3D scan-based classification of Chinese young female hand morphology

Abstract

To investigate the changes in hand morphology among young females, researchers employed 3D hand scanning to perform anthropometric measurement of 111 Chinese young women (20–26 years), enabling hand morphology classification for ergonomic applications. A total of 32 hand parts were measured and analyzed based on these models. The findings reveal that variables describing hand morphology are predominantly categorized into four types: finger width, finger circumference, finger length, and hand length. The typical indicators reflecting hand morphological characteristics include hand length, middle finger width, proximal circumference of the index finger, and ring finger length. Results revealed five distinct hand types: short/thin, short/wide, standard, long/thin, and long/wide. Compared to current national standards in China (GB/T 16252 − 1996), modern hand morphology showed significant increases in hand length (+ 3.3%) and metacarpal breadth (+ 8.3%). We propose a novel sizing system (5-size-5-fit) with 180/86 as the predominant type, optimized for ergonomic glove design. This study provides critical data references for the industrial design of hand appliances, while also offering potential implications for ergonomics and hand injury prevention.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-43547-z.

Introduction

Human anthropometric data constitute a fundamental resource for industrial design and ergonomic applications, with hand dimensions being particularly critical for the development of well-fitting hand appliances¹. The human hand, distinguished by its dexterity and functional complexity², exhibits considerable morphological variability across populations due to factors such as geographical region, age, gender, and occupational patterns. In recent years, evolving lifestyles and consumption patterns have heightened user expectations regarding the comfort, fit, and performance of hand-worn products. Nevertheless, many existing gloves and handheld devices exhibit poor ergonomic compatibility, largely due to inadequate or outdated hand sizing references. To address this, in - depth research focusing on specific social groups is imperative for enhancing the fit and comfort of hand appliances.

The measurement of hand morphology is an essential aspect of anthropometry, and also a necessary prerequisite for classifying hand size. However, the complex biological structure of human hand presents challenges for hand morphology research^3–6. Recently, the human torso has been the primary focus of contemporary research on body morphology measurement, while other body parts including the buttocks^7,8, neck^9–11, shoulders¹² and female breasts^5,13, as well as the categorization of the entire human body^14–19, have also garnered significant attention. In contrast, research on hand dimensions and morphology remains insufficient and requires further development. Existing hand anthropometric studies have predominantly centered on male participants²⁰ or employed limited measurement parameters²¹, resulting in a lack of targeted datasets for young female populations. Hand shape standards derived from traditional measurement methods in the past have gradually failed to keep pace with the development and changes of hand shapes^22,23. The current classification standard of hand morphology in China is unclear, so that factories can only refer to the size of past products when designing hand appliances such as gloves, which falls short in meeting the needs of hand appliances’ fitness and comfort for young females. This gap underscores the urgency of conducting updated, population-specific hand morphometric studies. Furthermore, hand morphology varies significantly across regions, ethnicities, ages, and genders, which justifies the need for focused studies on specific groups such as young Chinese females.

Traditionally, hand data was measured using tape measurement and two-dimensional (2D) measurement. While tape measurement was simple, it was time-consuming and inefficient. 2D measurement was relatively faster but had limitations such as complex image processing, great influence of light noise on error, and inability to measure hand circumference and other parameters^24,25. The advent of 3D scanning technology has dramatically improved the data collection phase of anthropometric research, enabling more effective hand data collection, faster acquisition rates, and the simultaneous measurement of length, width, and circumference data. However, the flexible deformation of human skin during 3D scanning introduces gaps between the obtained 3D human model and the parameters of the human body. Furthermore, the accuracy of measurement data obtained by different 3D scanning equipment varies. Compared with the data measurement of the human torso, the feature points of the hand were arranged closely, making the measurement more difficult. The currently used measurement data in GB/T 16,252 − 1996 “Hand Sizing System-Adult"²⁶ was still obtained by tape measurement, which was time-consuming and costly. Researchers frequently used 2D measurement for hand measurement. Li et al.²⁷ implemented the automated measurement of hand size based on the MLS image registration approach. A 2D color scanner was employed to capture color pictures of the hand, however, due to the limitation of 2D images, measurement data such as hand circumference could not be obtained. Fan²⁸ compared three measurement methods—3D scanning measurement, 2D measurement, and tape measurement—when measuring anthropometric data, and found that all hand parameters are suitable for 3D measurement. Evaluations by Xia et al.²⁹ and Kouchi et al.³⁰ found that tape measurement was typically time-consuming, inefficient, and fixed scanner was not cost-effective for small-scale scanning. The handheld scanner was a better choice for measuring anatomical dimensions of hand movement equipment. Ultimately, this study chose a portable 3D scanner as the measuring equipment after considering all variables, including cost, dependability, accuracy, and convenience of use.

