114 � 0 0 en Q) C) m E t:: Q) a. 10 8 6 4 2 o JOURNAL OF COSMETIC SCIENCE . . ............ ............ ............ -:· ............ :· ............ :· ............ � . . . . . . - . . . . . . . . . . . . . . : : Ea : : / . /: . . . . . . . . . . ............ � ......... ' ... : ............. : .......... . ··:··/······:············ /. . :/� � /: . : . . . . ............ , ................ . ·························"'························· . . . . . [!I I ·, . /. . . � . . ! : 4: : : : ........ - � 1/-·. riJ· - � - ........... ·· : ........... ·i ........... ) ........... -�: ........... . :a: i ! i m � i : i i : . . . . ' • I • o • . . . . . ' . . . . . --------1-�- ... ------..--------------------- 1.0 1.5 2.0 2.5 3.0 3.5 Svm (mm) 4.0 4.5 10 8 6 4 2 a � 0 0 en C: m Q) � Q) C) "'C ::J --, Q) . m z Figure 4. Quantitative roughness evaluation and human image assessment. The expert grades and nai"ve­ judge mean scores for the ten expert-image grading scale images are plotted versus the quantitative cellulite­ severity parameter Svm. The expert and na · ive scores were highly correlated (r = 0.96) and not significantly different. Both Svm and Sdr had the highest correlations to expert and na · ive scores (r � 0.86). BIOPHYSICAL CHARACTERIZATION OF CELLULITE Fifty-six females completed the study measurements. The mean BMI (kg/m2) was 34.6 ± 8 (range 21-57), and the mean weight was 204 ± 52 (range 128-331). The mean age was 44.3 ± 8 (range 21-60). For the ten males, the mean BMI was 33.2 ± 6 kg/m2 and the mean age was 41.8 ± 11.2 years. Table IV shows the results (mean ± standard error, range) for all of the biophysical evaluations (females and males). Correlations were found between expert image grades and quantitative roughness. Svm, Sz, and Spm had the highest correlations with image score (females), with coefficients of 0.69 (p 0.001), 0.68 (p 0.001), and 0.62 (p 0.001), respectively. The biophysical features that related to cellulite severity were identified from the rela­ tionships among all measurements (Table V, coefficients of �0.4 or higher, p 0.05). Cellulite severity, as measured by surface roughness (Svm) or live visual grade, was correlated to BMI (weight), thigh circumference, percent thigh fat, and the architecture

QUANTITATIVE MODEL OF CELLULITE 115 Table IV Morphological, Biophysical, and Anthropomorphic Measurements of Cellulite Females N = 54 Males N = 10 Mean±SE Range Mean ± SE Range Anthropomorphic data Weight (lb) 205.4 ± 7.0 128-331 225.6 ± 12.3 168-290 BMI 34.9 ± 1.2 21.3-56.8 33.2 ± 1.9 26.8-43.3 Age (years) 44.4 ± 1.2 21-60 41.8 ± 3.5 25-57 Thigh circumference (in) 26.9 ± 0.5 21-35 26.0 ± 0.8 23.0-30.5 Cellulite morphology Visual grade 1.0 ± 0.2 0.0--4.0 0.2 ± 0.1 0.0-1.0 Expert image score 4.7 ± 0.4 0.0-9.0 3.8 ± 0.8 0.0-7.0 Sa 1.1 ± 0.1 0.3-2.1 0.8 ± 0.1 0.4-1.3 Sq 1.4±0.1 0.4-2.6 1.0 ± 0.1 0.5-1.6 Sp 3.9 ± 0.2 1.2-8.7 3.4 ± 0.4 1.8-5.9 Sv 5.2 ± 0.3 1.5-12.9 4.4 ± 0.5 2.7-8.0 St 9.1±0.4 3.1-17.9 7.8 ± 0.8 5.0-12.8 Ssk -0.2 ± 0.1 -1.5-0.8 -0.2 ± 0.1 -1.2-0.1 Sku 3.8 ± 0.2 2.2-7.1 4.2 ± 0.7 2.3-8.9 Spm 2.6±0.1 1.0-4.8 2.1 ± 0.2 1.2-3.4 Svm 2.8 ± 0.2 1.0-7.1 2.3 ± 0.2 1.7-3.6 Sz 5.4 ± 0.3 2.0-11.8 4.4 ± 0.4 2.9-6.6 Sdr 0.9 ± 0.1 0.0-3.3 0.01 ± 0.0 0.00-0.01 Body composition % Fat thigh 44.4 ± 1.0 28.8-56.6 20.3 ± 1.7 12.5-27.1 % Lean thigh 53.6 ± 0.9 42.0-69.1 29.0 ± 1.9 19.9-37.0 % Fat subregion 36.8 ± 1.2 18.5-53.7 68.5 ± 1.8 61.2-77.1 3D ultrasound Dermal thickness 1.5 ± 0.0 1.2-2.1 1.7±0.1 1.2-2.0 Ultrasound surface area 352 ± 8 259-470 335 ± 20 283-460 Surface texture photography: red-band SD 14.8 ± 0.5 6.8-22.7 14.9 ± 1.6 9.7-24.0 Biomechanical properties Laxity 1.1 ± 0.0 0.6-1.8 1.0 ± 0.1 0.7-1.3 Laxity% 62.0 ± 0.9 41.8-72.3 60.9 ± 2.2 50.3-69.6 Elastic deformation 1.8 ± 0.0 1.2-2.6 1.7±0.l 1.4-2.0 Stiffness 143 ± 2 115-176 146 ± 4 125-158 Energy absorption 91.2 ± 1.4 67.3-121.4 87.9 ± 2.9 78.2-99.4 Elasticity (mm) 1.0 ± 0.0 0.6-1.5 1.0 ± 0.1 0.8-1.4 % Elasticity 55.3 ± 0.8 43.1-70.8 58.4 ± 2.0 51.7-68.2 of the dermal-subcutaneous border (ultrasound surface area, red-band SD from com­ pressed images). Cellulite severity was associated positively with tissue compliance (energy absorption) and negatively with stiffness (less stiff, greater severity). The parameters that predict cellulite severity were identified from multiple linear re­ gression modeling procedures. Expert image score and Svm were selected as the quan­ titative measures of cellulite severity (dependent variable). The expert image score could be predicted from subregion % fat (p 0.001) and Sdr (p = 0.01), giving the equation: Image Score= -4.2 + (0.22 · Subregion % Fat) + (0.91 · Sdr). The values of R, R2 , and adjusted R 2 were 0.82, 0.68, and 0.67, respectively. Cellulite severity (Svm) could be predicted from subregion % fat (p 0.001) and the dermal-subcutaneous surface area