microscopy-visualization-tools/De_convolution_microscope_animation.py at main · isvecova/microscopy-visualization-tools · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.widgets import Button
from matplotlib.widgets import Slider, RadioButtons
from scipy.signal import convolve2d
from skimage.restoration import wiener, unsupervised_wiener, richardson_lucy
from numpy.fft import rfft2, irfft2
from scipy.ndimage import binary_dilation
from skimage.draw import disk
import matplotlib.gridspec as gridspec

class InteractiveImageProcessor:
    def __init__(self,ax_drawing,ax_original,ax_psf,ax_noise,ax_blurred,ax_deconvolved,btn,s_pointsize,s_spread,s_noise,s_balance, point_size=5, image_size=(100, 100)):
        self.ax = ax_drawing
        self.image_size = image_size
        self.point_size = point_size
        self.image = np.zeros(image_size)  # Create an empty image
        self.points = []
        self.cid = ax_drawing.figure.canvas.mpl_connect('button_press_event', self)
        self.scatter = ax_drawing.scatter([], [], s=100)  # Visual representation of points
        self.plot_created = False
        self.plot_figure = []

        s_pointsize.on_changed(self.update_image_based_on_radius)
        # Call update function on slider value change
        s_spread.on_changed(self.process_image)
        s_balance.on_changed(self.process_image)
        s_noise.on_changed(self.process_image)
        btn.on_clicked(self.process_image)

        # Display initial images
        self.img_original = ax_original.imshow(self.image, cmap='gray')
        self.img_psf = ax_psf.imshow(self.image, cmap='gray')  # Placeholder for PSF
        self.img_noise = ax_noise.imshow(self.image, cmap='gray')  # Placeholder for Noise
        self.img_blurred = ax_blurred.imshow(self.image, cmap='gray')  # Placeholder for Blurred and Noisy Image
        self.img_deconvolved = ax_deconvolved.imshow(self.image, cmap='gray')  # Placeholder for Deconvolved Image


    def __call__(self, event):
        if event.inaxes != self.ax:
            return
        x, y = int(event.xdata), int(event.ydata)
        self.points.append((x, y))
        print(len(self.points))
        self.update_image_based_on_radius(s_pointsize.val)

    def update_image_based_on_radius(self, val):
        # Clear image
        self.image.fill(0)
        # Update image with new radius for each point
        for x, y in self.points:
            rr, cc = disk((y, x), radius=val, shape=self.image.shape)
            self.image[rr, cc] = 1
        self.image = np.flipud(self.image)
        self.update_plot()

    def update_plot(self):
        self.scatter.set_offsets(self.points)
        self.scatter.set_sizes([np.pi * s_pointsize.val**2] * len(self.points))
        self.ax.figure.canvas.draw_idle()

        self.img_original = ax_original.imshow(self.image,cmap='gray')
        self.process_image([])

    def process_image(self, event):

        print('processing image')

        spread = s_spread.val
        balance = s_balance.val
        noise = s_noise.val
        method = btn.value_selected

        x = np.linspace(-10, 10, 100)
        y = np.linspace(-10, 10, 100)
        x, y = np.meshgrid(x, y)
        psf = np.exp(-(x**2 + y**2) / spread)
        psf /= psf.sum()  # Normalize PSF so that it sums to 1
        blurred = convolve2d(self.image, psf, mode='same')
        noise_component = noise * np.random.normal(size=blurred.shape)
        noisy_blurred = blurred + noise_component
        if method == 'Wiener':
            deconvolved = wiener(noisy_blurred, psf, balance)
        elif method == 'Inverse':
            f_psf = rfft2(psf, s=noisy_blurred.shape)
            f_blurred = rfft2(self.image)*f_psf
            f_noise = rfft2(noise_component)
            with np.errstate(divide='ignore', invalid='ignore'):
                f_deconvolved =  f_blurred / f_psf + f_noise / f_psf
            deconvolved = irfft2(f_deconvolved)
        elif method == 'Iterative':
            deconvolved = richardson_lucy(noisy_blurred,psf)
        elif method == 'Unsupervised Wiener':
            deconvolved, chains = unsupervised_wiener(noisy_blurred,psf)