Beyond dimensional measurement, understanding hand typology is crucial for ergonomic design. Prior studies have highlighted morphological variations across ethnic groups and identified finger proportion indices as population-specific traits. For example, Imrhan³¹ compared and analyzed 24 hand items of Bangladeshi and other ethnic adults. Results showed that there were significant differences in hand size between Bangladeshis and other ethnic groups, and the differences in hand size between different ethnic groups should be considered in ergonomic design. Zhao^32,33 and Zhou et al.^34–36 conducted a classification study on the measurement data of hand size, results found that the ring index finger length had obvious population characteristics. Such findings underscore the need for demographic-specific hand models. This research focuses on young females as the target population, employing a handheld 3D scanner to acquire high-precision hand morphological data. Guided by the anatomical classification framework of the human torso and supported by 3D measurement technology, a total of 32 hand-related parameters were systematically obtained, including hand length, hand width, and other key dimensional indices. Through cluster analysis (e.g., K-means clustering) and analysis of variance (ANOVA), this study aims to elucidate the morphological variations of modern young females’ hands.

This research holds both technological and practical implications. Technologically, it demonstrates the efficacy of 3D scanning for detailed hand morphometry and provides a methodological reference for future studies. Practically, the derived hand typology and sizing system can directly inform the design of various hand appliances—including protective, athletic, and medical gloves—enhancing comfort, functionality, and injury prevention. Medically, the data may contribute to the design of rehabilitation aids and the study of hand strain related to morphology^37,38. By addressing the morphological specificities of modern Chinese young women, this study aims to bridge the gap between anthropometric research and industrial design^39,40, ultimately contributing to improved product ergonomics and user well-being.

Materials and methods

The measurement was conducted at the Ergonomic Laboratory of the Engineering Training Centre, College of Textiles and Clothing, Nantong University. An Artec Studio non-contact handheld 3D body scanner (Artec EVA, https://www.artec3d.cn/portable-3d-scanners/artec-eva) was employed, featuring an accuracy of 0.1 mm and a resolution of 0.2 mm. This scanner was solely utilized to generate 3D point cloud datasets of hands. With a scanning speed of 16 frames per second (fps), the scanned images were saved in STL format. Post-scanning, the data underwent processing using Artec Studio software (Artec Studio 20, https://www.artec3d.cn/3d-software/artec-studio), and the raw point cloud data was subsequently imported into Geomagic Studio (Geomagic Wrap, https://hexagon.com/products/geomagic-wrap) for further processing.​.

Prior to the experiment, informed consent was duly obtained from each participant, who signed the consent form. The participants’ hands were prepared by removing all unnecessary adornments, wiping away perspiration and oil stains, thereby ensuring that the 3D marking stickers adhered properly. Measurement points were determined based on the anatomical landmarks and descriptions specified in the Chinese national standard GB/T 16,252 − 1996 “Hand Sizing System – Adult,” as well as the commonly used parameters in glove pattern making. Accordingly, 4-mm-diameter, 0.1-mm-thick landmarks were attached to the predetermined locations on the right hand, as shown in Fig. 1(a) and 1(b). The detailed measurement positions are provided in Supplementary Table S1 online. During the scanning process, the participants stretched their hands and gently pressed them against a transparent plate, while maintaining a consistent straightening arc – defined as the inherent curvature formed from the metacarpophalangeal joints to the fingertips when the fingers are fully extended – to avoid measurement errors induced by excessive forward leaning or bending. Subsequently, the experimenter held the scanner and moved steadily around the participant to complete a full 360° scan, following the method shown in Fig. 1(c). After scanning, in Artec Studio, irrelevant point cloud data was eliminated, and outlier noises were removed. The remaining hand point cloud data was then subjected to overall registration and sharp fusion, ultimately yielding unclosed hand point cloud data.​.

Fig. 1.

Schematic diagram of a palm landmarks, b back landmarks, c the scanning method.

