Fig 1.
Light field microscope system used for 3D phenotypic imaging.
Green and red ray bundles show how different lateral positions in the sample are focused onto different elements into the microlens array. Abbreviations: C–condenser lens; MO–microscope objective; piezo–piezoelectric nosepiece focusing stage; TL–tube lens; sCMOS–scientific CMOS camera. Inset shows (from left to right): A raw light field micrograph of a swimming C. elegans specimen; a perspective (pinhole) view of the object formed by extracting the same pixel from each microlens subimage; a focused view of the worm formed by summing the signal from all pixels within a microlens subimage. Pixels using to create each image are shown in red.
Fig 2.
Extraction of C. elegans 3D midline skeleton from a light field image.
(a) On axis pinhole view with segmentation mask outline shown in green. (b) Combined depth estimates within the segmentation mask displayed using a colour scale. (c) Final extracted midline skeleton of the organism comprised of 25 linear segments. The orientation of each segment is described by an azimuthal (θ) and polar angle (ϕ).
Table 1.
Posture and locomotion metrics computed to compare behaviour of dpy-10 and dpy-13 mutants.
Fig 3.
Calibration of depth scaling factor.
Correspondence (red), defocus (blue) and combined (black) depth estimates versus known depth obtained by capturing a series of light field images as a paralysed C. elegans specimen was translated through a series of known axial positions. Defocus and correspondence estimates offset by + 100 μm and -100 μm for clarity.
Fig 4.
3D posture and movement of C. elegans during foraging.
(a) On axis pinhole views from a 30 second long light field image sequence of a wild type C. elegans organism moving within agarose gel. (b) Perspective views showing corresponding reconstructed mid-line skeletons. The total volume occupied by the animal during the sequence, represented by the grey bounding box, is 0.027 mm3. The nose (red) explores a volume of 0.009 mm3.
Fig 5.
The cuticle topography of dpy-10 and dpy-13 captured using atomic force microscopy.
Topography images of immobilised young adults captured in contact mode using a 2 nm probe tip as described in [21].
Fig 6.
Mean locomotion and posture metrics for dpy-10 (blue) and dpy-13 (orange) mutants moving in an agarose gel.
Error bars show ± standard error in the mean. Statistical significance of difference between measured values for each strain (confidence in rejection of the null hypothesis) indicated by the p-value shown in each plot. P-values computed using a two sample t-test (assuming unequal variances) applied to values measured for each strain, with the measured value for each individual worm was computed as the mean over all data captured for that animal. Statistical significance in difference between directional autocorrelation curves assessed by performing a two sample t-test on the decay constants of exponential fits applied to the mean data for each individual worm.
Fig 7.
(a) The first four eigenworms computed separately from azimuthal (θ, solid line) and polar (ϕ, dashed line) angles. (b) Normalised shape variance using different numbers of eigenworms. The first four eigenworms in both θ and ϕ are sufficient to capture 95% of the body shape variance.
Fig 8.
Time dependence of eigenworm projection amplitudes for a swimming wild-type worm.
(a) Relationship between the first and second eigenvalues computed for azimuthal, θ (top) and polar, φ (bottom) angles. (b) Variation in projection amplitudes in azimuthal and polar angles for the first eigenworm (left) can be approximated using a Rhodonea curve (right). (c) A least squares fit to the Rhodonea indicates that the frequency with which the animal reorients its plane of oscillation is approximately 8.5 times lower than the periodic in plane motion of the animal during swimming.