Stratified body shape-driven sizing system via three-dimensional digital anthropometry for compression textiles of lower extremities

Abstract

Poor fit has become one of reasons for high non-compliance in the use of garments made of compression textiles in venous deficiency treatments. A novel methodology to categorize lower body shapes and sizes has been established via three-dimensional digital anthropometric technology in this study based on 208 Hong Kong subjects aged 40–60 years. Three new parameters were introduced to classify body shapes, namely the “A-angle” for assessing the “alignment” of lower extremities, the “cosine values of the key angle” at the turning point for below-knee shape determination, and “gradient” for above-knee shape categories. The mathematical simulation via the interpolation function was employed to explore the characteristics of shape variation trends with the involvement of dynamic interactions of both circumferences (Cir) and heights (Hei) of lower extremities. The clustering analysis quantitatively segmented the sample population into three stratified leg morphologies (i.e. diamond, inverted trapezoid, and balanced leg shapes) in terms of the determined anthropometric landmarks along the lower extremities, in which the Cir(s) of the brachial (cB1), calf (cC), and thigh (cF) exhibited most obvious differences among the clustered lower limbs. The created stratified shape-driven sizing system and methodologies further involved the body shape classifications into the Cir-based size categories to cater for diverse body morphologies in product size selection, thus improving dimensional fitness and accurate treatment using compression textiles in practice.

Keywords

stratified body shape and size three-dimensional digital body scanning anthropometry compression textiles lower extremity

Chronic venous insufficiency (CVI) is a long-term progressive condition in which venous pooling reduces venous return, resulting in pain, itching, tiredness, varicose veins (VV), and even ulceration in the lower limb,¹ which affects 25–40% of women and 10–20% of men globally.² Compression therapy has been used in CVI treatment for centuries.³ Functional compression textiles with modalities of graduated compression stockings (GCSs) and bandages as an essential “source of pressure” have demonstrated their efficacies in promoting re-absorption of interstitial fluid, reduction of venous reflux, and improvement of venous return^4–6 in the physiotherapy of CVI. However, high non-compliance resulting from discomfort wearing perceptions and poor fit affected their effectiveness.^7,8 In a recent clinical survey it was reported that 20–43% of hospitalized patients wore incorrect sizes⁹ and 77% of respondents considered that initiating a hosiery fitting service would improve compliance.⁷ “Pressure dosage”, as the key indicator reflecting compression by textile interventions, is generated following action and reaction force pairs within Newton’s Third Law and is influenced by the geometric morphologies of the body where the pressure is located. To date, pressure dosages have been strictly regulated by standards from different organizations and countries¹⁰ while, in comparison, much less attention has been drawn to the classifications of body morphologies of users, which could be one of the reasons for poor fit and discomfort wearing perception in practice.

The available sizing system related to GCSs is commonly established in the laboratory based on in vitro wooden leg models or a “marking board”, as recommended by the Medical Compression Hosiery European Prestandard (ENV 12718). In practice, the circumference of the ankle is employed as the prioritized parameter to define the size of GCSs. The perimeters along other height levels as the reference values assist the size decision. This is largely due to the fact that the existing pressure levels of GCSs (e.g. Class I: 15–21 mmHg; Class II: 23–32 mmHg; Class III: 34–46 mmHg; and Class IV: ≥49 mmHg in terms of the European Prestandard¹¹ and other pressure levels indicated by Germany standard RAL-GZ387 and British standard BS:6612, etc.¹⁰) are commonly categorized according to the pressure exerted at the ankle area of the lower leg.^10–13

Irregular leg shapes may fit ankle perimeters in terms of the “size chart” given, but may exceed the calf size, thus leading to a tourniquet effect or ischemia or increased risk of thrombosis.¹⁴ In response, the “allowance” (or tolerance) with units of centimeters (cm) or inches (in) is added into the same size level to cater to wider leg sizes; however, this allowance may decrease the accuracy of the pressure dosage delivered in use.¹⁵ No existing sizing system takes body “shape” into account to fit diverse leg morphologies in compression therapy.

An early significant contribution to body shape classifications was made in the 1930s by William Sheldon et al., who introduced the concept of “somatotype” in their book The varieties of human physique.¹⁶ The somatotype is recognized as a valuable factor in creating custom-fit solutions, which has been applied in figure identifications for improvement of apparel pattern-making and fitting of mass-customized clothing.^17–20 Three-dimensional (3D) digital body scanning techniques have become the cornerstone of body-surface and anthropometric measurements for fashion industry and healthcare applications, including two-dimensional (2D) video silhouette images converted to 3D models, white light phase-based image capture, laser-based image capture, radio-wave linear array image capture, etc.^21–23 Studies on the anthropometry of lower limbs via 3D body scanning have been carried out, including lower body measurements for active body positions²⁴ and categorization of lower body shapes for adult females based on a multiple view analysis.²⁵ However, the existing studies are still based on body measurements or size optimization for common apparel purposes while lacking direct guidance for functional compression textiles with highly strict requirements on pressure dosage and body dimensions. Moreover, the studied subjects are mostly female^20,24,25 or young people,^18,19 while less attention was placed on the elderly, who are in the high risk group for suffering from CVI.

Relying on the controlled dimensions, compression textiles can be constructed to be smaller than the sizes of the body over which they are fitted, called “negative ease”, thus flexibly delivering calibrated pressure with fast molding to cater to different body shapes at an affordable cost.¹⁰ Raising precision in the identification of body shape and size are the prerequisites to ensuring the effectiveness of pressure dosage delivered by GCSs. However, no referable methodologies were reported on body shape and sizing identifications to guide the size selection of GCSs in practice. Therefore, in our present work, a novel methodological study has been conducted to create a new stratified shape-driven sizing system (S⁴) based on 3D digital body scanning, anthropometric analysis, and mathematical modeling, to improve the fit and wearing comfort of compression shells, thus enhancing treatment precision of compression textiles in CVI prophylaxis and physiotherapy. The developed methodologies in this study will also lay down the foundation for sizing systems for different body parts and purposes.

Methodology

Study sample

The prevalence of CVI increases with aging. A total of 208 volunteers (128 females and 80 males) aged 40–60 years (weight: 63 ± 15.13 kg, height: 162 ± 9.01 cm, body mass index (BMI): 23.82 ± 3.96 kg/m²) were recruited locally in Hong Kong via the convenience sampling method. The sources of recruited subjects were people who visited or studied at the Hong Kong Polytechnic University (PolyU), who had seen or were aware of the recruitment via reading information on “clinical screening of CVI” posted by the researchers in the campus. Diagnosis of CVI was based on clinical examination with direct observation of dilated veins on subjects’ lower limbs and the severity was graded using the Clinical-Etiology-Anatomy-Pathophysiology classification (CEAP) criteria for CVI¹⁰ (Table 1). It was found that among the 208 subjects recruited, 57.7% were suffering from VV on one or both legs, although the degree and affected areas varied. The percentage of subjects with VV among female (60.6%) was slightly higher than that among male (53.8%). The screening results indicated that approximately 60% of the subjects were in CEAP-C2, 38.3% in CEAP-C1, and 1.7% in CEAP-C3 on the day of clinical examination, indicating that the recruited subjects were the appropriate target users of GCSs.

Table 1.