In Geomagic Studio, the 3D model of the selected hand was closed (i.e., non-manifold edges repaired), and the optimally scanned 3D hand model was selected for data collection. The slightly raised marker stickers were clearly visible on the model, as shown in Fig. 1(c). The “Draw Curve” tool was employed to measure the anatomical dimensions between two landmarks in various directions. Each measurement was replicated three times, and the average value was calculated for subsequent analyses.

(Generated using Geomagic Wrap, https://hexagon.com/products/geomagic-wrap; and Adobe Photoshop 2023, https://www.adobe.com/cn/.)

To guarantee the reliability of the research findings, the sample size of this study was determined following statistical principles. In industrial production and scientific research, a 95% confidence level is typically adopted for sample size determination. The formula for calculating the minimum sample size is as follows:

1

In this formula, 1.96 denotes the critical value of the standard normal distribution associated with a 95% confidence level; σ represents the population standard deviation, which was determined in this study by referencing hand measurement data of adult Chinese females from the national standard GB/T 16,252 − 1996 “Hand Sizing System - Adult”; and Δ indicates the allowable error, which was set at 2 mm based on relevant research by Lee et al.^41–43 on the accuracy of three-dimensional human skin measurement.

Based on the calculation, the minimum required sample size is 56. To ensure the reliability of the research and mitigate the impact of errors, a total of 111 healthy Chinese female volunteers were ultimately recruited for the experiment. All participants were aged 20–26 years, right-handed, and had no severe injuries or obvious trauma to their right hand.

Ethics declarations

This study was approved by the Ethics Committee of Nantong University and conducted in accordance with the Declaration of Helsinki. All procedures followed relevant guidelines and regulations. Prior to the experiment, participants received a written information sheet detailing the study procedures, and all provided written informed consent. Participants fully understood the research purpose and consented to the use of their data solely for this study.

Results and discussion

Data processing

To ensure the reliability of the experimental results, data accuracy was rigorously verified prior to statistical analysis. A box-plot test was employed to quantify the deviations of the measured values from the mean. The results indicated that Samples No. 72, 3, 11, 12, and 71 exhibited significant deviations from the mean. The original data were meticulously verified to rule out data entry errors and other inaccuracies. These erroneous data points were subsequently rejected in accordance with the established sample criteria. Consequently, the final valid sample size was determined to be 106.

Testing the normality of data is a crucial step in data processing operations since it determines whether the data variables adhere to a normal distribution, allowing for the use of common statistical analysis techniques such as correlation analysis and linear regression analysis¹⁴. The Shapiro–Wilk (S–W) test is suitable for normality test of small data samples (less than 50 rows). While the Kolmogorov–Smirnov (K–S) test is appropriate for large data samples (more than 50 rows). There were 106 rows of valid data in this study, and the normality test results obtained by the S-W test were more inclined. The K–S test compares the frequency distribution f(x) with either the theoretical distribution g(x) or the distribution of two observations, with the null hypothesis is H₀: the two data distributions are consistent or the data conforms to the theoretical distribution. The value of D, where is the observed value, and when D > D (n, a), H₀ is rejected (P0.05), otherwise, it is accepted (P0.05). Results from Table 1 indicate that except for the index finger width (P=0.039) and palm thickness (P=0.010), the P values of other parts were all > 0.05. Hence, the H₀ was not rejected, suggesting the data followed a normal distribution. Moreover, Quantile–Quantile plots (Q–Q plots) and residual distribution diagrams were also used to test the normal distribution of data. The former is a graphical representation where x-axis shows the quantile of the normal distribution, and the y-axis represents the sample quantile. If the points formed by the two axes tend to fall on a straight line on the Q–Q plot, it indicates a linear association between the sample data and the normal distribution, implying the data is normally distributed. The residual distribution diagram represents the difference between the measured and expected cumulative probability. The y-axis represents the deviation between the theoretical and actual values. If the points on the diagram are uniformly scattered on the standard line (Y=0) without significant fluctuations, it indicates normal distribution. The normal Q–Q plots presented in Fig. 2(a) and Fig. 2(c) demonstrate high congruence between the cumulative probability distributions of the index finger width and palm thickness measurements against the theoretical diagonal. Crucially, as evidenced in Fig. 2(b) and Fig. 2(d), maximum residual fluctuations for these parameters remained within 0.6. These collective findings confirm normal distribution adherence for both proximal phalanx width and palm thickness data.

