Identifying Calcaneal Anatomical Regions of Interest (ROI) for Quantitative Ultrasound Application in Subadults

The study sample consisted of 498 retrospective multi-slice computed tomography (MSCT) scans (258 females, 240 males) of the calcaneus of patients aged two to 20 years. All individual scans were collected by a radiologist from the Queensland Health Enterprise PACS database, representing North-Eastern hospitals in Australia. As all samples were collected retrospectively, limited CT scans were available from young children as seen in Fig. 1. All scans were conducted between 2010 and 2020 at a maximum slice thickness of 4 mm with a CT scanning parameter range of 80–120 kV, 25–81 mA, and 4–46 table feed per rotation. Scans were excluded from collection if the radiology report described the presence of any metabolic or skeletal disorders that may affect growth, or trauma such as fractures to the calcaneus. It should be noted that scans were included if patients had fractures to surrounding bones but not to the calcaneus, with approximately 85 patients having traumatic fractures to either the tibia or fibula. At the Queensland Children’s Hospital, DICOM (Digital Imaging and Communications in Medicine) files were imported into OsiriXTM (Version 4.1, 64 bits; Visage Imaging GmbH, San Diego, CA) for deidentification with the metadata including the patient’s date of scan, age, and sex retained. Ethical approval was granted by The Children’s Health Queensland Hospital and Health Service Human Research Ethics Committee (LNR/19/QCHQ/51243), ratified by the Queensland University of Technology Research Ethics Unit (Approval No. 1900000946), and approved by the Queensland Government under the Public Health Act (Section 284) 2020 (RD008018).

Fig. 1figure 1

Frequency distribution of computed tomography (CT) samples per year of age per sex (F/red = female; M/blue = male)

ROI Location and Anatomical ROI Methods

All DICOM data was imported into Horos for quantitative measurements. To standardise the measurements between individuals, a virtual “base-plane” was created which would function as the floor plate, in which the patient’s foot would be ‘standing on’ during clinical application of QUS. This base-plane was created by drawing a horizontal line between the most inferior aspect of the first metatarsal sesamoid bones (if absent, the most inferior aspect of the first metatarsal head was used), and the most inferior aspect of the calcaneal body or calcaneal apophysis, depending on the age of the individual, as seen in Fig. 2a.

Fig. 2figure 2

Sagittal MSCT scans of the tarsus demonstrating standardisation of foot position and maximum region of interest selection. a Sagittal slice of tarsus, demonstrating the true horizontal base-plane (BP) (green line) from the inferior aspect of the lateral sesamoid bone of the 1st metatarsal (orange arrow) to the most inferior aspect of the calcaneus or calcaneal apophysis (if present) (white arrow). b Calcaneal inclination angle is measured from two anatomical landmarks: the inferior aspect of the calcaneal facet for cuboid (red arrow), and the most inferior aspect of the calcaneal apophysis (as indicated by the white arrow, as seen in panel a, b, and c, noting that the inferior aspect of the calcaneal apophysis and the inferior aspect of the calcaneal facet for cuboid cannot be seen in the same sagittal slice). c The maximum region of interest circle is drawn to not overlap with the calcaneal apophyseal growth plate, or cortical bone edges of the calcaneus. The centre point of region of interest is then found by two intersecting lines, one parallel to the base-plane horizontally (blue arrow) and one perpendicular to this line vertically (yellow arrow), with the ROI centre point indicated by the intersection of the two lines. Scan was taken from a 10 year-old female

The calcaneal inclination angle is used to methodologically standardise the base-plane in cases where the forefoot was not present within the field of view and the first metatarsal sesamoid bones were not present. To measure the calcaneal inclination angle, three points were used: (1) the most inferior point of the calcaneal facet for the cuboid, (2) the most inferior point of the calcaneal body or calcaneal apophysis, and (3) the base-plane created previously, as seen in Fig. 2b.

The location of the maximum region of interest (ROImax) and its diameter (mm) was identified by drawing a circle around the largest area of trabecular bone that avoided cortical bone edges of the calcaneus within a sagittal plane through the middle of the calcaneal body (Fig. 2c). This ROI represents the most suitable location for ultrasound transducer placement for accurate QUS assessment of bone quality. To determine which sagittal plane to use, three criteria were used: (1) if the entire foot was present within the field of view, the sagittal plane of the scan was adjusted so it passed through the posterior aspect of the heel and the body of the third metatarsal, (2) the sagittal slice contained the largest posterior height (superior to inferior) of the calcaneus, (3) the insertion of the calcaneal tendon could be observed on the superoposterior aspect of the calcaneus.

After identifying the centre point of the ROI, vertical and horizontal distances were measured from the centre point to three separate palpable landmarks including the superficial aspects of the lateral and medial malleoli, and the superior aspect of the calcaneal tendon insertion on the calcaneus (inferior to the retrocalcaneal bursa) to create three novel methods (Fig. 3). The calcaneal tendon method was applied to all CT scans, however, due to the retrospective design of this study, the malleoli methods were not applied to 85 CT scans as they contained fractures to the tibia or fibula.

