Intro

One of the core features of elmr is to enable data in the FAFB EM space to be transformed into one of a number of standard light level template brains. This transformation depends on manually selected landmarks. We can inspect these landmarks and to try to measure their consistency.

Setup

library(elmr)
library(knitr)
opts_chunk$set(fig.width=5, fig.height=5)
# set up for 3d plots based on rgl package
rgl::setupKnitr()
# frontal view
view3d(userMatrix=rgl::rotationMatrix(angle = pi, 1,0,0), zoom=0.7)

Landmarks

The standard landmarks used by elmr were specified by Davi Bock on the FAFB and JFRC2013 brains using the elm tool developed by Stefan Saalfeld and John Bogovic. We can plot the landmarks in the context of the JFRC2013 brain as follows.

plot3d(JFRC2013)
# note that the landmarks are in raw voxel coordinates and must be scaled
xyz=scale(elm.landmarks[,c("X","Y","Z")], scale = 1/voxdims(JFRC2013), center = FALSE)
xyz=as.data.frame(xyz)
rownames(xyz)=sub("Pt-","",elm.landmarks$Label)
spheres3d(xyz, col = ifelse(elm.landmarks$Use, "green", 'red'), radius = 4)

Landmarks based transform

The standard elm.landmarks set is used within to the elmr package to register a transformation that can map FAFB->JFRC2013 space in conjunction with the xform_brain functions provided by the nat.templatebrains packages.

Let’s first look at the FAFB landmarks

# note that these positions are in nm units and do not need scaling
xyz.fafb=elm.landmarks[,c("X1","Y1","Z1")]
rownames(xyz.fafb)=sub("Pt-","",elm.landmarks$Label)
kable(head(xyz.fafb))
X1 Y1 Z1
1 452638.6 93129.93 92560
2 522271.2 184057.51 96160
3 428701.7 173888.16 112440
6 597973.7 99741.26 77200
7 629661.1 172401.93 146120
8 629608.0 170950.13 185640

Now we can try to look at the consistency of the transform between each landmark pair as follows:

  1. Drop out each landmark pair in turn
  2. Calculate position of that landmark in target space using remaining landmark pairs
  3. Measure difference between calculated position and manually defined position
# record transformed position with leave one out
xyzt.loo=xyz
for(i in 1:nrow(xyz)) {
  thisreg <- reglist(tpsreg(xyz.fafb[-i, ], xyz[-i, ]))
  xyzt.loo[i,]=xform(xyz.fafb[i, , drop=FALSE], thisreg)
}
deltas=sqrt(rowSums((xyzt.loo-xyz)^2))
hist(deltas)

Let’s looks at the deltas (distance between leave one out vs manually defined landmark position) for each point.