Table 1.

Normality test of measurement parts.

Measurement part	K–S Test			S-W Test
	Statistics	DF	P-value	Statistics	DF	P-value
Little finger length	0.082	106	0.074	0.971	106	0.020
Ring finger length	0.064	106	0.200	0.985	106	0.293
Middle finger length	0.058	106	0.200	0.987	106	0.381
Index finger length	0.064	106	0.200	0.991	106	0.682
Thumb length	0.065	106	0.200	0.992	106	0.771
Dorsal length of little finger	0.072	106	0.200	0.977	106	0.058
Dorsal length of ring finger	0.038	106	0.200	0.989	106	0.542
Dorsal length of middle finger	0.063	106	0.200	0.984	106	0.238
Dorsal length of index finger	0.086	106	0.050	0.980	106	0.111
Dorsal length of thumb	0.081	106	0.087	0.985	106	0.285
Rhizosphere of little finger	0.067	106	0.200	0.984	106	0.250
Rhizosphere of ring finger	0.066	106	0.200	0.98	106	0.112
Rhizosphere of middle finger	0.074	106	0.185	0.967	106	0.010
Rhizosphere of index finger	0.071	106	0.200	0.969	106	0.015
Rhizosphere of thumb	0.048	106	0.200	0.988	106	0.462
Proximal circumference of little finger	0.058	106	0.200	0.991	106	0.687
Proximal circumference of ring finger	0.064	106	0.200	0.989	106	0.563
Proximal circumference of middle finger	0.061	106	0.200	0.992	106	0.802
Proximal circumference of index finger	0.078	106	0.115	0.988	106	0.447
Proximal circumference of thumb	0.059	106	0.200	0.983	106	0.210
Little finger width	0.059	106	0.200	0.990	106	0.592
Ring finger width	0.077	106	0.127	0.983	106	0.206
Middle finger width	0.072	106	0.200	0.987	106	0.365
Index finger width	0.089	106	0.039	0.966	106	0.009
Thumb width	0.04	106	0.200	0.987	106	0.375
Hand length	0.068	106	0.200	0.984	106	0.226
Back length	0.057	106	0.200	0.992	106	0.824
Hand breadth at metacarpal	0.084	106	0.066	0.986	106	0.358
Wrist width	0.086	106	0.052	0.970	106	0.016
Thenar width	0.084	106	0.060	0.979	106	0.091
Purlicue	0.061	106	0.200	0.990	106	0.589
Palm thickness	0.101	106	0.010	0.974	106	0.037

DF: degree of freedom.

Fig. 2.

a Normal Q–Q plot of index finger width. b Residual plot for index finger width. c Normal Q–Q plot of palm thickness. d Residual plot for palm thickness.

Descriptive statistics were utilized to obtain an overall understanding of the data’s basic features, including the degree of concentration, degree of dispersion, and distribution. The mean, standard deviation, skewness coefficient, and kurtosis coefficient were the fundamental descriptive statistical variables used most frequently for this purpose.

Descriptive statistical analysis enables the characterization of fundamental data attributes including central tendency, dispersion, distribution patterns, variation trends, and measurement ranges. The arithmetic mean serves as a key indicator of data location and concentration. It can be seen from Supplementary Table S2 online that the sample measurements exhibited the following variations: hand length ranged from 159.008 mm to 196.537 mm with a mean of 176.721 mm; hand breadth at metacarpal spanned 68.922 mm to 93.738 mm (mean = 82.338 mm); the finger length and width showed considerable variation, ranging from 46.251 to 106.582 mm and 11.177 to 23.409 mm; the proximal circumference of finger ranged between 40.688 and 64.758 mm; and the rhizosphere of finger was between 45.963 mm and 87.557 mm. Critically, the observed ranges for hand length and width comprehensively encompass the specification thresholds defined in the Chinese National Hand Size Standard. The standard deviation, defined as the square root of variance, quantifies data dispersion. Statistical results indicate that most parameters exhibited standard deviations below 5, signifying minimal fluctuations, with the exception of fundamental hand dimensions including hand length, dorsal length of finger, and hand breadth at metacarpal. Kurtosis and skewness coefficients, which statistically characterize distribution steepness and symmetry respectively, converged to 0 across the hand anthropometric dataset. This convergence demonstrates concentrated data distributions, collectively confirming the dataset’s suitability for hand shape classification and applied research endeavors.

