Monday, December 26, 2011

Bone microarchitecture in human fetuses: Result part 2

We present here additional unpublished results obtained on fetal femoral metaphysis. Three femurs of respective gestational age (17, 20 and 29 weeks) were obtained from the same fetuses as discussed in the previous section. They were embedded in methylmetacrylate and imaged with 3D SR micro-CT at an energy of 20 keV, and a voxel size of 10 jm. The upper limit of each reconstructed volume was chosen at the frontier between mineralized tissue bone and cartilage, the later having little contrast. Smaller VOIs avoiding periostal bone, with a total height between 2.5 and 4.5 mm, were selected in the cancellous bone region to get quantitative parameters of trabecular microarchitecture. Figure 4 illustrates a 2D transverse slice and a 3D display of the three femoral images. The 2D slices were taken at 3 mm below the top of the volumes. The 3D displays were made on the half 3D image to show the core of femurs. The 2D slices show the anisotropic elliptic shape of the sections perpendicular to the main direction of the femurs, with a large axis approximately 1,6 times larger than the small axis length. We may also note an increase of this length between 17 and 29 weeks (2.3 and 4.4 mm, respectively). The 3D displays clearly show the increase in size of the section as it goes towards the growth plate. The top of the 3D images corresponds to the cartilaginous epiphysis of the femur, which has little contrast in x- ray attenuation. Close to this region, bone trabeculae are very small, irregular, and less mineralized. Cancellous bone may clearly be seen either on the transverse sections as well as in the 3D rendering.




Figure 4 - 3D micro-CT images (voxel size: 10 jm) of microarchitec¬ture in a fetal human femoral metaphysis. From left to right, the ges- tational ages are 17, 20, 29 weeks. Top: Axial sections were taken in the mid-shaft, 3 mm below the cartilage junction, Bottom: 3D dis¬plays of the half 3D core volumes.

From a visual point of view, trabeculae appear thinner in the region close to the growth plate. The periostal region exhibits an irregular microstructure. This is illustrated in Figure 5 showing images selected from the 20- weeks femur. The image on left side shows a transverse slice cut at 8 mm from the top of the femur, and on the right side, a 3D display of a subvolume below this slice (height 2 mm). Concentric and parallel bone plates surround trabecular bone. This external shell may be interpreted as a primitive compacta which is described as irregular, multilayered and trabeculated due to remodeling. Note that at this stage the Haversian system is not yet in place. When looked closely, a large number of perforations may be seen in trabeculae. The images may also be used to evaluate the local degree of mineralization of bone.


Figure 5 - 3D microstructure in femoral diaphysis in a human fetus (ges¬tational age: 20 weeks). Left: transverse section showing periostal apposition, right: 3D display below this section (height: 2 mm).

Brighter lines, corresponding to mineralization fronts, may be seen at the center of most bone structures. This is illustrated in the left image of Figure 5, and medial lines may be observed in most bone slices. Due to the monochromaticity of the x-ray beam, the 3D image obtained in SR micro-CT reflects the map of the linear attenuation coefficient within the sample. Brighter gray levels correspond to higher mineral concentrations, and thus more ancient bone. Thus the images indicate that after initial bone deposition, bone expands symmetrically around this initial deposition, and at the same time the mineral content of the initial bone deposition increases. The external rings of bone in the periostal apposition region is in a less mineralized state (see Figure 5, left). The mean degree of mineralization of bone and its standard deviation, expressed in g/cm3, were computed on the entire volumes using the method described in, and are reported in Table III. Quantitative analysis was performed on cancellous bone VOIs using customized implementations of the different methods described.


Figure 6 - Evolution of trabecular bone volume and trabecular thickness measured on the 3D image, as a function of distance to cartilage junction

All parameters are summarized in Table III. The architectural parameters Tb.N, Tb.Th, Tb.Sp were derived using the 3D version of the MIL method. The mean derived-trabecular thickness in each sample was found between 52 and 58 |jm. The main trabecular orientation as well as the degree of anisotropy were also evaluated from the MIL method. All samples were found to have a strong anisotropy and the main trabecular orientation was in the longitudinal direction. The second and third orientations were mainly equivalent showing some isotropy in the transverse slices. The discrete thickness map was used to com-pute the model-independent trabecular thickness Tb.Th* averaged on each VOI, found between 70 and 82 jim. In addition, since the thickness map is by definition a 3D map of local thickness, it was possible to evaluate the mean model- independent thickness on each slice of the VOIs, and thus to follow the trabecular thickness as a function of the slice level. Similarly, the trabecular bone volume BV/TV was tracked in each slice. Since the main orientation of the sample was in the longitudinal direction, these evolutions may be expressed as a distance to the cartilage junction with a step equal to the slice thickness, i.e. 10 jim. A precise identification of the growth plate is difficult since x-ray attenuation only displays mineralized tissue. We observed on all samples an increase of trabec- ular bone volume and trabecular thickness with the distance to cartilage junction. Get smart and save money! Buy cialis canadian pharmacy

Table III - 3D quantitative parameters measured in the femoral metaphysis of three human fetus (gestational ages 17, 20, 29 weeks) from 3D synchrotron radiation micro-CT images.




Gestational age

17 weeks

20 weeks

29 weeks



BV/TV (%)

23.53

22.19

22.28



BS/BV (mm-1)

38.73

35.32

34.49

Morphometric parameters

Tb.N
Tb.Sp (jim)

4.77 158.14

3.62 219.87

3.76 208.32



Tb.Th (jim)

51.64

56.62

57.99



Tb.Th* (jm)

70.11

82.28

81.49

Anisotropy parameters

DA
Orientation

2.02 ZX

2.07 ZX

2.28 ZX



NV/BV (%)

7.53%

2.15%

2.68%

Geometric parameters

PV/BV (%)

29.72%

27.68%

34.50%



RV/BV (%)

62.74%

70.16%

62.82%

Connectivity parameter

Euler density (mm-3)

-112.52

-54.29

-64.83

Mineralization

Mean (g/cm3) Coefficient of variation
(%)

0.827 18

0.839 17

0.803 20

Figure 6 illustrates these evolution on the 29-week sample. This quantitative behavior confirms the visual appearance of trabeculae in 3D displays (see Figure 4) showing the smallest trabeculae close to the cartilage junction. The changes in trabecular thickness as a function of distance d was well fitted by a linear regression (R2=0.93, p<0,0001). However the increase slows slightly down when d is above 4.5 mm. If the linear regression is limited to the range [2-4.5 mm], the prediction is improved (R2=0.98, p<0,0001) and given by Tb.Th* = 15.03 d + 24.84. The trabecular thickness and trabecular bone volume as a function of the slice level are also significantly correlated (p<0,0001) but the variation of the trabecular bone volume shows some irregularities. This is explained by the fact that, if trabeculae are characterized by a specific thickness at a given level, the arrangement of bone volume is pseudo-periodic and thus bone mass varies from one slice to another. The connectivity was evaluated by measuring the Euler density in mm-3, i.e. the Euler number normalized by the analyzed VOI. Large negative numbers indicate a high degree of connectivity.

Finally, our new geometric method described in section II.2 was used in order to identify the proportion of plates and rods in the structure. The results showed a strong prevalence of rod volume, between 62 and 70% as compared to plate volume around 30%, in each of the analyzed VOIs (see Table III). The examination of the local map shows that regions closest to the growth plate are mainly rodlike.

No comments:

Post a Comment