about data extract in nodclsgbt.py about deeplung HOT 5 CLOSED

1 Could you please help me check the range of data = np.load(os.path.join(preprocesspath, pid+'_clean.npy'))
2 Sometimes the shape of data is smaller than bgx+CROPSIZE, bgy+CROPSIZE, bgz+CROPSIZE

from deeplung.

oh, the range of data is enough for cropsize, i didn't realize numpy is ok when out of range.
but , i still don't understand why you use 2-cropdatatmp to fill the cropdata, and i think cropdatatmp=np.array(data[0,bgx:crdx+cropsize/2,bgy:crdy+cropsize/2,bgz:crdz+cropsize/2) may be better

from deeplung.

I mean the min value and max value for the range.

from deeplung.

x ： 117 378
y： 160 392
z： 168 400

from deeplung.

I think it should be np.array(cropdatatmp), not np.array(2-cropdatatmp)

It should not affect the GBT results because tree classifier is not affected by input range. Maybe you can get better results after changing it.

from deeplung.