Clinical-Etiology-Anatomy-Pathophysiology classification (CEAP): clinical criterion of chronic venous insufficiency on different symptoms and severities

Anthropometric study

An NX-16 fully automatic body scanner ([TC]², NC, USA) producing a true-to-scale 3D body model was applied in the current anthropometry, which is based on white light with grid lines and consists of 16 sensors containing eight cameras to create a 3D point cloud with a density of 600,000–1 million points. Every four sensors are stacked at each corner of the cubic scanning booth to adjust for body levels. The scanner is reported to have low measurement noise and high accuracy (point accuracy: <1 mm; circumferential accuracy: <3 mm).^26–28 The female subjects wore the bras and briefs that they usually wore and males only wore briefs in body scanning. After completing the body mass and stature measurements, the subjects were required to stand still in the scanning booth with their feet at standardized distance (approximately 33 cm) landmarks on the platform. During the scanning, the subjects adopted a normal posture, looked ahead, and breathed naturally.

Match coefficient survey

Before determining the new solutions, a match coefficient survey (MCO%) was conducted on the 208 subjects regarding the four GCS brands from Gloria (I) (Italy), Koolfree (II) (Taiwan), JOBST (III) (USA), and Juzo (IV) (Germany), which are being sold on the local market, offering both knee-high and thigh-high GCSs in the size ranges of XS–XXL with pressures of 15–32 mmHg for treatment of CVI with CEAP C0-C3. The MCO% for a brand was defined as the percentage of the participants’ legs that fitted any one or more sizes in a range at all relevant measurement points.¹⁵ The types of “fitness” were divided into three categories in this study: (a) legs “fitting” at ankle and calf for knee-high styles, or legs “fitting” at ankle, calf, below knee (BK), mid-thigh, and groin for thigh-high styles; (b) legs that fit one or more sizes; and (c) legs that do not fit any sizes. The investigation results are summarized in Tables 2 and 3 and Figure 1, which confirmed our view that developing a new sizing solution was necessary to improve the fitness of compression textiles for the end users.

Table 2.

MCO% and numbers of the tested legs for knee-high graduated compression stockings at relevant landmarks

Brand	Pressure (mmHg)	Increments between sizes (cm)	Legs “fitting” at cB and cC (n, %)							Fit one or more sizes	No fit any sizes
Brand	Pressure (mmHg)	Increments between sizes (cm)		XS	S	M	L	XL	XXL	Fit one or more sizes	No fit any sizes
I	18–21	cB (2) cC (7–9)	n	n/a	115	39	n/a	n/a	n/a	154	56
I	18–21	cB (2) cC (7–9)	%		55%	19%				73%	27%
II	23–32	cB (2) cC (4–5)	n	72	116	62	11	n/a	n/a	261	14
II	23–32	cB (2) cC (4–5)	%	34%	55%	30%	5%			124%	7%
III	15–20	cB (3–4) cC (10–16)	n	n/a	130	124	4	n/a	n/a	258	2
III	15–20	cB (3–4) cC (10–16)	%		62%	59%	2%			123%	1%
IV	18–21	cB (2–2.5) cC (5–9)	n	52	89	71	2	n/a	n/a	214	15
IV	18–21	cB (2–2.5) cC (5–9)	%	25%	42%	34%	1%			102%	25%

Table 3.

MCO% and numbers of the tested legs for thigh-high graduated compression stockings at relevant landmarks

Brand	Pressure (mmHg)	Incrementsbetween sizes (cm)	Legs “fitting” at cB, cC, cD, cF, and cG (n, %)							Fit one or more size	No fit any size
Brand	Pressure (mmHg)	Incrementsbetween sizes (cm)		XS	S	M	L	XL	XXL	Fit one or more size	No fit any size
I	18–21	cB (2) cD (7–9) cF (8–12) cG (9–14)	n	15	79	29	n/a	n/a	n/a	123	87
I	18–21	cB (2) cD (7–9) cF (8–12) cG (9–14)	%	7%	38%	14%				59%	41%
II	23–32	cB (2) cC (4–5) cG (8–10)	n	57	44	29	5	1		136	74
II	23–32	cB (2) cC (4–5) cG (8–10)	%	27%	21%	14%	2%	0%		65%	35%
III	15–20	cB (3–4) cC (10–16) cG (21–24)	n	n/a	130	123	4	n/a	n/a	257	2
III	15–20	cB (3–4) cC (10–16) cG (21–24)	%		62%	59%	2%			122%	1%
IV	18–21	cB (2–2.5) cD (6–10) cF (15–19) cG (17–22)	n	60	92	63	4	1	n/a	220	37
IV	18–21	cB (2–2.5) cD (6–10) cF (15–19) cG (17–22)	%	29%	44%	30%	2%	0%		105%	18%

Figure 1.

Mapping of the studied subjects’ ankle, calf, and groin sizes and Cir size ranges of the four tested brands at the ankle and thigh regions.

Determination of key anthropometric landmarks

In the new solution, the “circumference” (Cir) along the landmarks of the lower limb and the “height (Hei)” (i.e. perpendicular distances) between the landmarks and the floor were considered to be equally important to define the global shapes and sizes of lower limbs. The circumferences and vertical heights from the floor along the 12 key anthropometric landmarks of lower limbs were determined, as shown in Figure 2. The determination method of the landmarks referred to the standards of GAL-GZ387/1 (Germany).

Figure 2.

Key anthropometric landmarks for body shape and sizing (c means circumference, e.g. cB, circumference of ankle B; and l means perpendicular height from the floor, e.g. lC, the height of the calf circumference relative to the floor. cY: heel girth; cA: length from heel to forefoot; cZ: length from heel to toe; cW: girth of forefoot. The foot size is for a selection of graduated compression stockings with closed or open-toe styles. AK section: above-knee part; BK section: below-knee part. a indicates the level of the floor.

Determination of the “alignment” of lower extremities

Three-dimensional digital body scanning demonstrated diverse silhouettes in multiple views. The studied CVI sufferers may also experience other lower limb diseases, such as osteoarthritis (OA) or knee arthralgia, resulting in malalignment of the lower extremity (i.e. knock-knees is common term for the X-shape leg, while bow-legs tends to refer to the O-shape leg). The leg malalignment would cause variations in medial-lateral shell fabrics stretching and length equilibrium in wear, thus negatively impacting pressure dosage delivery. Therefore, the “alignment” assessment of lower limbs was involved in the new sizing system to distinguish body shapes at the primary step. The “alignment” status is an intuitive parameter reflecting the contour of the lower limb, which has been proposed as an indicator for lower limb injuries.²⁹ In clinics, the “Q-angle” is a commonly used index to define the alignment of lower limbs^30–32 via defining internal anatomic structures. Instead of following the traditional concept and end-use of “alignment”, a new parameter has been created, named the “A-angle” in this study, as shown in Figure 3, which is worked as an approximate alignment indicator of the “groin-ankle” segment of the lower limb based on 3D surface-body scanning techniques to assist achieving body shape (silhouette) classifications purposes.

Figure 3.

Alignment determination of lower limb via the “A-angle” for body shape (silhouette) classification purposes.

