Abstract
To investigate the effect of structured bras and soft bras on breast shape, 46 female participants (Caucasian, BMI < 30, aged 18–45) were recruited for three-dimensional (3D) scanning. Participants were scanned in three conditions: wearing a provided structured bra, a provided soft bra, and nude. The impact of the bras on breast asymmetry was quantitatively studied. The change in breast shape and position from the nude condition to the condition when shaped by the bras was also explored. Contour maps that show the topographic shapes of the scans were generated to analyze these comparisons. Thirty-five measurements were extracted from spider web structures that were derived from the contour maps, and were used for statistical analysis. Eight measurements were found to be especially indicative of the shape variations introduced by the bras. Regression models were built to predict the in-bra shape given only the nude breast shape. Lastly, heat maps that visualize the shape variations from the nude-to-bra condition via colors were plotted on the surfaces of the 3D scans of the participants in bras, and were used for qualitative analysis. This study is helpful in understanding how bras interact with breast tissue, and can provide useful information for the improvement of bra designs for enhanced fitting or desired shaping effects.
Bras have become an important item of clothing for modern women. A well-designed bra can enhance a woman’s body contour, body attractiveness, and self-confidence. 1 It has been reported that the support provided by a bra is essential to prevent the elongation of the Cooper’s ligaments, which results in breast ptosis. 2 In addition, a well-fitted sports bra can provide constraints on the breasts, thus reducing the embarrassment and the symptom of breast pain caused by excessive breast movement during exercise. 3 The importance of bra fit and pressure comfort has been widely acknowledged. On the other hand, the design of a well-fitted, comfortable bra, or a bra with the desired shaping effect, requires knowledge of the three-dimensional (3D) shape of the breasts and, furthermore, the shape modification that the breasts can experience under the compression of a bra.
Although research has been intensively done to study breast shapes, or to evaluate the fit of bras,4–9 limited information was found that objectively assessed the impact of bras on breast shape. Scurr et al. 10 measured the nude breasts of 13 participants and studied the measurement changes over time after bra removal. An increase in the sternal notch to nipple distance (right: 2.8 mm, left: 3.7 mm) and a decrease in the height of the nipple (right: 4.1 mm, left: 6.6 mm) were found, whereas the nipple-to-nipple distance and the breast projection remained the same over time. The researchers, however, did not report the measurements of the breasts when bras were worn. Ashizawa et al. 11 compared the measurements obtained from the breasts of 11 Japanese participants with and without bras. They found an increase in the bust girth (range: 2–6 cm, mean: 3.4 cm, one outlier removed) and a decrease in the underbust girth (range: 0–6 cm, mean: 1.9 cm, one outlier removed) after a bra was put on. Four participants were removed due to limitations in measuring the height of the underbust line, but the other seven participants showed an increase in the line height after the bra was worn (range: 0.5–5 cm, mean: 2 cm). They also found that the most prominent point on the left breast moved upward (range: 0.3–3 cm, mean: 2 cm) and forward (range: 0.1–1.6 cm, mean: 0.9 cm), and that the distance between the left and right prominent points became wider with the bra on (range: 0–3 cm, mean: 1.4 cm). However, in another study where researchers investigated how respiration affects the calculated bra size, a mean difference of 1.9 inches (range: 0–6 inches) in bust girth, and a mean difference of 0.75 inches (range: 0–2 inches) in underbust girth between the two respiration states (i.e. inspiration and expiration) were reported. 12 Since the differences in circumferences between the nude breasts and the in-bra breasts can be very small, and respiration might play a significant role in those minor differences, the circumferential or other types of traditional measurements may not be the best tool to investigate the change in the shape of breasts provided by bras.
In order to fill the research gap, in this study, we endeavor to gain a deeper understanding on how bras interact with breast tissue. This can provide useful information for the improvement of bra designs. Also, most anthropometric studies conducted with 3D body scanners provided female participants with unstructured bras; however, this is not the type of bra worn by most women on a daily basis. Understanding the change in shape to a more commonly worn structured bra style can add value to these scan studies. This paper presents the results of an in-depth study regarding the shape differences between the nude breasts and the same breasts wearing either a structured bra or a soft bra. The following are the research questions to be answered.
What changes does the structured bra or the soft bra make in terms of breast asymmetry? What kind of measurements can effectively detect the shape differences between the nude breasts and the breasts with the structured bra/soft bra on? Are there any patterns in the shape modification of breasts, caused by the structured bra/soft bra? What is the relationship between the shape modification and the nude-breast shape? Is it possible to predict the in-bra shape given only the nude-breast shape? How can the shaping effect of bras be represented visually?
