img1.tif from wholeeye_input&segmentation.tif
- The channels are organised as follows:
C1 - nuclei
C2 - nuclei segmentation
C3 - segmentation dividing cells
C4 - nuclear envelope
C5 - segmentation nuclear envelope
- has no time. just a single confocal stack
img2.tif from xwingdatafish_forcoleman.tif
- blurry, untrackable Xwing data with x,y,z,t,c=1
img3.tif
blurry 3D stack of nuclei. Can't resolve fine structure. Mosaic labeling. Very sparse. no time.
img4.tif
relatively dense labeling. SPIM data. can't resolve fine structure. just nuclear marker. no time.
img5.tif from 20150611_injHSLAP_HSH2B_28-30hpfOn_5min_ROI1_REG-1.tif
blurry, sparse, with time, SPIM, nuclear stain.
img6.tif from
20_12_17_multiview H2B_RFP&BFP_Lap2bGFP_fish6_Multiview_RIF_Subset.czi
XYCZT. nuclear envelope and nuclear volume markers. Confocal w 5min time res. 400x400 crop from x,y = 600,900 (or 600,950?) Saving to tif reshapes to TZCYX.
img6_t0_zup.npy is img6, but only the first timepoint and with the z axis cubically interpolated to scale the axis by a factor of ≈5.
res042, res043 we see the cell membrane even during Meta/Ana - phase?
res000: movie across time of original img6 data. single slice z=60.
size in MB: gif: 27, tif:92, avi(jpeg): 7, png(one time point) 5
img007 was 141029Ath5LAP2b-GFP_bactin-rasmKate2_view1_Subset_cell1-1.tif
img008 was MAX_141029Ath5LAP2b-GFP_bactin-rasmKate2_view1_Subset_cell1.tif
img006_noconv was
raw = imread("/net/fileserver-nfs/stornext/snfs2/projects/myersspimdata/Mauricio/for_coleman/ath5lap2b_zflinedata_training/20_12_17_multiview H2B_RFP&BFP_Lap2bGFP_fish6.czi")
img006_noconv = raw[0,0,:,:,49:49+71,950:950+400,1147:1147+400,0]
img006_noconv = np.moveaxis(img006_noconv, 0,2)division.npz
shape (11, 10, 2, 189, 216)
No labels. Just a single division cropped out of img006.
labels006 the wavy hand-traced sparse labels for img006 created in ilastik.
Only exists for t=0.
labels_lut.tif 31 fully-annotated xy slices from various t and z, but almost all consecutive from t==0 and z=0.
labels_iso_t0.npy are the labels for time=0 on an isotropic resolution image. It's basically the same labels as labels_lut, but properly scaled in z.
(Actually... maybe not perfectly properly scaled in z. Should the labels stretch out with the pixels around them or stay const z-width?)
img006+lab normal img006 but with labels006 labels in the first time point
in the third channel.
img006_labels_000.tif just the 31 labeled xy slices from labels_lut, but all concatenated together in a row. 3 channel. labels are first channel.
img006_borderlabels boders around cells in first 10 xy slices in t=0 are drawn. Then each of these 2D cells has a centerpoint annotation.
anno000.npy
shape (130,2)
2D annotations from 3D annotation tool.
labels cells as correct '1', underseg '2', etc... pairs with lab000.npy
lab000.npy
shape (355,400,400)
label image
automated segmentation of upscaled image.
The labels don't change smoothly as we travel through z. sometimes no change at all.
point_anno.csv
csv with header: Area,Mean,Min,Max,X,Y,Ch,Slice,Frame,Counter,Count
cell centerpoint annotations for img006, t=1 (2nd timepoint)
extract them with lib.mkpoints
points = lib.mkpoints()
cen = np.zeros((71,400,400))
cen[list(points.T)] = 1
cen2 = convolve(cen, np.ones((5,)*3)))
iss = Stack(perm(np.concatenate([img[1], cen2[:,A]],1), 'zcyx','zyxc'))
img006.ilp - obviously matches with the img006 input data...