The three key reference lines and three key midpoints are adopted to determine the “A-angle” in the coronal plane of the point cloud image by 3D digital body scanning. Taking the right leg, for example, (Figure 3(a)), Line (i) connects the bottom point of crotch “T1” and the point “T2”. The point “T2” is the intersection point of Line (i) and the outline of the scanned body at the thigh part; Line (ii) horizontally passes through the mid-patella, which intersects with the outlines of the scanned body at points “K1” and “K2” of the medial and lateral sides of the knee section, respectively; Line (iii) horizontally passes through the thinnest girth of the ankle, which intersects with the outlines of the scanned body at “A1” and “A2” of the medial and lateral sides of the ankle section, respectively. Following the above, the three key midpoints are determined accordingly, that is, “T”, “K”, and “A”, locating along the three defined horizontal Lines (i), (ii), and (iii), respectively. The thigh–knee connection line (TK-L) can be further defined by linking the midpoints “T” and “K” of Lines (i) and (ii), and the knee–ankle connection line (KA-L) can be defined by linking the midpoints “K” and “A” of Lines (ii) and (iii). Then the angle formed between the connection lines of TK-L and KA-L represents the determined “A-angle”. The magnitudes, and positive or negative values of the A-angles reflect the alignment status of the lower limbs, for example, the closer to “zero” of the A-angle, the more aligned or “straight” are the lower limbs; with greater positive or negative values of the “A-angle”, the more the leg is subjected to “knock-kneed” or “bow-legged” shapes. For subjects with excessive “A-angle”, customized GCSs would be recommended, or more attention should be paid by nursing staff to determine their fitting of sized GCSs in practice.

Mathematical simulation of quantitative relationships between Cir and Hei

The morphology of lower extremities is dominated by both circumferences (Cirs) and perpendicular heights (Heis). However, the lower extremities with the similar Cirs may differ in Heis between individuals. These differences could result in variations of pressure gradient distributions by compression interventions, thus negatively influencing treatment efficacy. With consideration of interactions between Cir and Hei, the Cir was regarded as a function of the anthropometric values of Hei at the landmarks in this study. The variation trends of the Cir in terms of the independent variable Hei can be predicted via appropriate interpolation functions. Interpolation is a mathematical solution of constructing new data points within the range of discrete known data points, thus generating a formula that produces a graph passing through a given set of points for prediction and assessment.³³ The established interpolation function curves can reflect interactive characteristics between Cir and Hei under specific body shapes. These morphologic characteristics present a continuous variation in the 3D lower limb. The continuous variation can be simulated by applying available continuous numerical interpolation methods, including the basic linear interpolation (linear), the piecewise cubic spline interpolation (spline), and the shape-preserving piecewise cubic interpolation (pchip), etc.³⁴ MATLAB is the language of technical computing providing us with an interactive environment for numerical computation, visualization, and programming. In this study, MATLAB was employed to define the most appropriate numerical interpolation to quantify interactive characteristics between Cirs and Heis of the lower extremities (Figure 4).

Figure 4.

The Cir–Hei variation trends (Curves_Cir–Hei) of the specimens via multiple interpolation simulation methods.

Compared to the interpolation functions of “spline” and “linear”, the Curve_Cir–Hei of “pchip” closely approximated the real scanned anthropometric data to reflect the true body shape. Therefore, the “pchip” was adopted in this study to define quantitative relationships between Cir and Hei of the lower limbs for all specimens. The “pchip” curve is a cubic interpolation polynomial function between two landmarks and is continuously differentiable at every landmark with shape preserving and following monotonicity by appropriate slope selection. It well reflects the quantitative relationships between the studied parameters of Cir and Hei.

Determination of the “turning point” of the silhouette

The method shown in Figure 4 to apply the Curve_Cir–Hei to determine specific body shapes with describable features was of essential importance. For GCS size selection, the Cirs of cB, cB1, cC, cD, cE, cF, and cG are the seven major recognized landmarks in practice (Figure 2). Their girths and the relative positions of each girth level reflect body shape variations. Therefore, the “segment slope” (SS-Value) between each landmark and the “difference value” (D-Value) between the adjacent segment slopes were adopted in this study to determine the “turning point”, which is the most sensitive reflection of the lower body shape changes, especially at the BK portion. Figure 5 and formulas (1) and (2) show the coordinates of the studied landmarks along the Curve_Cir–Hei and the calculation methods of the SS-Value and D-Value relating to the landmarks

SS - Value = \frac{Δ y}{Δ x} = \frac{y_{2} - y_{1}}{x_{2} - x_{1}}

(1)

D - Value = | \frac{Δ y_{i}}{Δ x_{i}} - \frac{Δ y_{i - 1}}{Δ x_{i - 1}} | = | \frac{y_{i} - y_{i - 1}}{x_{i} - x_{i - 1}} - \frac{y_{i - 1} - y_{i - 2}}{x_{i - 1} - x_{i - 2}} |

(2)

where (x₁,y₁), (x₂,y₂), (x₃,y₃), and (x_i, y_i) are the coordinates of the studied landmarks. The calculation results showed that the highest D-Value was at the landmark cC with the maximum Cir for both males and females among the studied 208 subjects (Figure 6), indicating that point cC was the “turning point” of the outline shape of the lower limbs, which is therefore adopted as the remarkable reference in following leg shape classifications.

Figure 5.

Determination of the “SS-Value” and “D-Value” of the lower limb silhouette.

Figure 6.

Determination of the turning point along the tested landmarks of the lower extremities.

Shape classification of the lower extremities

Compared with the thigh-high GCS, the knee-high one was more popularly used in CVI treatment based on the clinical reports due to the convenience of use and cost effectiveness. However, there was no evidence on any significant difference between the two types of GCSs in correcting venous pathology,³⁵ while the ill-fitting thigh-high GCSs may cause pooling of blood lower in the legs.³⁶ To create an improved sizing system with considerations of diversities in leg shapes, the lower extremities were divided into the two parts in this study, that is, “above knee” (AK) and “below knee” (BK) (Figure 2), in order to enhance the fitness of both knee-high and thigh-high compression products.

Shape classification of the BK part

The angle between the adjacent segment slopes around the turning point cC determines the shape variations of the BK part. Due to the fact that cB is the determinant point for pressure classifications in GCS selection,¹⁰ cB is therefore adopted as the “adjacent” point of cC instead of cB1 in this study. Then the interpolation curves passing the landmarks of cD, cC, and cB become the basis of shape classification in the BK part. No serious swelling (edema) at the ankle is a prerequisite to ensure that Cir at the ankle is less than that at the calf and knee. Considering the monotonicity of the cosine function, the cosine values of the formed angles by the segments cB–cC and cC-cD (i.e. $∠ BCD$ ) range from 0 to π, which is used to define the leg shapes of the BK section in this study. The value of cos $∠ BCD$ can be calculated via formulas (3) and (4), as shown below

cos ∠ BCD = \frac{| BC |^{2} + | CD |^{2} - | BD |^{2}}{2 | BC | | CD |}

(3)

where

| BC | = \sqrt{| cC - cB |^{2} + | lC - lB |^{2}} | CD | = \sqrt{| cC - cD |^{2} + | lC - lD |^{2}} | BD | = \sqrt{| cC - cB |^{2} + | lD - lB |^{2}}

(4)

According to the monotonically decreasing property of the cosine function at the interval [0, π], the greater the cos $∠ BCD$ values, the smaller the values of angle $∠ BCD$ , implying that the BK section will present larger Cirs values at the calf (cC), and smaller ones at the ankle (cB) and BK (cD), which can be classified as Shape-I, that is, a “diamond shape” (Figure 7(a)). If the cos $∠ BCD$ values are smaller, then the values of $∠ BCD$ will be greater, indicating that the Cirs of the BK (cD) are closer to that of cC, leading to Shape-III, that is, an “inverted trapezoids shape” (Figure 7(c)). The intermediary values of cos $∠ BCD$ can be defined as Shape-II, named the “balanced shape” (Figure 7(b)). The rational value ranges for each type of BK shape will be further determined via the clustering method based on 3D anthropometry.