Methodology
The styles of the provided bras
The 3D scans obtained were processed in MATLAB®, with head, limbs, and the section below the underbust line removed. The remaining scan will be referred to as the bust scan. The locations of bust points and underbust level were identified visually.
Because the difference in breast shapes between the nude scan and the in-bra scan may not be large enough to be detected using traditional measurements, such as bust girth (in some cases there is very little difference visible), the topographic shapes of the breasts derived from their contour maps were used for analysis (Figure 1(a)). Scans were processed so that each scan had exactly 18,000 points on 100 horizontal slices (or transverse planes), while each slice contained 180 points, at the angles (with respect to the x-axis) of every 2° from –180° to 180°. The points were sorted in the same way for every scan. A total of 35 comparative measurements were then derived from these spider web structures, as shown in Figure 1(b). Pairwise Hotelling's T2-tests were performed to see whether the 35 measurements as a whole can detect the differences in shape. Post-hoc tests with Bonferroni correction were conducted to identify the specific measurements that are sensitive to shape modification. In addition, Asymmetry Indices
13
of the nude breasts and the in-bra breasts were calculated to quantify the impact of the bras on the breast asymmetry. Moreover, the relationships between the nude breast shape and the changes in shape provided by the structured bra and the soft bra, respectively, were investigated. Regression prediction models were then built to predict the in-bra shape given only the nude breast shape.
Breast topography shown by a contour map and the corresponding web structure (note: the black points at y = 0 are the bust points): (a) Contour map; (b) Spider web structure.
To visualize the changes in shape, for each participant, the 3D scan of the nude breasts was first aligned with the in-bra scan (either structured bra scan or soft bra scan) and the two scans were merged together. The alignment was based on the central axis (determined by averaging the x-coordinates and y-coordinates, respectively, of all the points on the bust scan) and the bust plane (defined by the averaged z-coordinates of the right and left bust points). Figure 2(a) shows the top view of the scans where the nude scan is presented in black while the in-bra scan is in red. Then a heat map was generated and presented on the surface of the in-bra scan (Figure 2(b)) by comparing and assigning a color for each pair of corresponding points from the two scans. This was done by calculating and recording the horizontal distance toward the central axis of each point. The difference in the point-to-axis distances between a point on the in-bra scan and its corresponding point on the nude scan can then be calculated (the nude scan served as the base case). After the calculation, red/orange colors were assigned to points where an increase in the distance, that is, expansion, was observed. Similarly, blue colors were assigned to points where a decrease in the distance, that is, compression, was observed (see Figure 2(b)). Another way of presenting this calculation result is by a 3D colored mesh (Figure 2(c)), in which case both the colors and surface heights were determined by the calculated difference values. The surface height can be positive, shown as upward protrusion, or negative, shown as downward indentation. The scan surface heat maps and the contour maps provide the basis for both quantitative analysis and qualitative analysis of the shape changes.
Visualization of the shape differences between scans (Color online only): (a) Top view of merged scans; (b) Scan surface heat map; (c) 3D colored mesh of difference value.
Results and discussion
Impact of bra on breast asymmetry
The Asymmetry Index (ASI) values of the nude-breast scans, the structured bra scans, and the soft bra scans were calculated and the overall distributions were plotted in a boxplot (Figure 3(a)). The median ASI values are 0.0536, 0.0237, and 0.0232 for the nude scan, the structured bra scan, and the soft bra scan, respectively (larger values indicate increased asymmetry). Both the structured bra scans and the soft bra scans have significantly lower ASI values compared with the nude scans (p-values < 0.0001 for both). This shows that, in general, both structured bras and soft bras can help to improve the breast symmetry. In addition, the ranges of the ASI values reduced for both of the in-bra cases. The standard deviation reduced from 0.0167 to 0.0097 (structured bra) and 0.0077 (soft bra).
Asymmetry Index (ASI) comparison: (a) Overall distributions; (b) Pairwise ASI difference between structured-bra scan & nude scan; (c) Pairwise ASI difference between soft-bra scan & nude scan.
Two representative cases to demonstrate the impact of bras on the breast asymmetry
Notes. A higher Asymmetry Index (ASI) value suggests a larger extent of breast asymmetry.
The left and right breasts are aligned based on the z-coordinates of their bust points (rather than aligned as they are on the body).