Fig. 3figure 3

Sagittal MSCT scans depicting coordinate measurements from superficial landmarks to the region of interest centre point (solid green circle). a Most medial slice of the medial malleolus and b most lateral slice of the lateral malleolus, with the blue and red dot representing the most superficial aspects of each malleolus, respectively. c The relative position of each superficial malleolus point is indicated by the blue and red dots which were pasted onto the same midsagittal slice as the region of interest circle. From there vertical (Y1, Y2) and horizontal (X1, X2) lines (black and white lines) were measured from the centre of the region of interest to the medial and lateral malleoli dots. d The calcaneal tendon (white X) can be seen, with the most superior aspect of the insertion of the calcaneal tendon identified by the yellow dot. e Vertical and horizontal red lines were measured from the centre point of the region of interest to the yellow dot to calculate Y3 and X3

It should be noted that the most superficial aspects of the distal tibial epiphysis and distal fibular epiphysis were used in individuals in which the secondary ossification of the malleoli had not yet commenced; for simplicity, we will refer to these points as the malleoli irrespective of patient age in this paper. In addition, the anatomical ROI method established by Jaworski et al. (1995) was applied to the sagittal CT scans by measuring the distance between the 5th metatarsal tuberosity and the most posterior aspect of the skin of the heel. A point was marked along the line at 1/3rd of the distance from the heel. From there a 1 cm line was drawn superior to this point, perpendicular to the line connecting the posterior heel to the tuberosity to identify the ‘Jaworski point’. Using the Jaworski et al. (1995) anatomical method, the ‘Jaworski point’ was created in 216 individuals where both the hindfoot and midfoot were visible within the field of view, where vertical and horizontal distances were measured from the ‘Jaworski point’ to the ROI centre point created in this study.

Inversion of the subtalar joint was measured using a coronal CT scan, in which a vertical line was drawn through the midline of the diaphysis of the tibia, with that line continuing until it passed through the most inferior aspect of the calcaneus. In the same coronal slice a second line was drawn from the most inferior aspect of the calcaneus to the most superior aspect, making sure that this line intersected the midpoint of the calcaneus. The midpoint was found by measuring the distance between the most lateral and medial aspects of the calcaneus in the coronal slice and identifying the half-way point. The angle tool was used to measure the angle between the two lines. Note that scans that contained fractures to the tibia were not used as they could affect the position of the tibial midline.

Flexion of the talocrural joint was measured in a parasagittal plane or in a pseudo-radiograph which was created by stacking the CT sagittal slices together, so structures over-lied each other and appeared as a lateral radiograph. A linear line was drawn from proximal to distal along the midline of the fibula (depending on if any fractures were present, the tibia may have been used) with the distal end of the line extending past the anterior border of the calcaneus. A second line was drawn longitudinally through the midline of the diaphysis and epiphyses of the 5th metatarsal, making sure that both lines intersected. From there the angle tool was used to measure the angle between the two lines. This approach was used as it virtually simulated the use of a goniometer.

A small number of our retrospectively collected CT scans contained fractures of the tibia, fibula, or tarsal bones, with exception of the calcaneus. For the lateral and medial malleoli methods, scans were excluded if they contained fractures to the distal aspects of the tibia and fibula, however they were not excluded for the remaining methods.

Statistical Methods

Polynomial and spline models were used to model the non-linear relationship between independent and dependent variables such as age (years), sex, region of interest diameter, and distance measurements from palpable landmarks to the centre of the region of interest. Dependent variables including flexion and inversion of the talocrural and subtalar joints, respectively, were measured in 103 individuals (20.68% of sample) ranging from 2 to 20 years, and then incorporated into the models. Modelling was performed in R Studio (2021, Ghost Orchid) with sex being specifically modelled. Differences between modelling types can be seen below [23].

Polynomial regression: this type of regression adds quadratic or cubic terms to the regression equation to better fit data that is curvilinear in nature.

Spline regression: due to polynomial regressions only being able to capture a certain amount of curvature in a curvilinear relationship, splines can be used to provide a smooth interpolation line between fixed points (knots). In our data set this was done by adjusting or increasing the degrees of freedom.

To accurately model polynomial and spline regressions in our data set, individuals under 5 years of age were grouped together to increase the number of samples in the 2- to 5-year age range.

Using the established anatomical Jaworski et al. (1995) method, the ‘Jaworski point’ was created. A vertical and horizontal distance from the ‘Jaworski point’ to the ROI centre point introduced in this study was measured. An independent samples T-test was used to determine if a sex difference existed. A one-way ANOVA with a Tukey post-hoc test was also used to determine if the distance measurements between these two points changed significantly with age.

留言 (0)

沒有登入
gif