oncogen.py

# -*- coding: utf-8 -*-
"""
Created on Fri Jul 19 17:44:53 2013

@author: КТ2
"""

import numpy as np
import matplotlib.pyplot as plt
import matplotlib.colors as mcol
from scipy import signal
from scipy.ndimage import filters

roi = np.zeros([100, 100])


class Tumor():
    def __init__(self, matrixsize, tum_size, tumorosity):
        if matrixsize % 2 == 0:
            raise ValueError()
        self.matrix = np.zeros([matrixsize, matrixsize])
        self.conturs = np.zeros([matrixsize, matrixsize])
        self.ms = matrixsize
        self.distances = self.matrix
        self.grow(np.array([matrixsize / 2, matrixsize / 2]), tum_size)

        for i in range(tumorosity):
            self.new_one(tum_size)

    def grow(self, startpoint, radius):
        x, y = startpoint

        log_mrx = np.ogrid[0:self.ms, 0:self.ms] - startpoint

        euclid_dist = np.sqrt(log_mrx[0] ** 2 + log_mrx[1] ** 2)
        self.matrix[euclid_dist < radius] = 1

    def new_one(self, maxsize=10):
        x, y = [np.random.randint(0, self.ms), np.random.randint(0, self.ms)]
        if self.matrix[x, y] == 1:
            self.grow(np.array([x, y]), np.random.randint(1, maxsize))
        else:
            self.new_one()

    def show_tumor(self, data1, data2):
        p1 = plt.subplot(121)
        p1.imshow(data1 / np.average(data1), cmap='gray')
        # p1.colorbar()
        p2 = plt.subplot(122)
        p2.imshow(data2 / np.average(data2), cmap='gray')
        #p2.colorbar()

        plt.show()

    def make_distanses(self):
        self.matrix_1 = self.matrix

        for i in range(50):
            self.matrix_1 = filters.generic_filter(self.matrix_1, self.erode, 3)
            self.distances += self.matrix_1
            print i
        # self.conturs[self.distances==1 OR self.distances==2 OR self.distances==3 OR self.distances==12]=1

        self.conturs[self.conturs == 0] = np.nan

    def make_perfusion(self):
        dist = np.copy(self.distances)

        self.perf_map = 2 / (1 + np.exp(dist + np.random.normal(0, 20, dist.shape)) ** 0.035)

        # self.perf_map/=np.average(self.perf_map)

    def erode(self, data):
        d = data[5]
        if d == 1 and 0 in data:
            return 0
        else:
            return d


a = Tumor(101, 20, 50)
tumors = []

cm = plt.get_cmap('bwr')
# cm.set_bad(color = 'k', alpha = 1)

for i in range(2):
    a = Tumor(151, np.random.randint(5, 40), 40)
    a.make_distanses()
    a.make_perfusion()
    p = plt.subplot(5, 5, i + 1)
    print np.average(a.perf_map), np.min(a.perf_map), np.max(a.perf_map)

    p.imshow(filters.gaussian_filter(a.perf_map, 2), alpha=1, cmap='gist_ncar', norm=mcol.Normalize(vmin=0.5, vmax=1.3))
    p.imshow(a.conturs, alpha=1, cmap=cm)

    plt.axis('off')

plt.show()