -np https://physionet.org/files/ptbdb/1.0.0/"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0NuB5J4vem3A"
      },
      "source": [
        "#download the wfdb libraries and associated databases\r\n",
        "#https://www.physionet.org/content/ptbdb/1.0.0/\r\n",
        "from IPython.display import display\r\n",
        "import matplotlib.pyplot as plt\r\n",
        "%matplotlib inline\r\n",
        "import numpy as np\r\n",
        "import os\r\n",
        "import shutil\r\n",
        "import posixpath\r\n",
        "\r\n",
        "!pip install wfdb\r\n",
        "import wfdb\r\n",
        "from wfdb import processing\r\n",
        "from PIL import Image "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yxvRBsgKku0W"
      },
      "source": [
        "#create new folder with only Patient_Samples\r\n",
        "!mkdir /content/Patient_Samples\r\n",
        "!mv /content/physionet.org/files/ptbdb/1.0.0/patient* /content/Patient_Samples/"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wfZFHeYklxa_"
      },
      "source": [
        "#get reason for admission and signal information for 1 second for the first channel and store in a numPy array\r\n",
        "patient_path = sorted(os.listdir('/content/Patient_Samples/'))\r\n",
        "\r\n",
        "#select channels, sampfrom, sampto parameters for training and testing\r\n",
        "samples = [] #determine the order\r\n",
        "classifications = [] #determine the classifiers\r\n",
        "recordings = [] #collect the recordings\r\n",
        "rec = [] #record for wfd library use\r\n",
        "signal_im = [] #store signals\r\n",
        "\r\n",
        "#obtain determine the relevant header files (hea)\r\n",
        "#do not delete the other associated files .dat, .html, .xyz, .hea\r\n",
        "for patient in patient_path:\r\n",
        "  file_path = sorted(os.listdir('/content/Patient_Samples/' + patient))\r\n",
        "  for filename in file_path:\r\n",
        "    if filename.endswith('.hea'):\r\n",
        "      #samples contains the actual file name in order\r\n",
        "      samples.append(filename) #list which stores the actual filename\r\n",
        "      #parameters can be modified\r\n",
        "      signals, fields = wfdb.rdsamp('/content/Patient_Samples/'+patient+'/'+filename[:-4], channels=[4], sampfrom=6000, sampto=10000)\r\n",
        "      #need to ensure the parameters are the same\r\n",
        "      record = wfdb.rdrecord('/content/Patient_Samples/'+patient+'/'+filename[:-4], channels=[4], sampfrom=6000, sampto=10000) \r\n",
        "      #retrieving reason for admission\r\n",
        "      diag = fields.get('comments')\r\n",
        "      admit = diag[4]\r\n",
        "      split_string = admit.split(': ', 1)\r\n",
        "      classifications.append(split_string[1]) #list which stores the classifications\r\n",
        "      recordings.append(record.p_signal[:,0]) #list of the recordings\r\n",
        "      rec.append(record) #store the whole recording for wfdb plotting use\r\n",
        "      \r\n",
        "      #image generation\r\n",
        "      plt.clf()\r\n",
        "      plt.plot(record.p_signal[:,0])\r\n",
        "      plt.axis('off')\r\n",
        "      figure = plt.gcf()\r\n",
        "      figure.set_size_inches(4, 2)\r\n",
        "      plt.savefig(\"/content/sample.png\", dpi=100, pad_inches=0, bbox_inches='tight')\r\n",
        "\r\n",
        "      #store images in a list of numpy arrays\r\n",
        "      img = Image.open('/content/sample.png') \r\n",
        "      numpydata = np.array(img)\r\n",
        "      signal_im.append(numpydata) "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oDtGOQm2Q1y6"
      },
      "source": [
        "#see classifications and signals for dataset specfied in range\r\n",
        "for n in range(400,401):\r\n",
        "  print(classifications[n])\r\n",
        "  print(samples[n])\r\n",
        "  print(recordings[n])\r\n",
        "  plt.imshow(signal_im[n])\r\n",
        "  wfdb.plot_wfdb(rec[n])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SIEE8ceA_pSs"
      },
      "source": [
        "#split the data by myocardial infarction and healthy control\r\n",
        "myocardial_class = []\r\n",
        "myocardial_record = []\r\n",
        "myocardial_image = []\r\n",
        "\r\n",
        "healthy_class = []\r\n",
        "healthy_record = []\r\n",
        "healthy_image = []\r\n",
        "\r\n",
        "for i in range(len(classifications)):\r\n",
        "\r\n",
        "  #myocardial --> 0\r\n",
        "  if classifications[i] == 'Myocardial infarction':\r\n",
        "    myocardial_class.append(0)\r\n",
        "    myocardial_record.append(recordings[i])\r\n",
        "    myocardial_image.append(signal_im[i])\r\n",
        "\r\n",
        "  #healthy --> 1\r\n",
        "  if classifications[i] == 'Healthy control':\r\n",
        "    healthy_class.append(1)\r\n",
        "    healthy_record.append(recordings[i])\r\n",
        "    healthy_image.append(signal_im[i])\r\n",
        "\r\n",
        "class_names = ['Myocardial infarction', 'Healthy']"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nXIRLeaOJNBQ"
