def different_slices(images,
labels,
roi_number,
SD_0,
k,
w,
organ_id,
ismax=False,
xlim=[200, 300],
ylim=[300, 200],
plot=True):
a = 50
voxelsize = np.array([0.976562, 0.976562, 2.5])
circles = 3
seed = 123
SD = [c * SD_0[0], c * SD_0[1], c * SD_0[2]]
assd_contour_rectum = labels[..., roi_number].copy()
for i in roi_z:
dx, dy, mask, t, L, roi_z = af_Sobel.assd_Sobel(
images[i, ...],
labels[..., roi_number][i, ...],
voxelsize,
a,
SD,
circles,
seed,
k,
w,
images,
labels,
organ_id,
smooth=True)
du = af_Sobel.plotting_assd(dx,
dy,
mask,
target_img1,
quiver=False,
plot=False)
assd_contour_rectum[i, ...] = du
assd_contour = np.array(assd_contour_rectum)
if plot:
plt.figure(figsize=(15, 15))
for i in range(9):
plt.subplot(3, 3, i + 1)
plt.imshow(images[roi_z[i], ...], cmap="gray", vmin=-250, vmax=250)
plt.contour(assd_contour_rectum[roi_z[i], ...],
levels=[0.5, 1.5, 2.5, 3.5, 4.5],
colors="red")
plt.contour(labels[..., roi_number][roi_z[i], ...],
levels=[0.5, 1.5, 2.5, 3.5, 4.5],
colors="blue")
plt.xlim(xlim[0], xlim[1])
plt.ylim(ylim[0], ylim[1])
blue_patch = mpatches.Patch(color='blue',
label="Clinician's contour")
red_patch = mpatches.Patch(color='red',
label='DU contour with random seed 1')
plt.legend(handles=[blue_patch, red_patch])
plt.axis('off')
return assd_contour
def find_c(target_img1,
target_label1,
images,
labels,
c_list,
SD_0,
organ_id,
xlim=[300, 190],
ylim=[300, 190],
plot_dice=True):
dice_list = []
a = 50
voxelsize = np.array([0.976562, 0.976562, 2.5])
circles = 3
seed = 123
w = 2
c_list = c_list
k = 1
df = pd.DataFrame(columns=['k', 'c', 'w', 'min/max', "dice"])
ismax = False
plt.figure(figsize=(15, 15))
for i in range(len(c_list)):
c = c_list[i]
SD = [c * SD_0[0], c * SD_0[1], c * SD_0[2]]
dx, dy, mask, t1, L1, roi_z = af_Sobel.assd_Sobel(target_img1,
target_label1,
voxelsize,
a,
SD,
circles,
seed,
k,
w,
images,
labels,
organ_id,
ismax=ismax,
smooth=True)
du1 = af_Sobel.plotting_assd(dx,
dy,
mask,
target_img1,
quiver=False,
plot=False)
dice = dice_coef(du1, target_label1, 1)
dice_list.append(dice)
df = create_log(df, c, k, w, ismax, dice)
plt.subplot(4, 5, i + 1)
plt.imshow(target_img1, cmap="gray", vmin=-250, vmax=250)
plt.contour(mask, levels=[0.5, 1.5, 2.5, 3.5, 4.5], colors="green")
plt.contour(du1, levels=[0.5, 1.5, 2.5, 3.5, 4.5], colors="red")
plt.xlim(xlim[0], xlim[1])
plt.ylim(ylim[0], ylim[1])
plt.title("c = " + str(c_list[i]))
plt.axis('off')
# plot dice
if plot_dice:
plt.figure(figsize=(10, 10))
plt.plot(c_list, dice_list, label="Dice coef")
plt.ylabel("Dice")
plt.xlabel("c*[1.7, 2, 2.5]")
plt.legend()
return df
def find_ismax(target_img1,
target_label1,
images,
labels,
w,
k,
c,
SD_0,
organ_id,
df,
xlim=[300, 190],
ylim=[300, 190],
ismax=False,
plot_dice=True):
print("SD: " + str(c) + "*[1.7, 2, 2.5]")
print("k: " + str(k))
dice_list = []
a = 50
voxelsize = np.array([0.976562, 0.976562, 2.5])
circles = 3
seed = 123
w = w
c = c
SD = [c * SD_0[0], c * SD_0[1], c * SD_0[2]]
k = k
plt.figure(figsize=(15, 15))
#df = pd.DataFrame(columns = ['k', 'c', 'w', 'min/max', "dice"])
ismax = False
for i in range(len(w_list)):
dx, dy, mask, t1, L1, roi_z = af_Sobel.assd_Sobel(target_img1,
target_label1,
voxelsize,
a,
SD,
circles,
seed,
k,
w,
images,
labels,
organ_id,
ismax=ismax,
smooth=True)
du1 = af_Sobel.plotting_assd(dx,
dy,
mask,
target_img1,
quiver=False,
plot=False)
dice = dice_coef(du1, target_label1, 1)
dice_list.append(dice)
df = create_log(df, c, k, w, ismax, dice)
plt.subplot(4, 5, i + 1)
plt.imshow(target_img1, cmap="gray")
plt.contour(mask, levels=[0.5, 1.5, 2.5, 3.5, 4.5], colors="green")
plt.contour(du1, levels=[0.5, 1.5, 2.5, 3.5, 4.5], colors="red")
plt.xlim(xlim[0], xlim[1])
plt.ylim(ylim[0], ylim[1])
if ismax:
plt.title("ismax = True")
else:
plt.title("ismax = False")
plt.axis('off')
return df