kable(cbind(elm.landmarks, delta=deltas))
Label Use X Y Z X1 Y1 Z1 delta
1 Pt-1 TRUE 571.4001 38.85996 287.05954 452638.6 93129.93 92560 4.8567731
2 Pt-2 TRUE 715.8113 213.29936 217.39349 522271.2 184057.51 96160 2.0271878
3 Pt-3 TRUE 513.0022 198.00197 217.79409 428701.7 173888.16 112440 8.6873515
6 Pt-6 TRUE 867.0125 31.91925 276.22344 597973.7 99741.26 77200 4.5722904
7 Pt-7 TRUE 935.2109 234.22952 351.51807 629661.1 172401.93 146120 6.6747676
8 Pt-8 TRUE 935.2109 232.62607 405.50079 629608.0 170950.13 185640 6.4982959
10 Pt-10 TRUE 509.7629 244.91917 362.20772 431157.3 165611.08 169920 8.8340362
11 Pt-11 TRUE 509.2284 281.12828 403.48102 431024.0 173148.67 207560 5.1295757
12 Pt-12 TRUE 509.2284 165.87835 401.67082 431057.1 125923.99 189520 6.7021690
14 Pt-14 TRUE 1106.8503 372.72023 222.01044 711682.1 254995.10 112320 12.4510042
17 Pt-17 TRUE 718.4719 242.59406 355.69466 531137.1 165586.38 164680 2.6243835
18 Pt-18 TRUE 776.9947 213.33267 378.56799 566543.7 148578.85 171000 8.3093447
19 Pt-19 TRUE 839.6388 287.10433 382.68932 587828.7 180069.25 180880 6.0548760
20 Pt-20 TRUE 601.2203 288.54680 382.68932 479270.4 176726.32 190640 2.5302671
21 Pt-21 TRUE 653.3550 221.57531 380.62865 500660.6 148115.02 175520 3.1633781
23 Pt-23 TRUE 646.4883 588.60129 222.43882 502568.4 335377.59 185680 14.3595881
24 Pt-24 TRUE 769.5130 605.94266 204.25736 555571.9 350062.58 171760 2.9367527
25 Pt-25 TRUE 1150.4870 201.70284 365.00699 733410.6 150225.52 148840 8.2068118
26 Pt-26 TRUE 1045.9179 168.05378 215.08408 685711.4 174524.46 68960 3.0133531
27 Pt-27 TRUE 389.1564 177.03002 206.10784 369059.9 163273.69 105480 4.8117727
39 Pt-39 TRUE 294.9652 198.55742 355.90246 324455.5 140507.97 176320 5.7458094
40 Pt-40 TRUE 241.6744 211.20539 350.83541 301541.1 160802.95 176320 7.4296983
45 Pt-45 TRUE 325.9132 597.71786 158.23298 340429.1 338520.68 171320 20.7364915
46 Pt-46 TRUE 197.7791 449.11545 319.63260 306709.7 285779.48 222520 22.7770038
47 Pt-47 TRUE 120.6732 274.52403 323.29304 233960.2 216887.97 181080 15.5495428
48 Pt-48 TRUE 162.5141 275.24054 425.51338 285325.0 165261.48 232120 16.4094093
49 Pt-49 TRUE 413.3632 161.16906 316.82330 382965.0 139994.70 148000 4.9865533
51 Pt-51 TRUE 732.8606 262.06885 208.87088 532888.7 206000.44 102280 2.8474895
52 Pt-52 TRUE 763.4152 262.35759 225.77664 549393.3 198858.96 104640 5.8249803
56 Pt-56 TRUE 1021.5445 379.96095 136.36467 669071.4 282885.13 73560 3.9146401
57 Pt-57 TRUE 1212.4784 422.50773 128.28765 744107.8 289188.76 100560 19.2798860
58 Pt-58 TRUE 1249.0229 401.20869 365.68265 756760.4 233828.39 201200 3.8537199
59 Pt-59 TRUE 1174.4594 631.83024 248.45789 745417.8 353730.87 198160 22.6614828
60 Pt-60 TRUE 1077.1313 573.66563 299.84596 674374.9 325483.14 198160 17.6697144
61 Pt-61 TRUE 1112.7070 339.48705 401.72907 683073.3 207169.29 198160 19.3194897
62 Pt-62 TRUE 1306.1523 263.30353 411.07188 786547.0 167538.31 198160 7.2968528
63 Pt-63 TRUE 1328.8859 469.20296 354.47855 786667.9 254601.10 219560 12.1283499
64 Pt-64 TRUE 549.6954 583.20048 162.32429 436921.0 324144.32 168120 13.3160542
65 Pt-65 TRUE 765.1116 188.70754 178.33252 543563.4 191829.65 66480 7.0929336
66 Pt-66 TRUE 785.3950 425.64347 15.25080 559862.1 319733.97 42880 7.2362758
67 Pt-67 TRUE 491.9622 447.69119 309.37694 416087.8 250364.48 202160 4.3195663
68 Pt-68 TRUE 636.6000 582.23100 304.45079 475821.1 301542.34 209200 17.8857285
69 Pt-69 TRUE 751.5368 622.07216 324.54432 549630.2 318599.41 227360 11.0236700
70 Pt-70 TRUE 248.6964 386.77664 133.64117 302955.0 281032.97 132480 12.9435630
71 Pt-71 TRUE 371.3233 241.13746 246.95856 352159.8 181155.82 132480 5.6993404