      },
      "source": [
        "#convert from list to numpy array of #im, row, col\r\n",
        "#images\r\n",
        "myocardial_class = np.array(myocardial_class)\r\n",
        "myocardial_imstack = np.stack(myocardial_image, axis=0)\r\n",
        "\r\n",
        "healthy_class = np.array(healthy_class)\r\n",
        "healthy_imstack = np.stack(healthy_image, axis=0)\r\n",
        "\r\n",
        "#seperate the training and testing data for images\r\n",
        "\r\n",
        "#train\r\n",
        "train_class = np.concatenate((myocardial_class[:64], healthy_class[:64]), axis = 0)\r\n",
        "train_image = np.concatenate((myocardial_imstack[:64,:,:,:], healthy_imstack[:64,:,:,:]), axis = 0)\r\n",
        "\r\n",
        "#test\r\n",
        "test_class = np.concatenate((myocardial_class[64:80], healthy_class[64:80]), axis = 0)\r\n",
        "test_image = np.concatenate((myocardial_imstack[64:80,:,:,:], healthy_imstack[64:80,:,:,:]), axis = 0)"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rgYwe3OIoTWy"
      },
      "source": [
        "#importing tensorflow and libraries\r\n",
        "%tensorflow_version 2.x\r\n",
        "\r\n",
        "import tensorflow as tf\r\n",
        "from tensorflow import keras\r\n",
        "from tensorflow.keras import datasets, layers, models\r\n",
        "from tensorflow.keras.models import Sequential\r\n",
        "\r\n",
        "#helper libraries\r\n",
        "import numpy as np\r\n",
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "B-uV-Ispsf3z"
      },
      "source": [
        "#data augmentation to avoid overfitting\r\n",
        "#dropout may lead to inconsistency\r\n",
        "\r\n",
        "data_augmentation = keras.Sequential(\r\n",
        "  [\r\n",
        "    layers.experimental.preprocessing.RandomFlip(\"horizontal\", \r\n",
        "                                                 input_shape=np.shape(train_image[1]))\r\n",
        "  ]\r\n",
        ")"
      ],
      "execution_count": 29,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Hfftn2pnbNKq"
      },
      "source": [
        "#https://www.tensorflow.org/tutorials/images/cnn\r\n",
        "#define the neural network for image classification\r\n",
        "num_classes = 2\r\n",
        "\r\n",
        "model = Sequential([\r\n",
        "  data_augmentation,\r\n",
        "  #layers.Conv2D(32, (3, 3), activation='relu', input_shape=np.shape(train_image[1])),\r\n",
        "  layers.experimental.preprocessing.Rescaling(1./255),\r\n",
        "  layers.Conv2D(16, 3, padding='same', activation='relu'),\r\n",
        "  layers.MaxPooling2D(),\r\n",
        "  layers.Conv2D(32, 3, padding='same', activation='relu'),\r\n",
        "  layers.MaxPooling2D(),\r\n",
        "  layers.Conv2D(64, 3, padding='same', activation='relu'),\r\n",
        "  layers.MaxPooling2D(),\r\n",
        "  layers.Dropout(0.2),\r\n",
        "  layers.Flatten(),\r\n",
        "  layers.Dense(128, activation='relu'),\r\n",
        "  layers.Dense(num_classes)\r\n",
        "])"
      ],
      "execution_count": 40,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Jw7g3o6R6vUg"
      },
      "source": [
        "#clear the model\r\n",
        "keras.backend.clear_session()"
      ],
      "execution_count": 39,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GLrL-6ip3V_7",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "71f76772-ca01-4d0e-da2c-90d810ff6c9f"
      },
      "source": [
        "#summary of the model architecture\r\n",
        "#https://www.tensorflow.org/tutorials/images/cnn\r\n",
        "model.summary()"
      ],
      "execution_count": 41,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "sequential (Sequential)      (None, 151, 310, 4)       0         \n",
            "_________________________________________________________________\n",
            "rescaling (Rescaling)        (None, 151, 310, 4)       0         \n",
            "_________________________________________________________________\n",
            "conv2d (Conv2D)              (None, 151, 310, 16)      592       \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 75, 155, 16)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 75, 155, 32)       4640      \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 37, 77, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 37, 77, 64)        18496     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 18, 38, 64)        0         \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 18, 38, 64)        0         \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 43776)             0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 128)               5603456   \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 2)                 258       \n",
            "=================================================================\n",
            "Total params: 5,627,442\n",
            "Trainable params: 5,627,442\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bgU1hD-2pPDE"
      },
      "source": [
        "#compile the model\r\n",
        "model.compile(optimizer='adam',\r\n",