Figure 7.

Shape classifications at the below knee (BK) part.

Shape classification of the AK part

The morphologic studies demonstrated that similar calf shapes presented distinct thigh shapes. Therefore, a new parameter, named the “gradient” (G_EG), between the landmarks cE and cG is employed to determine the variations of outline shape in the AK part of the lower extremities, which can be calculated by formula (5), and the range of G_EG is [0, K2]

G_{EG} = \frac{cG - cE}{lG - lE}

(5)

where cG and cE denote Cir at the E (patella) and G points (groin); and lG and lE denote the vertical heights (Hei) of the landmarks E and G from the floor. The larger the value of G_EG, the wider the groin part in the AK section, and vice versa (Figure 8).

Figure 8.

Shape classifications at the above knee (AK) part: (a) the AK part with a lower gradient (G_EG) (i.e. less slope); (b) the AK part with a higher gradient (G_EG) (i.e. greater slope).

Normality test on body shaping parameters

Before employing the clustering method to categorize the shapes of the lower limbs, normality tests on the studied parameters (cos $∠ BCD$ and G_EG) are required. The normal quantile–quantile plot (Q–Q plot) is an effective diagnostic tool for checking the normality of samples with n independent observations, which, however, is insufficient to provide conclusive evidence that the normal assumption is appropriate. Therefore, the Shapiro–Wilk test³⁷ was performed by SPSS software (23.0; IBM SPSS) to validate the calculated values of cos $∠ BCD$ and G_EG in the statistical analysis. The results showed that both cos $∠ BCD$ and G_EG values followed the normal distributions, as shown in Figures 9(a) and (b).

Figure 9.

Normality tests for values of cos $∠ BCD$ and G_EG. Q–Q plot: quantile–quantile plot.

Clustering method for shape classification

Clustering analysis is a class of techniques used to classify cases into groups (“cluster”) that are relatively homogeneous within themselves and heterogeneous between each other on the basis of a defined set of variables.³⁸ In this study, the K-means clustering method was adopted to further stratify the lower body shape based on the calculated cos $∠ BCD$ and G_EG values. Considering the operability in practice, three body types for the BK part (i.e. K = 3 for cos $∠ BCD$ ) and two body types for the AK part (i.e. K = 2 for G_EG) were determined in body shape classification.

Size classification of the lower extremities

For the same leg shape, the interpolation Curve_Cir–Hei (pchip) would display similar variation trends reflecting quantitative relations between Cir and Hei. Therefore, 3D body shape-driven sizing will be more efficient as compared with the traditional sizing system, which was merely based on separate Cir and Hei values without considerations of the shape factor.

For size classification, different methodologies can be used, including the following.

Sizing with a regularly specific increment. For example, the size system adopted by RAL-GZ387/1 is a representative one, which employs 1 cm as the increment from sizes 18 to 27 for almost all the tested landmarks along the lower limbs, without considerations of leg shape discrepancies and characteristics.

Sizing with an irregularly specific increment. Most of the commercial size charts adopt this method to select GCSs. Different brands applied different increments.

Sizing with the multiple percentiles method (SMPM). This solution is adopted in this study to build up a body-shape-driven sizing system, including the numerical values of “minimum”, “trimean”, “maximum”, and multiple “percentiles”, which can be further optimized based on the served population in the specific market or areas.

Work-flow of the S⁴ based on 3D anthropometry

Figure 10 depicts the work-flow of the established S⁴ based on 3D anthropometry. The “A-angle” is employed to distinguish the fabrication types of compression shells, for example, customized or non-customized GCSs. The aligned lower limb will be categorized into different body shapes in terms of cosine angle values at turning point (cC) of the BK part, based on which, the sizing system will be further determined via the SMPM, and the “gradient” values in the AK part will integrate with shape-driven sizing to improve the fit of both knee-high and thigh-high compression modalities.

Figure 10.

Work-flow of the stratified shape-driven sizing system based on three-dimensional body anthropometry.

Data analysis

Descriptive statistics was performed, including the mean, trimean, maximum, minimum, and standard deviation, in order to present substantial data characteristics. The percentile values were adopted to quantify the sizing classifications of the stratified body shapes. A paired sample t-test was used to determine if any statistically significant difference existed between body shapes, male and female, for the studied key landmarks at anatomic positions. The significance level (ɑ) was set at 0.05. All statistical analyses were conducted using SPSS for Windows (Version 23.0; IBM SPSS).

Results

General profile of 3D anthropometry

Figures 11(a) and (b) illustrated the general profile of anthropometries of the studied subjects for the tested landmarks along the lower extremities. It can be seen that the average Cir values (cm) at the key positions were 20.85 cm (±1.66) at the ankle (cB), 27.26 cm (±3.40) at the brachial (cB1), and 35.62 cm (±3.06) at the calf (cC) for the BK part; the corresponding Hei values (cm) at the ankle, brachial, and calf were 11.23 cm (±1.32), 19.77 cm (±1.11), and 31.06 cm (±1.88), respectively. Greater variations in dimensions were observed at cB1, cB1C-1, and cC of the BK and at cF and cG of the AK.

Figure 11.

General profiles of (a) Cir (cm) and (b) Hei (cm) for all the tested landmarks along the lower extremities.

Fitness survey

The percentages and numbers of subjects “fitting” each size for the studied four GCS brands are shown in Tables 2 and 3 and Figure 1. In general, the investigated subjects “fitting” size ranged from XS to L. The thigh-high GCSs had lower match coefficients (MCO%) than the brands’ knee-high ones, for example, the MCO% of “no fit any sizes” ranged from 1% to 41% for the thigh-high GCSs, and from 1% to 27% for the knee-high GCSs. The higher “increments” between sizes generally raised MCO%, for example, brand III, while they may conversely increase the variations of delivered pressure dosage in practical wear.³⁹ The MCO% values of “fit one or more sizes” were up to 73–124% for knee-high GCSs and 59–122% for thigh-high GCSs, implying that overlap existed between sizes, which may lead to confusion in size determination; hence, a more precise sizing system is highly desirable. Furthermore, as shown in Figure 1, the ankle (cB) sizes displayed more clear-cut Cir (cm) classifications than those at the thigh region (cG) (i.e. more overlapping between sizes at cG), which was consistent with the fact that there are larger variations in Cir (cm) at the thigh than at the ankle, as demonstrated in our present anthropometric study.

A-angle assessment and distributions

Figure 12 illustrates the profiles of the measured A-angle among the tested subjects. Based on the quantitative analysis, as shown in histogram and qualitative observation via 3D scanned images, the tested A-angle (σ) ranging from −10° to 20° can be divided into three segments, that is, σ < −7.0° representing the bow-legged oriented lower body shape (e.g. No. 52104 with σ of −7.38°), σ > 10° representing the knock-kneed oriented lower body shape (e.g. No. 60311 with σ of 18.03°), and −7.0°< σ < 10.0° representing the aligned lower body shape (e.g. No. 60506 with σ of 1.29°). A total 98.08% of subjects (n = 204/208) were involved in the aligned lower limbs for analysis of body shape classifications in this study.