Data classification analysis

Statistical Product and Service Solutions (IBM SPSS Statistics 25, SPSS Inc., Chicago, USA) was used to conduct R-type cluster analysis on the 106 data obtained as experimental samples. In order to achieve the purpose of dimensionality reduction, R-type clustering was used to cluster variables with similarities and choose representative variables for analysis from different classes. The analysis was performed using the between-groups linkage method with Pearson correlation coefficients as the measure of similarity, an approach that effectively merges clusters with the smallest average distance between groups and is suitable for identifying representative variable categories. Supplementary Fig. S1 online showed a genealogical map created using R-type clustering of 32 variables. Clustering category was represented on the x-axis, and relative distance (inter-group distance) of each category was represented on the y-axis, reset according to the distance ratio. The similarity between the classifications decreased as the relative distance increased. The optimal number of clusters was determined by analyzing the dendrogram structure. By adjusting the position of the reference line, a distinct separation into four main clusters was observed, where variables within each cluster were closely linked, while clear gaps existed between different clusters. After clustering¹², 32 measurement items were separated into four groups to create clear distinctions between variables.

Class 1 included little finger width, ring finger width, middle finger width, index finger width, thumb width, dorsal length of index finger, dorsal length of thumb. Class 1 called the width classification, mainly reflected the finger width.

Class 2 included proximal circumference of little finger, proximal circumference of ring finger, proximal circumference of the middle finger, proximal circumference of index finger, proximal circumference of thumb, rhizosphere of little finger, rhizosphere of ring finger, rhizosphere of middle finger, rhizosphere of index finger, thumb length, dorsal length of middle finger, wrist width, thenar width, purlicue. Class 2 called the circumference classification, mainly reflected the finger circumference.

Class 3 included the rhizosphere of thumb, little finger length, ring finger length, middle finger length, index finger length, dorsal length of little finger, dorsal length of ring finger, palm thickness, hand breadth at metacarpal. Class 3 called the length classification, mainly reflected the finger length.

Class 4 included the hand length and back length. Class 4 called the hand length, which mainly reflected the total length of the hand.

It could be seen from the above classification that the variables describing the hand morphology were mainly finger width, finger length, finger circumference and hand length.

After cluster analysis, typical indicators were selected from each classification. According to the formula and professional knowledge, the two principles of representative and easy measurement of the selected variables were integrated to determine the representative variables and extract the typical indicators. The eigenvalue of each indicator was calculated, and the equation is

2

Where, r is the correlation coefficient between and other indicators in the same classification, is the number of indicators of the classification where belongs, and the indicator with the largest is taken as the typical indicator.

The values of each indicator in the four classifications were calculated by Eq. (1), and the results were shown in Supplementary Table S3 online. The mean values of in four classifications were compared, the indicator with the largest in each classification was selected as the typical indicator. Finally, the middle finger width, the rhizosphere of index finger, the ring finger length and the hand length were extracted as the typical indicators to describe the hand morphology.

K-means clustering algorithm was applied to quickly analyze typical indicators and identify the characteristics of hand morphology. In this study, the algorithm used Euclidean distance as the similarity metric between samples and aimed to minimize the within-cluster Sum of Squared Errors through iterative optimization. The initial cluster centers were selected using the K-means + + method to reduce sensitivity to initial values. The iteration stopping criteria were set as either the change in cluster center coordinates being less than 10^− 4 or the number of iterations exceeding 500. The optimization algorithm followed these steps: (1) randomly select k samples as the initial cluster class centre; (2) calculate the distance between each sample in the data set and k cluster class centres, and assign them to the corresponding class with the smallest distance; (3) recalculate the distance of cluster class centres for each cluster class; (4) repeat until the number of iterations or the location of cluster class centre remains unchanged. To determine if the mean values of several groups are equal, analysis of variance (ANOVA) was utilized in this study. The null hypothesis is : ===. According to the number of components, analysis of variance may be separated into one-way ANOVA and multivariate analyses (MANOVA) of variance. One-way analysis of variance was utilized in this study since only the effect of typical indicator length on classification of hand size was studied. F-value was calculated by the equation.

3

Where, MSB is the variance of each group’s mean, MSE is variance of each distribution itself.

