3DCNN & MRIs

Steps

• Context (python, nibabel, Tensorflow)
• • Data exploration.
  • Preprocessing Input.
  • Model.
  • Training.
• Results.
• Q&A

Load data

import numpy as np
import pandas as pd
data = np.load("voxels_tagged.npy")
data.shape

[1108, 125001]

Pre-processing

(Separate classes)

class1 = frame[(frame['class']== 1)]
print("Total class 1:", len(class1))
class2 = frame[(frame['class']== 0)]
print("Total class 0:", len(class2))

Total class 1: 550

Total class 0: 558

Average brains

from sklearn.preprocessing import normalize
average_brain1 = normalize(class1).mean(0)
average_brain2 = normalize(class2).mean(0)
avg_vol1 = average_brain1.reshape(50,50,50)
print(avg_vol1.shape)

avg_vol2 = average_brain2.reshape(50,50,50)
print(avg_vol2.shape)

(50, 50, 50)

(50, 50, 50)

Comparing images

def compare_images(img1, img2):
    diff = img1 - img2  # elementwise for scipy arrays
    m_norm = sum(abs(diff))  # Manhattan norm
    z_norm = norm(diff.ravel(), 0)  # Zero norm
    return (m_norm, z_norm, diff)