Figure 12.

A-angle assessment and distributions among the tested lower extremities.

Body shape classification

The objective of clustering analysis is to categorize the sample population into homogenous groups. The three clusters as the representatives of basic body types were extracted with considerations of operability in practical sizing buildup. Using the defined cos $∠ BCD$ factor as the independent variable, Figure 13 plots the profiles of the clustered three body shapes with the scales of cos $∠ BCD$ values and subjects’ distributions, for example, cos $∠ BCD$ values ranging from −0.999 to −0.694 describe 42 subjects with inverted trapezoid leg shapes, those ranging from −0.693 to −0.511 describe 80 subjects with balanced leg shapes, and those ranging from −0.508 to −0.258 describe 44 subjects with diamond leg shapes.

Figure 13.

The clustered lower body shapes in terms of cos $∠ BCD$ values.

The detailed properties of the three clusters are described with descriptive statistics in Tables 4 and 5 for all the tested landmarks along the lower extremities. It can be seen that the most significant differences in the Cir values (cm) are in the cB1C-1, cB1C-2, cC, and cF regions among the clustered body shapes (<0.001), and the significant differences in the Hei values (cm) were found in the cF region (<0.005).

Table 4.

Anthropometric analysis on the stratified lower extremities at overall studied anatomic positions in circumference dimensions

Anatomic position	Diamond				Inverted trapezoid				Balanced				F	Sig. P
Anatomic position	Max.	Min.	Mean	S.D.	Max.	Min.	Mean	S.D.	Max.	Min.	Mean	S.D.	F	Sig. P
cB	25.95	18.06	21.63	1.76	24.30	16.74	20.13	1.55	24.31	17.86	20.93	1.58	7.903	.001**
cB1	33.46	24.18	28.43	2.07	32.64	21.11	25.65	2.34	56.54	21.40	27.62	4.11	7.174	.001**
cB1C-1	53.20	28.19	34.33	4.76	41.35	24.12	29.88	3.10	44.34	25.51	31.52	2.78	15.817	.000***
cB1C-2	44.55	30.59	37.04	3.16	40.01	26.87	32.50	2.54	44.34	29.13	34.67	2.55	25.701	.000***
cC	46.10	32.70	38.45	3.33	40.65	28.00	33.34	2.51	41.08	30.86	35.63	2.29	34.689	.000***
cD	43.69	30.49	35.80	3.08	40.68	27.07	32.67	2.64	39.52	28.00	34.15	2.43	12.694	.000***
cE	43.06	32.22	37.29	2.56	44.45	30.71	35.60	2.70	43.12	29.94	36.36	2.66	3.600	.030*
cEF1	47.67	35.27	40.04	2.80	47.15	28.27	37.09	3.74	45.36	29.24	38.60	3.29	7.099	.001**
cEF2	51.70	36.20	43.13	3.53	49.20	33.37	39.76	3.42	48.35	30.37	41.13	3.55	8.229	.000***
cEF3	56.39	37.49	46.61	4.29	52.80	35.74	42.97	3.76	51.46	34.24	44.48	3.76	7.881	.001**
cF	59.76	41.68	50.29	4.09	56.14	38.65	46.30	3.80	55.67	37.25	47.58	3.97	9.256	.000***
cG	68.43	48.04	56.45	4.35	66.32	45.43	54.04	4.26	75.43	39.65	54.82	5.40	2.196	.115 –
cY	38.95	24.35	31.84	2.79	33.55	26.36	29.52	1.67	35.26	25.87	30.56	2.06	10.572	.000***
cA	19.00	14.00	16.85	1.28	18.00	14.50	15.94	0.93	20.00	14.00	16.42	1.25	5.264	.006 –
cZ	27.19	15.48	22.29	2.83	25.66	16.04	21.36	2.02	25.49	14.98	21.56	2.05	1.701	.186 –
cW	28.04	13.26	21.03	3.11	23.13	13.80	19.35	2.11	26.33	1.10	19.34	4.01	3.036	.051 –

***P < 0.001; **P < 0.005; *P < 0.05.

Table 5.

Anthropometric analysis on the stratified lower extremities at overall studied anatomic positions in height dimensions

Anatomic position	Diamond				Inverted trapezoid				Balanced
Anatomic position	Max.	Min.	Mean	S.D.	Max.	Min.	Mean.	S.D.	Max.	Min.	Mean.	S.D.	F	Sig.
lB	13.25	9.50	11.55	1.03	15.00	8.75	10.88	1.3093	15.81	8.75	11.29	1.44	2.398	.094
lB1	22.75	17.75	19.95	1.05	21.25	18.00	19.65	0.7694	23.75	17.00	19.73	1.29	.682	.507
lB1C-1	26.10	21.12	23.66	1.12	24.84	21.80	23.51	0.8171	27.45	21.26	23.51	1.19	.257	.774
lB1C-2	30.42	24.50	27.37	1.43	29.79	24.88	27.36	1.1481	31.14	24.03	27.28	1.44	.073	.929
lC	35.76	27.74	31.09	1.87	34.81	27.70	31.22	1.5948	34.88	26.67	31.06	1.91	.108	.898
lD	40.61	32.74	36.86	2.22	40.83	30.93	35.26	2.2110	41.01	31.48	35.98	2.41	4.216	.017
lE	48.93	38.74	44.49	2.59	48.03	36.93	42.42	2.6214	50.89	38.67	43.54	2.85	5.224	.006
lEF1	53.54	42.74	48.94	2.70	52.68	42.91	47.66	2.4206	54.89	42.83	48.12	2.81	2.043	.133
lEF2	58.33	46.74	53.08	2.92	57.33	46.80	50.91	2.7378	58.89	46.03	51.87	3.06	4.834	.009
lEF3	63.11	50.74	57.23	3.22	61.98	46.91	54.17	3.6480	62.89	47.42	55.62	3.61	6.673	.002**
lF	68.01	54.74	61.54	3.53	66.67	52.93	59.66	3.3781	67.48	53.79	60.22	3.42	2.758	.067
lG	77.46	62.74	69.09	4.08	76.43	60.79	67.73	3.5736	76.81	60.77	67.96	3.72	1.338	.266

***P < 0.001; **P < 0.005; *P < 0.05.

Figure 14 demonstrates the representative Curve_Cir–Hei simulated by the interpolation function. It can be seen that the Curve_Cir–Hei of the “diamond” leg shapes presented obvious apophysis at the ankle–knee segments (Figures 14(a) and (b)), while it showed more gentle curves in those of the inverted-trapezoid leg shapes (Figures 14(e) and (f)). The “balanced” leg shapes displayed moderate fluctuations, as shown in Figures 14(c) and (d). The simulated Curve_Cir–Hei provides an intuitive approach to profile lower body shapes with varying Cir and Hei values.

Figure 14.

Simulated Curve_Cir–Hei by the interpolation function for the three different stratified lower limb shapes: (a), (b) diamond shape; (c), (d) balanced shape; (e), (f) inverted trapezoid shape.