When F-value is large, it indicates that one or more distributions are far removed from the others, and each distribution is highly concentrated, meaning that the variance of each distribution is minimal. Thus, is rejected. Table 2 showed the ANOVA results of typical indicators obtained when hand data were clustered into 3, 4 and 5 categories according to 4 typical indicators.

Table 2.

K-means analysis of typical indicators.

Categories and items		Clustering		Error		F statistics	P-value
		MSB	DF	MSE	DF
3	Hand length	3.101	2	4.932	103	0.629	0.535
	Proximal middle finger width	286.918	2	18.351	103	15.635	0
	Rhizosphere of index finger	495.018	2	6.590	103	75.117	0
	Ring finger length	2854.452	2	14.108	103	202.329	0
4	Hand length	5.474	3	4.881	102	1.122	0.344
	Proximal middle finger width	447.240	3	11.003	102	40.649	0
	Rhizosphere of index finger	306.305	3	7.352	102	41.664	0
	Ring finger length	1934.588	3	13.316	102	145.280	0
5	Hand length	14.415	4	4.521	101	3.189	0.016
	Proximal middle finger width	343.733	4	10.783	101	31.878	0
	Rhizosphere of index finger	252.739	4	6.513	101	38.803	0
	Ring finger length	1547.797	4	9.612	101	161.023	0

MSB: mean squared between; MSE: mean squared error.

The P-values of the test were calculated using F statistics. As demonstrated in Table 2, when the hand data were clustered into five categories, all P-values were below 0.05. This result indicated significant differences in the mean values across groups, confirming that fast clustering into five categories represented the optimal solution. Based on the clustering outcomes, hand morphology was classified into five types. A detailed analysis of the data at each cluster center was conducted to name and characterize the hand morphologies (Table 3). The most prevalent type, designated as the standard morphology, constituted 63% of the total young female population. Individuals in this group exhibited slender knuckles and aesthetically pleasing hand shapes, with the data for each hand component primarily falling within the range of the mean value. The long and wide morphology accounted for 32%, featuring larger hand dimensions with longer fingers and wider palms. The remaining hand morphologies represented smaller proportions. Analysis showed that due to improvements in living standards, people’s hands have undergone changes, with increased sizes compared to the past. Furthermore, the rise in the proportion of long and wide hand shapes suggested that the current hand morphology has experienced a slight increase in hand length and metacarpal hand breadth compared to historical data.

Table 3.

Classification of hand morphology.

Categories	Hand morphology	Character	Hand sizing	Proportion (%)
1	Short/thin	Short hands and thin knuckles	160/72,160/79	3
2	Short/wide	Short hands and wide knuckles	170/93	1
3	Standard	Normal hands and thin knuckles	170/72,170/79,170/86,180/72,180/79,180/86,190/79	63
4	Long/thin	Longer hands and thin knuckles	200/79	1
5	Long/wide	Longer hands and wide knuckles	180/93,190/86,190/93,190/93,200/86	32

To gain insights into the hand characteristics of young females, the data were compared with the mean and standard deviation of women’s hands in Chinese national standard GB/T 16,252 − 1996 “Hand Sizing System-Adult” (Table 4). The findings suggested growth in the hand length, hand breadth at the metacarpal and index finger length, with a slight reduction in index finger breadth. Specifically, the analysis indicated a 3.346% increase in hand length, 8.339% increase in hand breadth at the metacarpal, 4.832% increase in index finger length, and a 1.624% reduction in index finger width. The increase in hand length and hand breadth at metacarpal further verified the conclusion obtained from the classification of hand morphology, that is, the contemporary hand morphology was larger than before, with thinner knuckles. The widening of palm, increase in hand length, and thinning of knuckles were the primary changes observed in modern hand morphology. It should be clarified that there are differences in age structure between the sample of this study and the national census population of adult females on which GB/T 16,252 − 1996 is based. Therefore, this comparison aims to reveal the contemporary characteristics of this specific young subgroup rather than question the generalizability of the standard. This provides an empirical basis for the potential refinement of size standards for specific populations in the future.

Table 4.

Analysis of basic data changes in female.

