script_convert_mask

import numpy as np
from PIL import Image, ImageDraw
from scipy import misc
import pandas as pd
import xmldataset
import matplotlib.pyplot as plt

##
filename = 'TCGA-RD-A8N9-01A-01-TS1'
with open(''.join(['./data/Amit_Sethi_data/Annotations/',filename,'.xml']), 'r') as myfile:
    data=myfile.read()

profile = """
Annotations
    Annotation
        Regions
            Region
                Id = dataset:Vertices
                Vertices
                    Vertex
                        X = dataset:Vertices
                        Y = dataset:Vertices"""

output = xmldataset.parse_using_profile(data,profile)
df = pd.DataFrame.from_records(output['Vertices'])
last_id = 0
for i in range(df['Id'].shape[0]):
    if np.isnan(float(df['Id'][i]))==0:
        last_id = df['Id'][i]
    else:
        df['Id'][i] = last_id
df['Id'] = df['Id'].astype('int')
df['X'] = df['X'].astype('float')
df['Y'] = df['Y'].astype('float')
df.to_pickle(''.join(['./data/Amit_Sethi_data/Annotations/',filename]))
df = pd.read_pickle(''.join(['./data/Amit_Sethi_data/Annotations/',filename]))


# polygon = [(x1,y1),(x2,y2),...] or [x1,y1,x2,y2,...]
# width = ?
# height = ?

origimg = misc.imread(''.join(['./data/Amit_Sethi_data/Tissue images/',filename,'.png']))
width, height = origimg.shape[0], origimg.shape[1]

img = Image.new('L', (width, height), 0)
uniq_ids = df['Id'].unique()
j = 0
polygon = []
for i in range(df['Id'].shape[0]):
    if df['Id'][i] == uniq_ids[j]:
        polygon.append((df['X'][i],df['Y'][i]))
    else:
        if len(polygon)>=2:
        ImageDraw.Draw(img).polygon(polygon, outline=0, fill=1)
        polygon = []
        j = j+1
mask = np.array(img)
np.save(''.join(['./data/Amit_Sethi_data/Annotations/',filename]),mask)
##
plt.subplot(1,2,1)
plt.imshow(origimg)
plt.subplot(1,2,2)
plt.imshow(mask)