        "              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),\r\n",
        "              metrics=['accuracy'])"
      ],
      "execution_count": 42,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3QeYzY-I01_T",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "cc3041aa-d7f2-4a5a-bf67-fb70047343ff"
      },
      "source": [
        "#image classification training, can manipulate epochs\r\n",
        "epochs = 10\r\n",
        "history = model.fit(train_image, train_class, epochs=epochs, \r\n",
        "                    validation_data=(test_image, test_class), shuffle = True)"
      ],
      "execution_count": 43,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/10\n",
            "4/4 [==============================] - 6s 2s/step - loss: 2.6325 - accuracy: 0.5031 - val_loss: 1.3506 - val_accuracy: 0.5000\n",
            "Epoch 2/10\n",
            "4/4 [==============================] - 6s 1s/step - loss: 0.8504 - accuracy: 0.6052 - val_loss: 0.6941 - val_accuracy: 0.5000\n",
            "Epoch 3/10\n",
            "4/4 [==============================] - 6s 1s/step - loss: 0.7210 - accuracy: 0.4729 - val_loss: 0.6909 - val_accuracy: 0.5000\n",
            "Epoch 4/10\n",
            "4/4 [==============================] - 6s 1s/step - loss: 0.6794 - accuracy: 0.5437 - val_loss: 0.6512 - val_accuracy: 0.6875\n",
            "Epoch 5/10\n",
            "4/4 [==============================] - 6s 1s/step - loss: 0.6517 - accuracy: 0.6417 - val_loss: 0.6284 - val_accuracy: 0.6250\n",
            "Epoch 6/10\n",
            "4/4 [==============================] - 6s 1s/step - loss: 0.6165 - accuracy: 0.6729 - val_loss: 0.5969 - val_accuracy: 0.6875\n",
            "Epoch 7/10\n",
            "4/4 [==============================] - 6s 1s/step - loss: 0.6084 - accuracy: 0.6552 - val_loss: 0.5612 - val_accuracy: 0.6875\n",
            "Epoch 8/10\n",
            "4/4 [==============================] - 6s 1s/step - loss: 0.5740 - accuracy: 0.7167 - val_loss: 0.5990 - val_accuracy: 0.6875\n",
            "Epoch 9/10\n",
            "4/4 [==============================] - 6s 1s/step - loss: 0.5343 - accuracy: 0.7281 - val_loss: 0.5731 - val_accuracy: 0.6250\n",
            "Epoch 10/10\n",
            "4/4 [==============================] - 6s 1s/step - loss: 0.4974 - accuracy: 0.7490 - val_loss: 0.5843 - val_accuracy: 0.6562\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_IqFGCQb1JjJ",
        "outputId": "3cedff2d-b9ae-4873-9d1a-1f6cce5bb825"
      },
      "source": [
        "#evaluating the model for image classification\r\n",
        "test_loss, test_acc = model.evaluate(test_image,  test_class, verbose=2) \r\n",
        "print('Test accuracy:', test_acc)"
      ],
      "execution_count": 44,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1/1 - 0s - loss: 0.5843 - accuracy: 0.6562\n",
            "Test accuracy: 0.65625\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 499
        },
        "id": "Vp2rP5f0EG78",
        "outputId": "f058b21e-23f8-487b-87ee-237b5514d6b9"
      },
      "source": [
        "#evaluating the model for image classification\r\n",
        "acc = history.history['accuracy']\r\n",
        "val_acc = history.history['val_accuracy']\r\n",
        "\r\n",
        "loss = history.history['loss']\r\n",
        "val_loss = history.history['val_loss']\r\n",
        "\r\n",
        "epochs_range = range(epochs)\r\n",
        "\r\n",
        "plt.figure(figsize=(8, 8))\r\n",
        "plt.subplot(1, 2, 1)\r\n",
        "plt.plot(epochs_range, acc, label='Training Accuracy')\r\n",
        "plt.plot(epochs_range, val_acc, label='Validation Accuracy')\r\n",
        "plt.legend(loc='lower right')\r\n",
        "plt.title('Training and Validation Accuracy')\r\n",
        "\r\n",
        "plt.subplot(1, 2, 2)\r\n",
        "plt.plot(epochs_range, loss, label='Training Loss')\r\n",
        "plt.plot(epochs_range, val_loss, label='Validation Loss')\r\n",
        "plt.legend(loc='upper right')\r\n",
        "plt.title('Training and Validation Loss')\r\n",
        "plt.show()"
      ],
      "execution_count": 46,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x576 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QSuwi8yw1Ysf"
      },
      "source": [
        "#the prediction vs the actual for the test case for image classification\r\n",
        "for n in range(0,32):\r\n",
        "  print(n)\r\n",
        "  predictions = model.predict(test_image)\r\n",
        "  print('Neural Network Prediction: '+ class_names[np.argmax(predictions[n])])\r\n",
        "  if test_class[n] == 0:\r\n",
        "    print('Actual Condition: Myocardial infarction')\r\n",
        "  if test_class[n] == 1:\r\n",
        "    print('Actual Condition: Healthy')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LWk-4Ch_9MO6"
      },
      "source": [
        "#10 epochs with shuffle on seems to be good got 75 and then 85 percent\r\n",
        "#15 epochs with shuffle on seems to be good got 81 percent\r\n",
        "#removing data augmentation got 78% and 71%\r\n",
        "\r\n",
        "#models being generated are inconsistent"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}