72 Pt-72 TRUE 125.3698 356.45723 424.67776 276788.0 219417.40 262680 16.6889349
74 Pt-74 TRUE 297.1745 286.05935 401.86582 328376.8 172800.07 224200 12.3145756
75 Pt-75 TRUE 120.6515 261.44368 422.62512 234767.8 172806.20 231640 20.4604900
76 Pt-76 TRUE 371.7625 362.30048 386.60631 372970.9 200278.04 221400 9.9392250
77 Pt-77 TRUE 863.6581 20.51800 305.28403 593593.3 92800.43 83480 5.2978426
78 Pt-78 TRUE 877.3450 69.27787 262.22208 602858.0 116568.92 72080 4.9387303
79 Pt-79 TRUE 887.6924 116.68930 198.29597 610278.8 156940.02 53160 4.1762134
80 Pt-80 TRUE 456.7993 163.75185 410.38070 413853.6 120770.68 191640 4.8779902
81 Pt-81 TRUE 457.2808 92.26043 414.81480 413799.3 90902.71 166320 8.3478615
82 Pt-82 TRUE 543.1608 46.12096 309.74346 441965.7 86907.20 110360 2.2727861
83 Pt-83 TRUE 610.5969 134.29354 334.08200 475282.3 125516.26 138120 3.3060558
84 Pt-84 TRUE 687.3255 150.56257 339.67920 517256.1 128105.25 138120 5.0033435
85 Pt-85 TRUE 637.3402 37.40585 246.20518 485233.0 96212.32 71640 6.9772825
86 Pt-86 TRUE 601.4145 48.82580 235.55270 457888.0 104535.82 71560 7.5135967
87 Pt-87 TRUE 470.4498 113.37605 185.97815 412043.1 138743.20 73360 9.7471167
88 Pt-88 TRUE 646.9302 287.95844 201.12036 497269.3 219264.50 105080 6.8055324
89 Pt-89 TRUE 617.9160 393.59540 128.45568 479040.7 271503.39 106320 7.1815581
90 Pt-90 TRUE 539.2479 159.18572 362.26769 448664.2 130197.79 162440 3.6823690
91 Pt-91 TRUE 599.1029 299.87976 339.71151 475916.7 189951.39 167480 4.0787974
92 Pt-92 TRUE 524.0257 228.59402 80.51068 429789.2 217571.25 44040 4.5038170
94 Pt-94 TRUE 568.7742 195.81066 147.47252 447827.5 191431.30 65080 7.1549798
95 Pt-95 TRUE 559.2319 216.64859 115.06092 451509.5 210571.92 56680 2.0295053
96 Pt-96 TRUE 694.7296 181.04882 208.66967 512860.3 173756.69 83760 3.7317246
97 Pt-97 TRUE 522.7734 153.64999 152.36817 424630.5 170571.37 65520 6.4315287
98 Pt-98 TRUE 723.9280 208.76311 156.20774 521821.9 191059.99 65920 9.4240092
99 Pt-99 TRUE 561.4849 406.58976 139.86903 439456.2 277536.72 114040 8.7170864
102 Pt-102 TRUE 589.7660 374.11829 65.80798 456343.4 278586.81 70040 5.4709799
104 Pt-104 TRUE 528.9954 249.36157 52.77704 429666.7 236504.64 40720 2.6290145
105 Pt-105 TRUE 559.3112 311.49054 38.99692 446608.2 266996.18 40720 4.4235895
106 Pt-106 TRUE 536.0237 280.85395 42.25106 428356.2 251917.06 39440 3.6249780
108 Pt-108 TRUE 529.8430 190.66314 99.95127 443220.7 198582.95 46960 8.8340324
109 Pt-109 TRUE 588.3022 180.80908 73.79142 465835.1 208589.43 30640 3.2212746
110 Pt-110 TRUE 523.6692 255.82408 96.18389 432486.9 227589.09 60560 2.0600276
111 Pt-111 TRUE 519.9193 323.34459 95.62596 429673.5 249409.19 75440 5.6314508
112 Pt-112 TRUE 644.9661 227.95424 49.43940 492185.1 238768.76 27200 4.9893856
114 Pt-114 TRUE 596.8888 313.78083 28.38294 467809.3 269292.47 39960 3.9160562
115 Pt-115 TRUE 549.6726 232.40734 27.72951 435030.4 233285.12 26120 5.8291358
116 Pt-116 TRUE 793.1537 291.40008 203.01114 552445.9 221911.47 96760 7.4515988
117 Pt-117 TRUE 932.0076 190.30624 207.65353 623021.1 182913.48 78400 6.3006672
118 Pt-118 TRUE 913.0380 127.35328 214.32475 619919.5 154904.04 69920 4.0765125
119 Pt-119 TRUE 1049.5826 275.41781 194.25617 678227.0 226038.51 77000 3.8476223
121 Pt-121 TRUE 778.8555 329.78345 110.36329 553018.5 264730.38 62080 2.3813873
122 Pt-122 TRUE 766.8180 238.30891 129.28220 546434.1 223179.57 44560 5.3464239
126 Pt-126 TRUE 821.1940 165.31615 170.20768 580468.1 184025.15 57240 5.6133324
127 Pt-127 TRUE 814.9952 231.24334 335.40372 568773.4 173760.52 141240 9.5395721
