摘要目的:开发一款基于照片建模技术的病理性瘢痕三维形态量化评估软件，并验证其在临床应用中的准确性和可行性。方法:采用前瞻性观察性研究方法。解放军总医院第一医学中心2019年4月—2022年1月收治59例符合入选标准的病理性瘢痕患者（共107处瘢痕），其中男27例、女32例，年龄33（26，44）岁。以照片建模技术为基础，自主研发具有患者基本信息采集及瘢痕拍照、三维重建、模型浏览、报告生成等功能的病理性瘢痕三维形态测量软件。采用软件和临床常规方法（游标卡尺、彩色多普勒超声诊断仪及弹性体印模注水法测量）分别测量瘢痕的最长径、最大厚度及体积。统计建模成功瘢痕的数量、分布、患者数及使用软件和临床常规方法测量的瘢痕最长径、最大厚度、体积，统计建模失败瘢痕的数量、分布、类型及患者数。采用一元线性回归分析及Bland-Altman法分别分析软件和临床常规方法测量瘢痕最长径、最大厚度、体积的相关性和一致性，分别计算组内相关系数（ICC）、平均绝对误差、平均绝对百分误差。结果:对54例患者的102处瘢痕成功建模，瘢痕位于胸部43处、肩背部27处、四肢12处、面颈部9处、耳廓6处、腹部5处。使用软件和临床常规方法测量的瘢痕最长径、最大厚度、体积分别为3.61（2.13，5.19）、3.53（2.02，5.11）cm，0.45（0.28，0.70）、0.43（0.24，0.72）cm，1.17（0.43，3.57）、0.96（0.36，3.26）mL。5例患者5处增生性瘢痕或耳廓瘢痕疙瘩建模失败。使用软件和临床常规方法测量的瘢痕最长径、最大厚度、体积均呈显著线性相关（ r值分别为0.985、0.917、0.998， P<0.05）。使用软件和临床常规方法测量的瘢痕最长径、最大厚度、体积ICC分别为0.993、0.958、0.999（95%置信区间分别为0.989~0.995、0.938~0.971、0.998~0.999）。软件和临床常规方法测量的瘢痕最长径、最大厚度、体积一致性较好。Bland-Altman法分析显示，分别有3.92%（4/102）、7.84%（8/102）、8.82%（9/102）的瘢痕最长径、最大厚度、体积在95%一致性界限以外；在95%一致性界限以内，2.04%（2/98）的瘢痕最长径误差在0.5 cm以上，1.06%（1/94）的瘢痕最大厚度误差在0.2 cm以上，2.15%（2/93）的瘢痕体积误差在0.5 mL以上。软件与临床常规方法测量的瘢痕最长径、最大厚度、体积的平均绝对误差分别为0.21 cm、0.10 cm、0.24 mL，平均绝对百分误差分别为5.75%、21.21%、24.80%。 结论:基于照片建模技术的病理性瘢痕三维形态量化评估软件可以实现对大部分瘢痕的三维建模和形态参数测量。该软件测量结果与临床常规方法测量结果具有较好的一致性，误差在临床可接受范围内，能够作为病理性瘢痕临床诊断与治疗的辅助方法。

abstractsObjective:To develop a quantitative evaluation software for three-dimensional morphology of pathological scars based on photo modeling technology, and to verify its accuracy and feasibility in clinical application.Methods:The method of prospective observational study was adopted. From April 2019 to January 2022, 59 patients with pathological scars (totally 107 scars) who met the inclusion criteria were admitted to the First Medical Center of Chinese PLA General Hospital, including 27 males and 32 females, aged 33 (26, 44) years. Based on photo modeling technology, a software for measuring three-dimensional morphological parameters of pathological scars was developed with functions of collecting patients' basic information, and scar photography, three-dimensional reconstruction, browsing the models, and generating reports. This software and the clinical routine methods (vernier calipers, color Doppler ultrasonic diagnostic equipment, and elastomeric impression water injection method measurement) were used to measure the longest length, maximum thickness, and volume of scars, respectively. For scars with successful modelling, the number, distribution of scars, number of patients, and the longest length, maximum thickness, and volume of scars measured by both the software and clinical routine methods were collected. For scars with failed modelling, the number, distribution, type of scars, and the number of patients were collected. The correlation and consistency of the software and clinical routine methods in measuring the longest length, maximum thickness, and volume of scars were analyzed by unital linear regression analysis and the Bland-Altman method, respectively, and the intraclass correlation coefficients (ICCs), mean absolute error (MAE), and mean absolute percentage error (MAPE) were calculated.Results:A total of 102 scars from 54 patients were successfully modeled, which located in the chest (43 scars), in the shoulder and back (27 scars), in the limb (12 scars), in the face and neck (9 scars), in the auricle (6 scars), and in the abdomen (5 scars). The longest length, maximum thickness, and volume measured by the software and clinical routine methods were 3.61 (2.13, 5.19) and 3.53 (2.02, 5.11) cm, 0.45 (0.28, 0.70) and 0.43 (0.24, 0.72) cm, 1.17 (0.43, 3.57) and 0.96 (0.36, 3.26) mL. The 5 hypertrophic scars and auricular keloids from 5 patients were unsuccessfully modeled. The longest length, maximum thickness, and volume measured by the software and clinical routine methods showed obvious linear correlation (with r values of 0.985, 0.917, and 0.998, P<0.05). The ICCs of the longest length, maximum thickness, and volume of scars measured by the software and clinical routine methods were 0.993, 0.958, and 0.999 (with 95% confidence intervals of 0.989-0.995, 0.938-0.971, and 0.998-0.999, respectively). The longest length, maximum thickness, and volume of scars measured by the software and clinical routine methods had good consistency. The Bland-Altman method showed that 3.92% (4/102), 7.84% (8/102), and 8.82% (9/102) of the scars with the longest length, maximum thickness, and volume respectively were outside the 95% consistency limit. Within the 95% consistency limit, 2.04% (2/98) scars had the longest length error of more than 0.5 cm, 1.06% (1/94) scars had the maximum thickness error of more than 0.2 cm, and 2.15% (2/93) scars had the volume error of more than 0.5 mL. The MAE and MAPE of the longest length, maximum thickness, and volume of scars measured by the software and clinical routine methods were 0.21 cm, 0.10 cm, 0.24 mL, and 5.75%, 21.21%, 24.80%, respectively. Conclusions:The quantitative evaluation software for three-dimensional morphology of pathological scars based on photo modeling technology can realize the three-dimensional modeling and measurement of morphological parameters of most pathological scars. Its measurement results were in good consistency with those of clinical routine methods, and the errors were acceptable in clinic. This software can be used as an auxiliary method for clinical diagnosis and treatment of pathological scars.