Table 6 illustrates that the G_EG values varied from 0.33 to 1.19 among the clustered body shapes in the AK section. The higher G_EG is greater than the lower one by 46–66% in trimean values for the three clustered body shapes. Higher G_EG reflected wider opening and a lower one reflected a narrow opening at groin.

Table 6.

The clustered gradients (G_EG) along the AK part

Body shape	Cluster	N	$\bar{x}$	S.D.	CV (%)	Percentiles
Body shape	Cluster	N	$\bar{x}$	S.D.	CV (%)	Min.	25th	Trimean	75th	90th	97.5th	Max.
Diamond	Lower G_EG	32	0.66	0.05	8%	0.52	0.58	0.65	0.70	0.73	0.75	0.76
Diamond	Higher G_EG	10	0.90	0.08	9%	0.79	0.87	0.83	1.03	1.08	1.10	1.11
Balanced	Lower G_EG	51	0.63	0.09	14%	0.33	0.43	0.64	0.65	0.71	0.74	0.75
Balanced	Higher G_EG	29	0.87	0.11	12%	0.77	0.87	0.85	1.08	1.14	1.18	1.19
Inverted trapezoid	Lower G_EG	24	0.64	0.07	11%	0.47	0.54	0.66	0.68	0.72	0.75	0.75
Inverted trapezoid	Higher G_EG	20	0.85	0.09	11%	0.77	0.84	0.89	0.99	1.03	1.06	1.06

(a-1), (a-2) Diamond lower leg with higher and lower G_EG; (b-1), (b-2) balanced lower leg with higher and lower G_EG; (c-1), (c-2) inverted trapezoid shaped lower leg with higher and lower G_EG.

Figure 15 shows the comparison between males and females for the Cir and Hei values (cm) at all the tested landmarks along the lower extremities. It can be seen that most significant differences exist around the mid-thigh (cEF1–cEF3) in Cir (<0.001) and at the thigh (cG), mid-thigh (cF), knee (cE), BK (cD), and ankle (cB) in their Hei values (<0.001). Whether the statistical significant differences in body dimensions lead to significant differences in pressure delivered still needs to be determined via objective pressure assessment, which would provide an evidence-based reference for a decision if gender difference is shown to impact the course of compression product design and development.

Figure 15.

Comparison of the tested anatomic positions in Cir and Hei values between males and females.

Buildup of the shape-driven sizing system

The size system based on the three body shapes is established via the multiple percentiles method, as shown in Table 7 (1–3). The seven key anatomic positions are determined in the newly created size chart following the widely accepted measurements in commercialized products. The seven size increments are defined from size-I (min.) to size-VII (max.) based on the stratified lower body shapes of the study subjects. For similar ankle–knee segmental shapes, the Cir (cm) values corresponding to the clustered lower and higher G_EG are provided for knee–thigh sections for increasing the fitness of thigh-high GCSs in choice. The established new size system clearly distinguished the size characteristics and differences between lower body shapes. For example, the ankle with similar sizes of Cir (cm) (e.g. 18.0–18.9 cm) presented greater Cir (s) in the calf (31.9 cm) of the diamond shaped leg but less in the inverted trapezoid shaped leg (29 cm). With the increase in sizes, the differences from the ankle to the calf in the AK part become greater for the diamond-shaped leg (13.4–17.6 cm) than for balanced leg shape (12.1–15.3 cm) with respect to a similar ankle Cir. The different dimensions in the BK part, especially at the turning point cC, may enable the tourniquet effect created by traditional GCSs to be avoided. In general, the size ranges from I to VII in the present study are located in the ranges of XS–L compared with those applied in the commercial compression products, as shown in Figure 1. Large-scale populations involved in the anthropometry would allow the differences in body sizes among the three clustered body shapes to be more obvious.

Table 7.

Stratified shape-driven size system

Cir (cm) with increased sizes (min., percentiles, Trimean, max.)
Positions	Min.		25%		Trimean		75%		90%		97.5%		Max.
Positions	I		II		III		IV		V		VI		VII		Hei (cm)
(1) Diamond
cB	18.5		19.1		20.3		21.5		22.5		22.7		22.7		10.6
cB1	23.1		27.5		29.2		31.0		32.1		32.9		32.9		19.5
cC	31.9		33.5		36.0		37.2		40.2		40.3		40.3		29.0
cD	29.6		31.6		33.9		36.1		37.3		37.4		37.4		34.6
cE	32.2^a	32.8^b	34.9	34.9	37.2	37.4	39.1	40.4	40.4	41.9	40.9	42.8	41.2	43.1	41.9
cF	41.7	45.2	38.9	45.9	48.9	52.0	54.3	56.1	56.8	58.3	58.1	59.4	58.5	59.8	57.6
cG	48.0	52.0	52.0	56.1	54.7	58.8	56.9	64.3	58.7	66.8	59.6	68.0	59.9	68.4	64.6
(2) Balanced
cB	16.7		18.9		19.9		20.8		21.6		23.6		24.1		10.5
cB1	21.7		24.8		26.2		27.5		28.9		31.6		33.4		19.8
cC	28.2		31.0		32.7		34.0		35.5		39.2		39.4		30.6
cD	27.3		30.7		32.3		33.7		35.4		39.2		39.4		34.8
cE	29.9^a	32.3^b	33.2	34.9	35.9	37.0	39.8	40.2	41.8	41.8	42.8	42.6	43.1	42.9	42.0
cF	37.3	43.4	41.4	46.5	46.2	49.9	49.7	52.6	52.1	54.4	53.4	55.4	53.8	55.7	58.5
cG	39.7	51.2	48.6	55.4	52.6	57.9	63.7	66.5	66.2	71.9	67.4	74.5	67.9	75.4	66.7
(3) Inverted trapezoid
cB	18.0		19.3		20.2		21.4		22.5		24.4		24.8		10.4
cB1	23.7		26.1		27.4		28.9		32.5		34.9		39.8		19.5
cC	29.0		33.0		34.7		36.5		37.8		39.9		40.8		29.8
cD	27.4		31.7		33.3		35.5		36.3		38.0		39.3		34.2
cE	30.7^a	33.1^b	34.1	35.4	35.2	35.9	39.9	41.0	41.3	43.1	42.0	44.1	42.2	44.5	41.8
cF	38.7	42.9	43.0	45.9	45.8	47.1	51.8	52.0	53.8	54.4	54.7	55.7	55.0	56.1	58.5
cG	45.4	52.9	50.7	54.8	52.8	57.5	61.1	58.5	59.6	64.2	60.2	65.8	60.4	66.3	65.9

The Cir value (cm) at cE, cF, cG positions with lower G_EG, that is, narrower openings at the thigh parts.

The Cir value (cm) at cE, cF, cG positions with higher G_EG, that is, wider openings at the thigh parts.

Discussion

New parameters and elements created in the S⁴ system

Three new parameters were introduced to classify body shapes in this study, namely the “A-angle” for assessing the “alignment” of lower extremities, “cosine $∠ BCD$ ” for BK shape determination, and “gradient G_EG” for AK shape classification. The clustering analysis quantitatively segmented the sample population into three homogenous groups (i.e. diamond, inverted trapezoid, and balanced leg shapes) based on the mathematical simulation. The landmark cC was determined as the “turning point”, sensitively reflecting shape variations along the key circumference levels of the BK section. The introduction of the “gradient” (G_EG) provided a new solution to differentiate AK dimensions into higher and lower G_EG clusters based on the stratified body shapes. Based on the widely recognized landmarks adopted by standard GAL-GZ387/1 (Germany), the new body shape-driven sizing (S⁴) methodology added two more landmarks between the brachial and calf segments (cB1C-1, cB1C-2) and three more between the knee and mid-thigh (cEF1, cEF2, cEF3) (Figure 2) to refine the shape and size characteristics of the studied lower bodies, which is expected to be a simple but more effective way to improve the selection of compression products without altering the long-term formed size selection habits much.