According to Figure 3(b), compared with their nude scans, 44 of all the structured bra scans (95.7%) show a decrease in the ASI value, that is, an improvement in the breast symmetry, while two (4.3%) show an increase in the ASI value, that is, a decline in the breast symmetry. Seven cases (15.2%) have changes (either increase or decrease) that are less than 0.01, meaning the changes are so trivial that they can be ignored. However, the decrease in the ASI value can be as much as 0.0912, and there are 10 participants (21.7%) whose ASI values reduce more than 0.04. According to observation, a significant improvement in the breast symmetry (where the decrease in the ASI value exceeds 0.06) tends to happen to nude breasts that are severely uneven. On the other hand, the increases in the ASI values are always less than 0.011.
For the soft bra scans (Figure 3(c)), the distribution looks very similar to that in Figure 3(b). Forty-four scans (95.7%) show an improvement in the breast symmetry, while two (4.3%) show otherwise. Four cases (8.7%) have a trivial increase or decrease in the ASI value. The decrease in the ASI value can be as much as 0.0817, and there are 11 participants (23.9%) whose ASI values reduced more than 0.04. Similar conclusions can be drawn as well. A significant improvement on the breast symmetry tends to happen to nude breasts that are severely uneven, while a decrease in the breast symmetry is very rare and is a trivial change (always less than 0.01).
Analysis of the extracted measurements
Extracted from the web structures (Figure 1(b)), the 35 measurements include the following: (a) the overall radius (the median value of all radii); (b) 12 radii (the median values among layers) at 12 different angles (the included angles between the x-axis and one of the auxiliary lines); (c) 12 between-layer radius increments at the 12 angles (one increment value at an angle); (d) the median perimeter of all the 11 rings (each layer is essentially a ring); (e) the overall perimeter ratio between adjacent layers (the median value of 10 ratios: outer-ring-perimeter to inner-ring-perimeter); (f) the median area of all the 11 rings; (g) the overall area ratio between adjacent layers (the median value of 10 ratios: outer-ring-area to inner-ring-area); and (h) six diagonal radius differences (e.g. radius at 30° minus radius at 120°). 13
Hotelling's paired T2-tests
Pairwise post-hoc tests with Bonferroni correction (partial)
Note. * implies statistically significant at a 0.05 level; more * suggests significant at a lower level (0.01 or less).
As shown in Table 4, some measurements appear to be insignificant for both dataset pairs (e.g. #15, #34). Some appear to be significant only for one of the dataset pairs (e.g. #7, #14). However, there are a few measurements that are significant for both dataset pairs at an extremely low significance level: #20, #21, #22, #23, #24, #25, #26, #28.
Measurements #20–#25 are the between-layer radius increments at 30°, 60°, 90°, 120°, 150°, and 180°, divided by the corresponding median radius (i.e. Measurements #2–#13), respectively (see Figure 4(a) and the Appendix as supplementary material). Clearly, the shape modification caused by bras mostly happens at the upper section of the breast. Measurement #26 is the overall perimeter ratio between layers (Figure 4(b) only presents one of the 10 perimeter ratios; the median value of the 10 ratios was used in the analysis), and Measurement #28 is the overall area ratio between layers (Figure 4(c) only presents one of the 10 area ratios; the median value of the 10 ratios was used in the analysis). Those two overall ratios depend highly on the fullness and the extent of the protrusion of the breast. As demonstrated in Figure 4(d), the perimeters and the areas (of cross-sections) drop more dramatically from the base to the top for a tall cone than for a short cone. They drop more gradually for a hemisphere than for a cone. Therefore, the two ratios will decrease if the breast protrusion reduces but the fullness of the breast increases. This is exactly what can be observed from the majority of the in-bra scans, compared with their corresponding nude-breast scans. Clearly, under the compression of a bra, the breast becomes less prominent and less conical.
Demonstration of significant measurements: (a) Radius increments between adjacent layers at 60° (note: this figure only shows 4 layers, and 3 increments. In fact, there are 11 layers and 10 increments. The median value of all the 10 increments was obtained then divided by the median radius at the same angle); (b) Perimeter ratio between layers- Outer ring (black) to Inner ring (red); (c) Area ratio between layers- Outer ring (black) to Inner ring (red); (d) The ratios depend on the fullness and the degree of protrusion.
