Figure 3.

Representative segmentation endpoints for the low- and high-resolution 3D HMRN data sets. Representative digital section images from a low- ([A] upper) and high- ([A'] lower) resolution 3D HMRN are shown. Magnified regions of unsegmented LC beams are shown in (B, B'), respectively. An LC beam with its central capillary is shown by blue arrows in both (B, B'). Note that an algorithm may easily segment this single beam as two (smaller) beams if the capillary space is considered an LC pore. Because they contain more detail, this is more likely to occur within high-resolution HRMNs. Since our previous report²¹ we adjusted the segmentation algorithm to achieve consistent inclusion of the capillary within the LC beam in both the low- and high-resolution HMRNs of this report by visual inspection ([C, C'], respectively). Note that LC beam segmentation is a 3D process in that data from seven section images on either side of a given section image are included in the assignment of beam borders (D, D'). Once segmented, the algorithm fills in the LC beam capillary space by classifying each capillary lumen as connective tissue. See Figure 4 for a higher-magnified version of LC beam segmentation within (C', D').