Power Spectral Density: Collagen D-bands (Part 2)

In a previous post, collagen D-bands were measured by sectioning an image. One sample of the D-band period was found to be 54 nm-- much less than the 67 nm found in TEM of stained histological specimens. While some variation is expected depending upon hydration, this is a significant discrepancy.

A fast Fourier transform (FFT) of the image is shown. A circular ring reflects the periodicity of the D-band spacing. This ring in the FFT is also broken reflecting the fiber orientation. Most of the intensity is between 11 and 1 o'clock as this reflects the orientation of the fibers with this D-band period. Another portion is around 2 o'clock representing the fibers at the top of the image that are diagonal in the image field. This is one useful application of the FFT: quantitating the orientation of structures in the image field.

The second image is 2D isotropic power spectral density. This reflects the power of image signal strength as a function of spatial frequency. The first maximum is at 64 nm and reflects the D-band spacing. As this includes spatial frequency information from the entire image, this is a much more robust method of estimating the D-band spacing than manually sectioning a couple D-band in an image.

Image Sectioning: Collagen D-bands (Part 1)

Images can be sectioned using the sectioning tool. In this case an image of human collagen was taken with PeakForce tapping. The image was flattened to remove large scale spatial relationships not related to the structure of individual collagen fibrils.

The D-bands arising from the staggering of tropocollagen structural units is clearly visualized on individual collagen fibrils. By drawing the sectioning tool across individual fibers, an effective digital cross section of the fiber along that line is produced. In this case the spacing between two specific bands is found to be 54 nm.  While the D-band period is a function of such environmental parameters as hydration, this is far less than the expected 67 nm seen in TEM of stained fibrils.

Application of the sectioning tool can be problematic as measurements are then biased according to the operator's selection of targets and placement of the dimensioning cursors. It is human nature of select features that are the least ambiguous and simplest to dimension by interacting with the image. As an example in this case I chose to section a long fiber in a cluster of long fibers. Removing operator bias would require a sampling methodology which covered more of the image field, including less "attractive" looking fibers, and statistically combining these measurements.