Relationship between the nude shape and the shape modification
The difference values between the in-bra (structured bra or soft bra) data and the nude-breast data were calculated for each of the 35 measurements. The difference values will be referred to as the structured bra/soft bra shape modification values. The correlations between the shape modification values and the nude-breast data were calculated as well. A substantial portion of measurements show strong negative correlations for the structured bra modification values: 27 out of the 35 measurements (77.1%) have correlation values (with the nude-breast data) that are less than –0.5 (negatively correlated); 18 measurements (51.4%) have correlation values that are less than –0.7 (strongly negatively correlated). Fewer measurements show strong correlations with the nude-breast data for the soft bra modification values: 18 measurements (51.4%) have correlation values that are less than –0.5 (negatively correlated); nine measurements (25.7%) have correlation values that are less than –0.7 (strongly negatively correlated). In terms of the aforementioned significant measurements (#20, #21, #22, #23, #24, #25, #26, #28), all of them have strong correlation values for both the structured bra and the soft bra modification values (median: –0.81, minimum: –0.92, maximum: –0.63).
In-bra shapes predicted by the nude-breast shape based on the regression models
Qualitative analysis
Two representative cases to demonstrate the shaping effect of bras
According to the observation of all 46 participants, most of the structured bra contour maps show a certain degree of lift-up effect. Most soft bra contour maps appear to be flattened to a higher extent vertically than horizontally (see Case II in Table 6). The lift-up effect can be more easily observed in the heat maps. For the structured bra heat maps, red/orange colors, which imply expansion, can usually be found at the upper section of breasts, while blue/cyan colors, which imply compression, can usually be found at the lower section (Table 6 and Figure 2(b)). This shows that the whole breasts had been pushed up to a higher level. In addition to push-up, a push-in effect can be observed for some participants (Case I in Table 6, and Figure 2(b)). Their breasts were pushed slightly toward the center (the push-in is not large enough to form cleavage). Although only the same style of bra was used during scanning, many other participants do not show a push-in effect at all. Their breasts were pushed upward but also rotate toward the sides (Case II in Table 6). It appears that the same style of bra can have a different shaping effect on different people, but this could also be due to how the bras were donned (some people may prefer to move their breasts by hand toward the center intentionally).
The red/orange colors at the central bridge area are usually caused by the gap between the bra and the skin surface, rather than by the shape modification of breasts. This pattern in the heat map happens with the soft bra more often than with the structured bra, but if the structured bra is poorly fitted and the bridge is not in contact with the skin, this will appear in the heat map (Figure 2(b)). Among the soft bra heat maps, the typical shapes of the red/orange area are the triangle (Case II in Table 6) and inverted trapezoid (Case I). Observation shows that if the two breasts are already fairly close together on the torso without the bra, the inverted trapezoidal shape will be more likely to appear, but since most nude breasts are separated, the triangular shape shows up more often. On the other hand, because the soft bras used in this study were soft and highly stretchable, there are a few instances that do not have the red/orange area. These instances show mild yellow at the bridge area, suggesting the gap between garment and skin is small (some other cases show no gap at all). These are usually the cases where breast size is not as large.
There are three participants who have very flat breasts. They do not have much deformable tissue on their breasts. Therefore, not only are their nude breast shapes quite different from the rest of the participants, but also the manner in which the bras influence their breast shapes is also very different, and inconclusive. It is hard to tell whether the red/orange areas are caused by expansions in the shape or by garment gaps. There is no general trend that can be summarized from the contour maps or the heat maps, due to the limited number of participants with flat breasts and the large differences in bust shapes among them.
Lastly, high asymmetry in the color distribution, which represents shape variations in the breast, can be observed for some cases (e.g. high expansion appears only on one side of the breasts). This is as expected because, in general, both the structured bras and the soft bras can improve the breast symmetry to some extent (as demonstrated in the Impact of bra on breast asymmetry section), and the improvement in symmetry could mean equalization when there was initially more variation on one breast than the other.
Conclusions
To investigate the impact of structured bras and soft bras on the breast shape, this study recruited 46 female participants (Caucasian, BMI < 30, aged 18–45) for 3D scanning. Participants were asked to wear a structured bra, a soft bra, and no bra during scanning. Three scans were obtained for each participant and were processed in MATLAB®. Contour maps that show the topographic shapes of the breasts were generated and studied. Thirty-five measurements were extracted from the spider web structures simplified from the contour maps, and were used for statistical analysis. Scan surface heat maps that represent the shape modification via colors were adopted for qualitative analysis. This study is very useful for researchers to understand how bras interact with the breast tissue, and for bra designers to improve their designs to achieve desired shaping effects.
The impact of bras on breast asymmetry was firstly explored. The breast asymmetry was quantified by calculating the ASI. A higher value in the ASI implies a lower degree of symmetry between the left and right breasts. The overall boxplot shows that, in general, both structured bras and soft bras can help to improve breast symmetry. The ASI differences between a nude scan and one of the two in-bra scans were also calculated for each individual participant. The individual analysis shows that, for both types of bras, a significant improvement in the breast symmetry is more likely to happen to nude breasts that are severely asymmetric.
