Evaluation of bone mineral density by Digital X-ray Radiogrammetry

Theoretical and technical aspects

During the last three decades, radiogrammetry has been utilised to assess bone health and measure bone mineral density. Although measuring the cortical thickness in metacarpal bones from hand radiographs is a classical method for evaluating bone health and measuring the bone mineral density. The use of digital X-ray radiogrammetry (DXR) has developed this method and increased its precision. Special image processing is used by Digital X-ray radiogrammetry (DXR) to minimise the errors, which were happened with conventional radiography and it improves the sensitivity of this technique [1]. Image processing automates the location of regions of interest (RIO) for analysis of radiographs and calculates different bone status parameters.

Digital X-ray radiogrammetry evaluates bone mineral density and bone age from hand radiographs by using a web-based software like the BoneXpert system, which is computer software developed specifically for measuring and calculating both bone age and bone mass in children [2], [3,]. Before using BoneXpert software, there were other computerised systems. For example, Rijn et al. 2004 [5] used different software called Pronosco/ Sectra X-posure System. It determined digital X-ray radiogrammetry (DXR)-bone mineral density (BMD). This software was designed to measure (BMD) in adults, but Rijn and his team tried to analyse bone mineral components in paediatric population. Pronosco/ Sectra X-posure System was not useful in children below 10 years, and it was not able to adapt the size of the measurement region to the size of the hand.

BoneXpert software is web-based software calculates the amount of cortical bone in the metacarpal bones to determine the bone age and bone health index (BHI). It measures the cortical thickness (T), length (L), and width(W) of the three middle metacarpals from posterior- anterior view hand and wrist radiographs for non-dominant hand, and results are expressed as the bone health index (BHI) by using the following formula: BHI = T / (LW)O.33 .[6]. BHI SDS is automatically calculated based on a large cohort of Caucasian children [7].

BoneXpert determines and reconstructs bone boundaries automatically. It can compute the diaphysis and epiphysis boundaries. Two points define the bone axis of metacarpal relate to the proximal and distal ends of the diaphysis. However, the length (L) of the bone is measured along with this axis (epiphysis included). A region of interest (ROI) is placed at 44% from the proximal end of the bone, extending 25% of L [8]. In this region (ROI), the inner and outer borders of the cortical bone are defined. The boundary of the outer cortical border is a connected line at sites with maximal gradient, while the inner boundary is the line of maximal intensity [9]. Also, the cortical thickness and width are measured from this ROI.

Figure 1 below has shown the dimensions of metacarpal bones and ROI that measures to calculate BHI, the radiograph on the left has shown: cortical thickness (T), width (W), medullary diameter (M), and The radiograph on the right has shown the length (L) of middle metacarpal bones. From these dimensions, BHI and the cortical area are measured easily.

Figure 1: The radiographs illustrate the metacarpals measurements and the regions of interest (ROI). The radiograph on the left shows cortical thickness (T), width (W), medullary diameter (M), and the radiograph on the right shows the length (L) of middle metacarpal bones. “Comparison of radiogram metrical metacarpal indices in children and reference data from the First Zurich Longitudinal Study” (Martin, David D., et al., 2012).

BoneXpert cannot measure the BHI if the bones are reconstructed incompatibly or if the quality of the image is unsatisfactory, such as haziness or decrease in contrast. Also, the boundaries should not be too sharp or too blurred, and the soft tissue noise between the metacarpal bones should not be too high [8][10].

Its technique is quick and straightforward, after acquiring the digital X-ray image, it stores the image in PACS and sends the original DICOM image to the BoneXpert software, which then automatically analyses the image and sends back an annotated DICOM image to PACS. Then the radiologist can extract and read the results

High-resolution peripheral quantitative computed tomography (HRpQCT):

Principles and technical background

Contemporary progress in clinical imaging modalities permits the bone structure to be evaluated within the body. Quantitative computed tomography has provided three-dimensional macro- and microstructure information of the cortical and trabecular portions of the bone, and this information about the bone was obtained previously by taking invasive transiliac bone biopsies. High-resolution peripheral quantitative computed tomography (HR-pQCT) is the most recent quantitative computed tomography technique for evaluating bone structure and strength in the appendicular skeleton (distal radius and tibia) by measuring volumetric bone mineral density (vBMD) and assessing bone micro architecture in the trabecular and cortical bone compartments. Moreover, HR-pQCT allows finite element analysis (FEA), which provides objective parameters of bone status and strength [11],[12],[13].

The principal of HR-pQCT is similar to Quantitative computed tomography QCT and Peripheral quantitative computed tomography pQCT techniques; it is a 3D radiographic imaging methodbased on the linear X-ray absorption coefficients of the tissues through which it passes. It uses a particular bone mineral phantom to convert Hounsfield Units (HU) of the CT scan into bone mineral density units. This calibration phantom has numerous material concentrations with X-ray attenuation similar to the bone [14]. However, HR‐pQCT differs from QCT, pQCT techniques in terms of the resolution and the field of view. It can achieve higher spatial resolution with an isotopic voxel size of 82 μm(XtremeCT; Scanco Medical AG, Brüttisellen, Switzerland). A new HRpQCT scanner was invented and developed with a resolution of 64 - 52 μm HRpQCT (The XtremeCT II is the new generation HR-pQCT). This resolution is high enough to determine the bone micro architecture. Additionally, the ionizing radiation dose of the HR-pQCT scan is low than that of QCT and pQCT. It is around 3 μSv per HR-pQCT scan, while in QCT and pQCT, the radiation doses are 90 μSv and 10 μSv, respectively.

Most previous studies used a standard protocol recommended by the manufacture to use HR-pQCT in research work. Before the scan, the forearm or ankle is immobilised in a carbon cast fixed to minimize motion artifact during the scan. To determine the region of interest, an explore projection image of the distal radius or tibia is acquired to mark the reference line. The tomographic region span is 9.02 mm in length (110 slices). Then the scan region is fixed 9.5 mm proximal from the mid joint line in the radius, while 22.5 mm proximal from the tibial plafond in the tibia [15]. Figure-2 shows the HR-pQCT scan region at the distal radius and radius (A, and B), respectively.

Regarding children, there are some modifications to determine the ROI to avoid radiation exposure to the epiphyseal growth plate. Burrows, et al (2010) selected a specific region of interest (ROI), which was at a site 8% of the tibial length from the tibial platform. Therefore, the growth plate of the tibia will be protected from the harmful effect of ionizing radiation [16].

Figure-2: The HR-pQCT scan region for (A) the distal radius and (B) the distal tibia. The green area represents the ROI. This is collected from Burghardt, Andrew J., Thomas M. Link, and Sharmila Majumdar, "High-resolution computed tomography for clinical imaging of bone micro architecture." Clinical Orthopaedics and Related Research® 469.8 (2011): 2179-2193.

The raw projection images are reconstructed to generate two-dimensional grayscale images. A filter with edge enhancement and specific threshold procedure both is used to extract the cortical bone portion than the trabecular portion. The densitometric measurements and the bone morphometric analysis are obtained that is based on the segmentation of the cortical and trabecular compartments. The total volumetric bone mineral density (in mg /cm3); trabecular and cortical bone density bone (in mg /cm3); trabecular and cortical thickness (in mm); Bone volume fraction (in %) and cortical porositycortical pores (in %) can be determined from HRpQCT scan. In addition, HRpQCT has resolution and signal-to-noise ratio higher enough to measure finite element analysis FEA and assess the bone strength [12], [17].