Measurement items	Subjects		Hand Sizing System-Adult
	Mean (mm)	Standard deviation	Mean (mm)	Standard deviation
Hand length	176.721	8.259	171.000	7.600
Hand breadth at metacarpal	82.338	5.018	76.000	3.700
Index finger length	69.189	3.733	66.000	3.800
Index finger width	16.724	2.075	17.000	0.900

The optimized design of hand size specifications is largely dependent on the variation characteristics of hand morphology. The rational optimization of hand size specifications is related to the coverage rate of the size and the utilization rate of the size standard. Based on the hand shape classification results and the changes in basic hand dimension data, considering the increase in both hand length and hand width among young females, the hand width was relaxed with a 7 mm grade difference. The optimized hand size specifications were formulated according to the principle of coverage rate ≥ 5%. The optimized hand size specifications for young females are 5-size-5-fit. As can be seen from Table 5, the hand length of most young females was 170 ~ 190 mm, and the hand width was 79 ~ 93 mm, and the size coverage within this range was relatively high. The selection of centre size should be combined with the average value in the high-frequency area⁴⁴. Therefore, 180/86 was selected as the centre size with the highest coverage 22%. Compared with the national standard GB/T 16,252 − 1996 “Hand Sizing System-Adult” in the national adult women’s centre size 170/75, the hand length and width had increased by 5.88% and 14.667%, respectively. Based on the hand data required for producing hand appliances, grading was established for typical indicators (see Supplementary Table S4 online). Combined with the centre value of each cluster in k-means clustering, the middle finger width, the rhizosphere of index finger, the ring finger length were set to 2.5 mm, 5 mm and 4 mm, respectively.

Table 5.

Hand sizing coverage of young female.

Hand length	Hand width
	72 mm	79 mm	86 mm	93 mm	100 mm
160 mm	2%	1%	-	-	-
170 mm	3%	7%	13%	1%	-
180 mm	1%	12%	22%	10%	-
190 mm	-	4%	12%	5%	1%
200 mm	-	1%	5%	-	-

To validate the practical performance of the proposed “5-size-5-fit” sizing system, glove prototypes were developed in accordance with both the new central size and the original central size specified in the national standard. Wearing trials involving participants demonstrated that gloves based on the new central size exhibited superior fit and higher conformity compared to those designed per the original standard, effectively confirming the enhanced adaptability of the new sizing system for hand-worn products.

The basic measuring part of the hand involved in the production and design of hand appliances will be taken as the control part². The correlation analysis between hand length and five finger length, dorsal length of finger, finger width, finger circumference, rhizosphere of finger and other variables indicated that the ratio of ring finger length, middle finger length, index finger length, dorsal length of ring finger and dorsal length of middle finger was greater than 70%. The dorsal length of index finger accounted for more than 60% of the hand length. The ratio of thumb length and proximal circumference of little finger to hand length exceeded 50%. The little finger length, the thumb width, the rhizosphere of index finger, and the proximal circumference of fingers except the little finger occupied relatively low. In general, finger length, dorsal length of finger and hand length had strong positive correlations, while finger circumference, finger width and hand length had slightly lower correlations. During the production and design of hand appliances, the reasonable design of finger length and dorsal length of finger was slightly more important than the finger circumference and the finger width. Based on the control parts data and hand length, the regression equation could be established to calculate the corresponding control part size of a hand size, as shown in Table 6.

Table 6.

Correlation and regression analysis between control site and hand length.

Measurement part	Correlation	Sig	Regression equation
Phalanx thumb width	0.256	0.008	Y1=6.840 + 0.065X
Little finger length	0.441	0	Y2=16.588 + 0.231X
Ring finger length	0.700	0	Y3=12.009 + 0.338X
Middle finger length	0.718	0	Y4=17.928 + 0.333X
Index finger length	0.715	0	Y5=12.073 + 0.323X
Thumb length	0.551	0	Y6=16.365 + 0.238X
Little finger back length	0.384	0	Y7=14.447 + 0.200X
Ring finger back length	0.741	0	Y8=22.775 + 0.389X
Middle finger back length	0.761	0	Y9=28.319 + 0.391X
Index finger back length	0.641	0	Y10=29.052 + 0.330X
Thumb back length	0.372	0	Y11=33.978 + 0.140X
Phalanx circumference of little finger	0.510	0.009	Y12=34.073 + 0.075X
Phalanx circumference of ring finger	0.328	0.001	Y13=34.478 + 0.102X
Phalanx circumference of middle finger	0.306	0.001	Y14=37.708 + 0.102X
Phalanx circumference of index finger	0.274	0.009	Y15=38.450 + 0.093X
Phalanx circumference of thumb	0.343	0	Y16=35.867 + 0.124X

In conclusion, the study provided a methodological reference for accurately obtaining data on hand morphology and obtained results that are of certain practical significance in hand morphology classification, hand model design, and virtual garment fitting. And this classification system can be used to refine hand morphology classification systems, establish basic data size databases of young Chinese women’s hands, and support hand appliance manufacturers. Future research might include additional factors such as the subjects’ region, age, and gender to facilitate comparative analyses.