To date, some brands have achieved the necessary segmentations in size system for GCS selection, for example, the Gloria medical stocking (Gloria Med S.P.A, Italy) has sub-segmentations of “balanced” and “max.” under each S, M, L, and XL category in product selection. However, it is still a size-centered system without involving shaping elements. Although the size, to a great extent, reflects body morphologies, the spatial variations of circumferences along the height directions of the lower limbs still lack sufficient expression, for example, the lower body with similar girths at the ankle, calf, and thigh may cause different fit and pressure profiles in practice. The malalignment of the lower limbs could adversely impact the equilibriums of dimensions and stretching of compression shells along and around the bodies. This is one of reasons for introducing the “A-angle” as the first step of GCS selection in the new S⁴ system. In addition to malalignment, the complications of swelling and lymphedema may also seriously deform the morphologies of single or double lower limbs, resulting in asymmetry in legs, which will be included into the “customized-made” category of the new sizing system to ensure fitness in use; for the asymmetric legs without serious deformations or swelling, users still can follow the established sizing system to determine their suitable sized legwear for left and right legs individually, since most of the GCSs with knee-high or thigh-high styles have separate left and right sides within one pair for possible selection. Meanwhile, for the newly introduced “A-angle”, we will attempt to adopt multi-midpoints to define the TK-L and TA-L lines via regression methods in our following studies to further validate the reference connection lines that represent the alignment of lower extremities by the 3D “surface-body” scanning method in future studies.

Value of the tested samples

Whether the sizing systems appearing in the existing compression product market are based on healthy or patient bodies is uncertain. Based on our knowledge, the subjects participating in the present anthropometry were diagnosed as CEAP C1-C3. Three categories may not fully represent complex lower body shapes, while considering the production cost and operability in practice, the three categories used in this study was the most efficient cluster number with sufficient participants located in shape groups when compared to two, four, or five clusters. In addition, the recruited subjects are high-risk CVI sufferers and are also the targeted consumers for the purchase and use of GCSs; therefore, the anthropometric data obtained in this study are considered to be highly applicable.

Meanwhile, it is noted that measuring a circumference is sensitive to the measuring plane angle just like cross-sectional areas vary with the cut plane angle. It is possible for the circumferences to be measured differently according to leg posture. In this study, the standardized cross-sectional cut plane set by the NX-16 scanning system was used. All the subjects were required to be positioned in a standing posture with separated feet on preset footprints, so as to standardize the leg posture and to avoid any anomalies occurring in the scanned image at the crotch area. The slightly separated legs in the body scan may result in potential deviations in measurements compared to the values obtained when subject to being in a standard upright standing position. Therefore, an illustration presenting the body measurement condition is highly necessary when a sizing chart is introduced to the users for sizing choice.

The stratified body shapes by the S⁴

It was found that the lower limbs with similar ankle girths (e.g. 21 cm) may not find proper sizes for the calf or thigh.¹⁴ For example, the lower limb with a “diamond” calf shape could produce a tourniquet effect when wearing the normal sized GCS, since the calf girths in the diamond shaped leg may exceed the range of standardized settings (e.g. ranges of calf girths in a diamond shape are 38.45 ± 3.33 cm, those in the balanced shape are 35.63 ± 2.29 cm, although both lower limbs have similar ankle girths of 21 cm, as shown in Tables 4 and 5), which may generate negative pressure gradients from the ankle to the calf (e.g. lower pressure at the ankle but higher at the calf), thus decreasing blood flow rather than enhancing it.³⁹ If the size chart adopted wider calf ranges in the setting, a wider variation in pressure dosage will be delivered, which could decrease the accuracy of compression therapy. It was found that changes in the lower limb girths by only 1.5–2.3 cm would vary pressure dosages by 7–24%, as demonstrated by Thomas et al.⁴⁰ Therefore, the created 3D shape-driven sizing (involving clustered “diamond”, “inverted trapezoid”, and “balanced” leg shapes) is an optimal way to improve fitness in practice.

Our studies of the “MCO%” survey obtained consistent findings with those conducted by Macintryre et al.¹⁵ The slope degrees along the thigh contour increase the difficulties in fitness of the thigh part. Compression shells with wider sizes would tend to slip down smaller thighs, while those with smaller sizes would produce a tight-bandaging effect, thus impeding blood circulation. The parameter of “gradient” (G_EG) adopted in this study can be used to refine the thigh morphologies, for example, the ankle–calf with similar sizes may require GCSs with different thigh dimensions for size choice.

Considering that the length of the compression garment (e.g. GCS) would influence the matching of longitudinal pressure gradient distribution with the corresponding leg segments, the 166 subjects among the 208 subjects studied with similar groin length (lG) ranging from 60–70 cm fall into the three leg shape cluster analysis in this study, so as to raise the precision of sizing ranges. The other 42 subjects with lG greater or less than 60–70 cm will be involved in separate sizing systems following the same (S⁴) methodology.

Conclusion

In this study, three parameters, namely the “A-angle”, “cosine angle value”, and “gradient”, are introduced to the new S⁴ via 3D digital anthropometry and mathematical analysis to refine classification of lower body morphologies and to improve size selection of compression textiles. Mathematical simulation via the interpolation function is employed to explore characteristics of shape variations with the involvement of dynamic interactions of both circumferences (Cir) and heights (Hei) of the lower extremities. The clustering analysis quantitatively segmented the sample population into three stratified leg shapes (i.e. diamond, inverted trapezoid, and balanced leg shapes), in which the circumferences of the brachial (cB1), calf (cC), and thigh (cF) exhibited the most obvious differences among the clustered lower limbs. The developed new sizing system is also a starting point of the one-stop quick response manufacturing process for compression textiles (Figure 16). The input anthropometric data will be analyzed by the established S⁴ system before the fabrication and delivery of new textiles, thus promoting fitness and treatment accuracy of the compression textiles in use. With the accumulation of a users’ database, the data pool may vary, including parameter values of the A-angle, cosine values of the turning point angles, and the gradient. Therefore, further identifications and validations of the shape-driven sizing system are required in a large-scale survey or for a specific consumers’ market, so as to ensure the categorized lower limb shapes and sizes work effectively for diverse populations.

Figure 16.

Application of the new stratified shape-driven sizing system (S⁴) in a one-stop quick response model. 3D: three-dimensional.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Innovation and Technology Fund (ITF) of the Hong Kong SAR Government through the collaborative research project ITP/049/14TI and by the Hong Kong Polytechnic University through the start-up research project 1-ZE7K.

References

Heather

. The science of compression therapy for chronic venous insufficiency edema. J Am Coll Clin Wound Spec 2009; 1: 20–24.

Shammeri

Aihamdan

Ai-hothaly

. Chronic venous insufficiency: prevalence and effect of compression stockings. Int J Health Sci 2014; 8: 231–236.