More quantitative analysis was done on the unconventional measurements. The results of the pairwise Hotelling's T2-tests show that the 35 measurements are able to detect the shape modifications. The post-hoc tests with Bonferroni correction revealed eight measurements to be especially significant. Among them, six measurements are the radius increments between adjacent layers at certain angles. They all relate to the angles on the upper section of the breast. The other two measurements are the overall perimeter ratio and the overall area ratio between adjacent layers. It was found that under compression of a bra, as the breast becomes less prominent and less conical, both ratios will decrease. Furthermore, to study the relationship between the nude shape and the shape modification introduced by the bras, the correlations between the shape modification values (calculated by subtracting the nude-breast data from the structured bra/soft bra data) and the nude-breast data were obtained for each measurement. A substantial portion of measurements show strong negative correlations, including all of the eight significant measurements. In addition, regression prediction models were built to predict the in-bra shape given only the nude breast shape. Similarities can be observed between the predicted shape and the true shape.
As for the qualitative analysis, according to observations, most of the contour maps of the soft bra scans indicate that the breasts are flattened to a higher extent in the vertical direction. Both the contour maps and the scan surface heat maps show that the breasts had been pushed up by the structured bras. Some participants’ breasts were also slightly pushed in, whereas many other participants do not show a push-in effect at all. A gap between the bra and the skin surface can sometimes be observed in the heat maps of the soft bras. The shape of the gap relates to how separate the breasts are.
In this study, we concentrated on comparisons between the nude breast shape and the structured bra shape, and between the nude breast shape and the soft bra shape. Similar methods could be used to investigate changes in shape between bra styles, by comparing the structured bra to the soft bra. This could result in methodologies to convert scan data from the many body scan studies conducted with soft bras to more useful structured bra shapes.
There are several limitations in this study. Firstly, even though the scan surface heat map can be a good visual tool to evaluate shape modifications on the torso, it may not work well for the neck and the shoulders. The generation of the heat map relies highly on the alignment of scans and is very sensitive to displacement and posture changes. Although participants were asked to keep the same posture during scanning, slight posture changes cannot be avoided, especially for the arms. The placement of arms also directly influences the shoulder areas on the scan. Perfect alignment at shoulder areas and armhole areas is almost unachievable. This is the reason why some abrupt and irregular dark red or dark blue colors appear at the shoulder and armhole areas, as well as at the neckline area for some participants, on the heat maps. It is also possible that the wearing of different types of bras could be an influencing factor in body posture: the weight distribution of the breasts and the forces generated by the different straps and bands interacting with the body may have an effect on the position of the shoulders and neck. Secondly, this study was unable to draw informative conclusions for the flat-chested population, partly because of the small sample size (only three participants with flat chests were in the sample). Neither quantitative analysis nor qualitative analysis could reach consensus for these three cases. Although the flat-chested population is a minority among the female population, their need for improved bra designs cannot be overlooked. Future study could be done specifically for this population. In addition, the sample size is relatively small. The 46 participants may not be sufficient in capturing all the breast shape variation in the population (for those aged 18–45, Caucasian females, with BMI < 30). Moreover, although the prediction regression models predict the in-bra shapes to some extent, there is room for improvement. This is also due to the limited number of participants. Better regression models can be built with increased sample size. In addition, since the shape modification is related to the nude breast shape, a categorization of the nude-breast shape into different groups before regression may improve the accuracy of the prediction. Also, the regression models depend highly on the specific bra style and the fit of the bras; thus, it cannot be generalized to other designs and styles. Future studies can look into the material properties of the bras and refine the models for an improved generalizability. There also may be variation introduced in the manner in which participants donned the bras; that is, some participants may have manually centered the breast tissue in the bra cup in different ways, resulting in different results. A study in which the bras are donned in a standard manner may result in different outcomes. Lastly, merely comparing breast shape is not sufficient for understanding the relationship of the bust to the torso. Therefore, additional information, for example, the relationship of the bust point to the sternum (a potential measure of how high the bust is lifted by a bra), is needed for bra designers.
Supplemental Material
Supplemental material for Detection and comparison of breast shape variation among different three-dimensional body scan conditions: nude, with a structured bra, and with a soft bra
Supplemental Material for Detection and comparison of breast shape variation among different three-dimensional body scan conditions: nude, with a structured bra, and with a soft bra by Jie Pei, Jintu Fan and Susan P Ashdown in Textile Research Journal
Footnotes
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by Cornell Institute of Fashion and Fiber Innovation (CIFFI).
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