Conclusion

This study systematically acquired 32 hand morphological parameters from 111 young adult females using a non-contact handheld 3D scanner, followed by classification and optimization of hand dimensions. The key findings are summarized as follows:

1. Hand morphology exhibits significant contemporary evolutionary characteristics. The mean values of hand length and metacarpal hand breadth were 176.721 mm and 82.338 mm, respectively. Compared with the national standard GB/T 16,292 − 1996, these measurements exhibited significant increases of 3.346% and 8.339%, respectively, indicating notable morphological changes in the younger population.

2. The hand morphological structure can be summarized into four key dimensions. R-type cluster analysis revealed four distinct dimensional clusters: finger width, finger circumference, finger length, and hand length. This result underscores the multidimensional and complex structural characteristics of the human hand, necessitating comprehensive parameter consideration in ergonomic design.

3. A scientific classification system of five hand shape types was established. K-means rapid clustering analysis categorized hand morphologies into five distinct types: short/thin, short/wide, standard, long/thin, and long/wide. This classification system provides a scientific basis for understanding inter-individual variations in hand morphology.

4. A 5-size-5-fit hand type system was established with 180/86 as the central dimension. Based on clustering results of key indicators, the grading intervals were determined as 2.5 mm for middle finger width, 5 mm for index finger root circumference, and 4 mm for ring finger length, enabling precise dimensional gradation.

5. Regression equations were developed for 16 key hand dimensions (e.g., thumb width, finger length, dorsal finger length, finger circumference) with hand length as the predictor variable. The significant regression outcomes provide a scientific reference for manufacturers to design hand-held appliances that conform to the morphological characteristics of target users.

In summary, leveraging non-contact 3D scanning technology, this study not only updated the contemporary anthropometric data on hand morphology of young adult Chinese females but also established a systematic design support system for hand equipment encompassing three dimensions: classification, sizing, and prediction. Notably, the methodology adopted herein exhibits a certain level of replicability. Overall, this work possesses certain scientific value and considerable potential for engineering applications.

Notwithstanding the aforementioned contributions, this study has inherent limitations that should be acknowledged. Primarily, the research scope was confined to young adult Chinese females, leading to a relatively narrow sample coverage. For future research, the sample pool could be further expanded to include diverse age groups to establish a more comprehensive database of hand morphology and sizes for Chinese females. This expansion would help enhance the generalizability and applicability of the research findings. Additionally, future research could conduct a comparative analysis of hand data obtained from non-contact 3D scanning and traditional 2D measurement to verify the higher measurement accuracy of the former.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

S–W test

Shapiro–Wilk test

K–S test

Kolmogorov–Smirnov test

Q–Q plots

Quantile–Quantile plots

DF

Degree of freedom

ANOVA

Analysis of variance

MANOVA

Multivariate analyses

MSB

Mean squared between

MSE

Mean squared error

Author contributions

Yanru Zhai: Conceptualization, Methodology, Formal analysis, Writing-original draft.Yongjie Bian: Data collection, Visualization, Writing-original draft.Yue Shen: Validation, Writing-review & editing.Xuefeng Yan: Methodology, Resources provision, Funding acquisition.Xiaoyan Li: Writing–review & editing, Supervision.

Funding

This research was financially supported by the following funding agencies: The Major Projects of the Natural Science Foundation of Jiangsu Province (Grant Nos. 20KJA540001, 22KJA540001). The Emergency Management Science and Technology Project of Jiangsu Province (Grant No. YJGL-YF-2020-5). The Nantong Science and Technology Projects (Grant Nos. JB2022004, MS22021003). The Nantong Basic Science Research Project (Grant No. JC2021041).

Data availability

Data is provided within the manuscript or supplementary information files.

Declarations

Competing interests

The authors declare no competing interests.