Hohlbaum

. Zur geschichte der kompressionstherapie. I. Phlebol Proktol 1987; 16: 241–255.

Partsch

. Compression therapy: clinical and experimental evidence. Ann Vasc Dis 2012; 5: 416–422.

Mosti

Partsch

. Duplex scanning to evaluate the effect of compression on venous reflux. Int Angiol 2010; 29: 416–420.

Lattimer

Kalodiki

Kafeza

et al.

Quantifying the degree graduated elastic compression stockings enhance venous emptying. Eur J Vasc Endovasc Surg 2014; 47: 75–80.

Tandler

. Challenges faced by healthcare professionals in the provision of compression hosiery to enhance compliance in the prevention of venous leg ulceration. EWMA J 2016; 16: 29–33.

Raju

Hollis

Neglen

. Use of compression stockings in chronic venous disease: patient compliance and efficiency. Ann Vasc Surg 2007; 21: 790–795.

Winslow

Brosz

. Graduated compression stockings in hospitalized postoperative patients: correctness of usage of size. Am J Nurs 2008; 108: 40–45, 47–51.

10.

Liu R, Guo X, Lao T, et al. A critical review on compression textiles for compression therapy: textile-based compression interventions for chronic venous insufficiency. Text Res J. 2017; 87: 1121–1141.

11.

CEN European Prestandard: Medical compression hosiery. Management Center of the European Committee for Standardization, Brussels, August 2001, pp.1–43.

12.

Veraart

JCJM

Neumann

. Pressure differences of elastic compression stockings at the ankle region. Dermatol Surg 1997; 23: 935–939.

13.

Partsch

Braun

. Interface pressure and stiffness of ready made compression stockings: comparison of in vivo and in vitro measurements. J Vasc Surg 2006; 44: 809–814.

14.

Maclellan

. Compression profiles of antiembolic stockings. Aust NZ J Phlebol 2002; 6: 9–14.

15.

Macintyre

Kent

McPhee

. Do anti-embolism stockings fit our legs? Leg survey and data analysis. Int J Nurs Stud 2013; 50: 914–923.

16.

Sheldon

Stevens

Tucker

. The varieties of human physique: An introduction to constitutional psychology, New York: Harper & Brothers, 1940.

17.

Simmons

Istook

Devaragjan

. Female figure identification technique (FFIT) for apparel. Part I: describing female shapes. JTATM 2004; 4: 1–16.

18.

Chen

. Analysis of upper physical characteristics based on angle measurements. Text Res J 2011; 81: 301–310.

19.

Hrženjak

Doležal

Ujević

. Sizing system for girls aged 13-20 years based on Body types. Text Res J 2015; 85: 1293–1304.

20.

Faust

Carrier

. A proposal for a new size label to assist consumers in finding well-fitting women’s clothing, especially pants. Text Res J 2009; 79: 1446–1458.

21.

Braganca S, Arezes PM and Carvalho M. An overview of the current three-dimensional body scanners for anthropometric data collection. In: Arezes PM, et al. (eds) Occupational safety and hygiene III. London: Taylor & Francis Group, 2015, pp.149–153.

22.

Treleaven P and Wells J. 3D body scanning and healthcare applications. Computer 2007; 40: 28–34.

23.

Istook

Hwang

. 3D body scanning systems with application to the apparel industry. J Fashion Market Manag 2011; 5: 120–132.

24.

Choi

Ashdown

. 3D body scan analysis of dimensional change in lower body measurements for active body positions. Text Res J 2011; 81: 81–93.

25.

Song

Ashdown . Categorization of lower body shapes for adult females based on multiple view analysis. Text Res J 2011; 81: 914–931.

26.

D’Apuzzo N. Recent advances in 3D full body scanning with applications to fashion and apparel. In: Gruen A and Kahmen H (eds) Optical 3-D measurement techniques IX. Vienna, Austria, 2009.

27.

Yu W. Development of a three-dimensional anthropometry system for human body composition assessment. PhD Dissertation, The University of Texas at Austin, August 2008, pp.29–32.

28.

NX 16 Tutorial guidance. http://apparel.edgl.com/Media/DocumentLibrary/tc2.pdf (accessed 12 October 2016).

29.

Nguyen

Boling

Levine

et al.

Relationships between lower extremity alignment and the quadriceps angle. Clin J Sport Med 2009; 19: 201–206.

30.

Ilahi

Kohl

. III Lower extremity morphology and alignment and risk of overuse injury. Clin J Sport Med 1998; 8: 38–42.

31.

Powers

. The influence of altered lower-extremity kinematics on patellofemoral joint dysfunction: a theoretical perspective. J Orthop Sports Phys Ther 2003; 33: 639–646.

32.

Horton

Hall

. Quadriceps femoris muscle angle: normal values and relationships with gender and selected skeletal measures. Phy Ther 1989; 69: 17–21.

33.

Epperson JF. Interpolation and approximation. In: Epperson JF (ed.) An Introduction to numerical methods and analysis. Hoboken, New Jersey: John Wiley & Sons, 2013, pp.171–262.

34.

Siam J. Numerical analysis. In: Kahaner D, Moler C and Nash S (eds) Numerical methods and software. Upper Saddle River, New Jersey, USA: Prentice Hall, 1988, pp.238–246.

35.

Sajid

Tai

NRM

Goli

et al.

Knee versus thigh length graduated compression stockings for prevention of deep venous thrombosis: a systematic review. Eur J Vasc Endovasc Surg 2006; 32: 730–736.

36.

Hameed

Browse

Immelman

et al.

Should knee-length replace thigh-length graduated compression stockings in the prevention of deep-vein thrombosis. S Afr J Surg 2002; 40: 15–16.

37.

Ghasemi

Zahediasl

. Normality tests for statistical analysis: a guide for non-statisticians. Int J Endocrinol Metab 2012; 10: 486–489.

38.

Landau S and Everitt B. A handbook of statistical analyses using SPSS. Boca Raton, Florida: Chapman & Hall, CRC Pressure LLC, 2004, pp.309–339.

39.

MacLellan

. Compression profiles of anti-embolic stockings. Aust NZ J Phlebol 2002; 6: 9–14.

40.

Thomas

Toyick

Fisher

. Graduated external compression and the prevention of deep vein thrombosis, Bridgend: Surgical Materials Testing Laboratory, 2000.

Stratified body shape-driven sizing system via three-dimensional digital anthropometry for compression textiles of lower extremities

Abstract

Keywords

Methodology

Study sample

Anthropometric study

Match coefficient survey

Determination of key anthropometric landmarks

Determination of the “alignment” of lower extremities

Mathematical simulation of quantitative relationships between Cir and Hei

Determination of the “turning point” of the silhouette

Shape classification of the lower extremities

Shape classification of the BK part

Shape classification of the AK part

Normality test on body shaping parameters

Clustering method for shape classification

Size classification of the lower extremities

Work-flow of the S4 based on 3D anthropometry

Data analysis

Results

General profile of 3D anthropometry

Fitness survey

A-angle assessment and distributions

Body shape classification

Buildup of the shape-driven sizing system

Discussion

New parameters and elements created in the S4 system

Value of the tested samples

The stratified body shapes by the S4

Conclusion

Footnotes

Declaration of conflicting interests

Funding

References

Work-flow of the S⁴ based on 3D anthropometry

New parameters and elements created in the S⁴ system

The stratified body shapes by the S⁴