        # Normalize each image for better visibility
        psf_normalized = (psf - np.min(psf)) / (np.max(psf) - np.min(psf))
        noise_component_normalized = (noise_component - np.min(noisy_blurred)) / (np.max(noisy_blurred) - np.min(noisy_blurred))
        noisy_blurred_normalized = (noisy_blurred - np.min(noisy_blurred)) / (np.max(noisy_blurred) - np.min(noisy_blurred))
        deconvolved_normalized = (deconvolved - np.min(deconvolved)) / (np.max(deconvolved) - np.min(deconvolved))

        # self.img_psf.set_data(psf_normalized)
        # self.img_noise.set_data(noise_component_normalized)  # For noise, normalization might not be needed as we visualize the raw noise pattern
        # self.img_blurred.set_data(noisy_blurred_normalized)
        # self.img_deconvolved.set_data(deconvolved_normalized)
        self.img_psf = ax_psf.imshow(psf_normalized, cmap='gray')  # Placeholder for PSF
        self.img_noise = ax_noise.imshow(noise_component_normalized, cmap='gray')  # Placeholder for Noise
        self.img_blurred = ax_blurred.imshow(noisy_blurred_normalized, cmap='gray')  # Placeholder for Blurred and Noisy Image
        self.img_deconvolved = ax_deconvolved.imshow(deconvolved_normalized, cmap='gray')  # Placeholder for Deconvolved Image
        fig.canvas.draw_idle()

        plt.show()


# Setup the figure and axes
# Setup the figure and axes
fig = plt.figure()
gs = gridspec.GridSpec(5, 3, height_ratios=[0.1, 0.1, 0.5, 1, 1])  # Define the grid layout
ax_drawing = fig.add_subplot(gs[0:3, 0])
ax_drawing.set_title('Click to add fluorophores')
ax_drawing.set_aspect('equal')
ax_drawing.set_xticks([])
ax_drawing.set_yticks([])
ax_drawing.set_xlim(0,99)
ax_drawing.set_ylim(0,99)
ax_original = fig.add_subplot(gs[3, 0])
ax_original.set_title('Object')
ax_original.set_xticks([])
ax_original.set_yticks([])
ax_psf = fig.add_subplot(gs[3, 1])
ax_psf.set_title('Point spread function')
ax_psf.set_xticks([])
ax_psf.set_yticks([])
ax_noise = fig.add_subplot(gs[3, 2])
ax_noise.set_title('Noise')
ax_noise.set_xticks([])
ax_noise.set_yticks([])
ax_blurred = fig.add_subplot(gs[4, 0])
ax_blurred.set_title('Image')
ax_blurred.set_xticks([])
ax_blurred.set_yticks([])
ax_deconvolved = fig.add_subplot(gs[4, 2])
ax_deconvolved.set_title('Deconvolved image')
ax_deconvolved.set_xticks([])
ax_deconvolved.set_yticks([])
ax_button = fig.add_subplot(gs[4, 1])
ax_button.set_title('Deconvolution method')
ax_button.axis('off')
ax_slider_pointsize = fig.add_subplot(gs[0, 1])
ax_slider_spread = fig.add_subplot(gs[0, 2])
ax_slider_noise = fig.add_subplot(gs[1, 1])
ax_slider_balance = fig.add_subplot(gs[1, 2])

# Initial parameters
initial_spread = 20
initial_balance = 0.1
initial_noise = 0.005

btn = RadioButtons(ax_button, ('Wiener', 'Unsupervised Wiener', 'Inverse', 'Iterative'))

s_pointsize = Slider(ax_slider_pointsize, 'Object size', 1, 20, valinit=5, valstep=1)
s_spread = Slider(ax_slider_spread, 'PSF spread', 5, 50, valinit=initial_spread)
s_noise = Slider(ax_slider_noise, 'Noise level', 0, 0.01, valinit=initial_noise)
s_balance = Slider(ax_slider_balance, 'Wiener balance', 0, 0.5, valinit=initial_balance)

processor = InteractiveImageProcessor(ax_drawing,ax_original,ax_psf,ax_noise,ax_blurred,ax_deconvolved,btn,s_pointsize,s_spread,s_noise,s_balance)

plt.tight_layout()
plt.show()