128 Pt-128 TRUE 988.4672 148.08581 377.28153 658380.7 134538.09 155240 2.8092561
129 Pt-129 TRUE 991.9385 171.66403 374.07560 658486.8 144747.32 155240 0.8946411
130 Pt-130 TRUE 955.0846 81.15812 394.58469 644408.4 104909.96 146160 3.3820125
131 Pt-131 TRUE 858.8542 137.36019 305.95222 585811.2 138028.77 109960 7.9843254
132 Pt-132 TRUE 902.7680 109.49854 349.66070 616437.7 117813.37 130960 3.7968063
133 Pt-133 TRUE 1063.1324 180.50395 294.78720 693213.5 164460.49 113480 1.2879857
139 Pt-139 TRUE 1031.4633 60.62717 298.21962 685748.2 117615.93 97360 4.0234049
140 Pt-140 TRUE 1072.7554 89.46014 292.45409 702673.0 128121.18 103640 4.5195348
141 Pt-141 TRUE 906.2379 36.43737 347.67361 617878.2 90004.29 110640 1.1635802
142 Pt-142 TRUE 894.3160 179.67182 111.32961 611880.8 221748.47 22560 5.9632264
143 Pt-143 TRUE 806.9353 307.39564 41.45364 563795.2 280366.60 24360 3.5135552
144 Pt-144 TRUE 840.9156 187.58365 109.41779 586630.0 218165.18 24360 2.8986408
145 Pt-145 TRUE 882.2956 302.07558 46.33199 604025.1 281980.35 24360 3.5220758
146 Pt-146 TRUE 918.1660 212.11489 98.60478 619392.5 233507.39 21320 3.0237052
147 Pt-147 TRUE 895.6440 278.92859 131.98689 612176.8 239146.97 56800 6.1187261
148 Pt-148 TRUE 760.4791 261.92395 138.44280 544916.4 238404.45 56800 6.9567334
150 Pt-150 TRUE 1019.6740 349.16788 149.33955 664468.4 266903.26 65720 6.3876473
151 Pt-151 TRUE 920.9986 280.81417 97.82217 627242.5 257156.22 39080 5.5171652
152 Pt-152 TRUE 921.7953 243.70019 69.01316 618928.3 257152.53 15920 4.8130192
153 Pt-153 TRUE 1026.1469 218.11495 158.36097 669259.1 208512.87 46280 4.7513239
154 Pt-154 TRUE 815.5029 393.53224 138.96906 569609.9 281005.27 90880 2.9149836
155 Pt-155 TRUE 794.5535 205.37576 107.90936 568153.1 215132.53 29160 5.2031648
156 Pt-156 TRUE 801.2243 199.57122 76.64452 568700.7 225895.95 10560 2.6632709
158 Pt-158 TRUE 814.9698 322.09382 38.20410 568697.9 283356.34 29360 2.4646616
159 Pt-159 TRUE 838.3869 377.84149 71.16519 583877.6 290936.45 63640 7.0485530
160 Pt-160 TRUE 411.8688 62.66130 313.62719 394760.5 91104.42 119360 10.7763448
161 Pt-161 TRUE 496.8204 48.48879 281.87705 417463.3 91287.59 105160 7.6300593
162 Pt-162 TRUE 428.9324 105.46846 259.46298 387742.4 120397.12 112400 5.6982507
163 Pt-163 TRUE 348.1907 135.07246 325.80300 351378.7 120400.07 153960 5.0459193
164 Pt-164 TRUE 709.1743 290.44481 186.96499 519852.8 220066.31 101920 4.1074545
165 Pt-165 TRUE 699.9468 306.79750 238.26757 521274.8 208667.13 125880 6.2903583
166 Pt-166 TRUE 608.1120 187.71150 274.79127 471356.9 156196.13 119480 3.4605347
167 Pt-167 TRUE 714.0975 140.92095 311.99102 520683.3 131647.18 119480 4.2230713
168 Pt-168 TRUE 772.3868 182.61529 314.80399 549523.6 144648.80 128720 5.7080171
169 Pt-169 TRUE 768.2635 248.16935 300.42741 549526.6 181341.05 139520 3.3483182
170 Pt-170 TRUE 826.6157 180.43207 267.93905 581265.9 162330.58 105360 6.2424722
171 Pt-171 TRUE 785.7634 208.03504 306.12177 560689.1 163361.97 133400 5.3466226
172 Pt-172 TRUE 811.7054 59.21032 162.54342 566206.7 134796.13 28680 5.0805028
173 Pt-173 TRUE 322.4938 314.56546 221.63341 326640.0 225641.20 141360 6.3571177
174 Pt-174 TRUE 405.4023 406.32913 182.38558 370744.3 267765.51 141360 3.4199451
175 Pt-175 TRUE 370.4399 96.40938 329.08686 365886.3 100756.18 138560 3.5423043
176 Pt-176 TRUE 912.5393 228.67122 427.44708 621995.4 159644.05 202720 5.2879125
177 Pt-177 TRUE 976.2923 123.25038 411.18404 646015.3 122117.82 165280 7.0213588
178 Pt-178 TRUE 1129.1425 227.70428 380.81028 722705.1 167172.74 162120 7.7402841
179 Pt-179 TRUE 341.2364 173.62318 404.95490 351948.7 124957.93 194040 3.4439352

