STimage-1K4M

Crop spot-level image

To perform spot-level analysis, you need to crop the slide images into spot-level. We used the following code to crop the spot-level images. We note here although all the images are named as png, they are actully a mixture of jpg, tiff and png. We suggest you using PIL.Image.open since it can automatically detect image type.

import sys
import pandas as pd
from PIL import Image
import matplotlib.pyplot as plt


Image.MAX_IMAGE_PIXELS = 3080000000000000000000000000000000000

def crop_image_all(path,slide):
  # We note here that although
  temp_image = Image.open(f'{path}/image/{slide}.png')
  temp_image = temp_image.convert('RGB')
  coord = pd.read_csv(f'{path}/coord/{slide}_coord.csv')
  r = coord.r[0]
  for i in range(len(coord.index)):
      yaxis = coord.yaxis[i]
      xaxis = coord.xaxis[i]
      spot_name = coord.iloc[i,0]
      temp_image_crop = temp_image.crop((xaxis-r, yaxis-r, xaxis+r, yaxis+r))
      temp_image_crop.save(f"./crop/{spot_name}.png")
        #plt.scatter(xaxis,yaxis,c='r')



index = int(sys.argv[1])
meta = pd.read_csv('meta_all_gene.csv',sep='\t')

slide = meta.slide[index]
tech = meta.tech[index]

path=f'./{tech}/'
crop_image_all(path,slide)
Search