And then visualise them in 3D, colouring each point by the magnitude of the delta.

clear3d()
view3d(userMatrix=rgl::rotationMatrix(angle = pi, 1,0,0), zoom=0.7)
jet.colors <-
  colorRampPalette(c("#00007F", "blue", "#007FFF", "cyan",
                     "#7FFF7F", "yellow", "#FF7F00", "red", "#7F0000"))

plot3d(JFRC2013)
spheres3d(xyz, col = jet.colors(10)[cut(deltas, 10)], radius = 4)
# nb these are numbered from 0 like the elm landmarks
texts3d(xyz, texts = rownames(xyz), adj = c(1,1))

Based on this, I would definitely suggest checking the following landmarks

75/48 43 46 68 23

Compare the leave one out landmark position with manually defined position

clear3d()
view3d(userMatrix=rgl::rotationMatrix(angle = pi, 1,0,0), zoom=0.7)
for(i in 1:nrow(xyz)){
  segments3d(rbind(xyz[i, , drop=FALSE], xyzt.loo[i, , drop=FALSE]), col=jet.colors(10)[cut(deltas, 10)][i], lwd=3)
}
points3d(xyz)
texts3d(xyz, texts = rownames(xyz), adj = c(1,1))
plot3d(JFRC2013)

Now of course this is not just a measure of consistency in each landmark point, but it will also depend strongly on the extent to which the neighbouring points can define the required registration; in areas requiring substantial non-rigid deformation this may be a problem; this will be further emphasised when the neighbouring points are distant. We can check the latter point like so:

library(nabor)
nearest3=rowMeans(nabor::knn(xyz, k=4)$nn.dists[,-1])
plot(deltas~nearest3)
deltabynearest3=lm(deltas~nearest3)
abline(deltabynearest3)

summary(deltabynearest3)
## 
## Call:
## lm(formula = deltas ~ nearest3)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -9.1418 -2.0414 -0.1602  1.6765 14.0253 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  0.22861    0.74626   0.306     0.76    
## nearest3     0.20458    0.02155   9.491   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 3.522 on 133 degrees of freedom
## Multiple R-squared:  0.4038, Adjusted R-squared:  0.3993 
## F-statistic: 90.08 on 1 and 133 DF,  p-value: < 2.2e-16

So there is a correlation between neighbour distance and our calculated delta; with an R^2 of about 0.4 it’s reasonably strong. This may be explained by Davi placing more landmarks in areas where he is more certain.

Perhaps the ideal thing would be to find a set of landmarks on one side of the brain and then find exactly the same landmarks on the other side of the brain. For the JFRC2013 brain we can exactly mirror points from one side to the other using an image-based registration distributed with the nat.flybrains package.

We can find the mirror image position of all the points that we used like so:

#  nb this is conditioned on being able to find CMTK
if(isTRUE(nzchar(cmtk.bindir()))){
  xyzm=mirror_brain(xyz, brain = JFRC2013)
  
  clear3d()
  spheres3d(xyz, col='red', rad=2)
  spheres3d(xyzm, col='green', rad=2)
  plot3d(JFRC2013)
  view3d(userMatrix=rgl::rotationMatrix(angle = pi, 1,0,0), zoom=0.7)

}

One could then write out a table of landmarks including these mirrored positions like so:

xyzm.pixels=scale(xyzm, scale = voxdims(JFRC2013), center = FALSE)
colnames(xyzm.pixels)=c("XM","YM","ZM")
write.table(cbind(elm.landmarks, xyzm.pixels), 'elm.landmarks-with-mirror.tsv',
            col.names = T, sep='\t', row.names = F)

If one were then to take the mirror image landmarks in JFRC2013 space and manually pick the corresponding locations in EM space, one should be able to compare two different paths to the same point (one mirroring in FAFB, the other mirroring in JFRC2013) and see whether there is a difference in the calculated position. Differences would presumably reflect the uncertainty in picking positions in the EM volume (and a small contribution from the mirroring uncertainty in JFRC2013, which nevertheless expect to be <1 µm based on analysis in Manton et al 2014).