目的 基于输卵管系膜囊肿的核磁共振（MRI）、多模态超声等图像特征及临床生化指标构建机器学习模型,并验证其在输卵管系膜囊肿中的诊断价值。方法 回顾性分析本院2021年4月—2024年4月收治的178例疑似输卵管系膜囊肿患者的影像学资料,根据术后病理结果分为输卵管系膜囊肿组（120例）与非输卵管系膜囊肿组（58例）。采用LASSO回归和XGBoost算法筛选变量,通过多因素Logistic回归分析筛选独立影响因素,ROC曲线评估模型诊断效能。结果 LASSO回归筛选出病灶边缘、病灶形状、病灶大小、T1囊肿内低信号、T2囊肿内高信号、WBC、HDL-C共7个相关参数;维恩图对两种算法中的变量取交集后进行Logistic分析得到5个重要参数,包括病灶边缘、病灶大小、T1囊肿内低信号、T2囊肿内高信号、WBC;多因素分析证实各变量为独立影响因素（P<0.05）;ROC曲线显示模型AUC为0.91,敏感度为89.66%,特异度为75.83%;预测模型检出113例患者,误诊12例,漏诊7例,准确度高于单一诊断。结论 基于磁共振、多模态超声等图像特征构建的机器学习预测模型,具有较高诊断效能。
Abstract
Objective To investigate the diagnostic value of a machine learning model based on magnetic resonance imaging (MRI), multimodal ultrasound, and clinical biochemical indicators for mesosalpinx cysts. Methods A retrospective analysis was conducted on MRI and ultrasound image features of 178 patients with suspected mesosalpinx cysts admitted to Gansu Provincial Maternity and Child-Caring Hospital between April 2021 and April 2024. Based on surgical pathology, patients were classified into a mesosalpinx cyst group (n=120) and a non-mesosalpinx cyst group (n=58). Variables were selected using LASSO regression and the XGBoost algorithm. Independent predictors were identified by multivariate logistic regression, and the model's diagnostic performance was assessed with receiver operating characteristic (ROC) curves. Results LASSO regression selected seven parameters most relevant to mesosalpinx cyst diagnosis: lesion margin, lesion shape, lesion size, T1-weighted intracystic hypointensity, T2-weighted intracystic hyperintensity, white blood cell (WBC) count, and high-density lipoprotein cholesterol. After intersecting variables from both algorithms via a Venn diagram, five key parameters were retained for logistic analysis: lesion margin, lesion size, T1-weighted intracystic hypointensity, T2-weighted intracystic hyperintensity, and WBC count. Multivariate logistic regression confirmed all variables as independent predictors (P<0.05). ROC curve analysis demonstrated the model's diagnostic performance with an area under the curve of 0.91, sensitivity of 89.66%, and specificity of 75.83%. The predictive model correctly identified 113 patients, with 12 misdiagnoses and 7 missed diagnoses, achieving higher accuracy compared to single diagnostic methods. Conclusion The machine learning model incorporating MRI and multimodal ultrasound image features shows high diagnostic efficacy for mesosalpinx cysts.
关键词
核磁共振 /
多模态超声 /
输卵管系膜囊肿 /
LASSO /
XGBoost
Key words
Magnetic resonance /
Multimodal ultrasound /
mesosalpinx cyst /
LASSO /
XGBoost
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] CHEN J, LI C, ZHANG H, et al.Tubal mesosalpinx cysts combined with adnexal torsion in adolescents: a report of two cases and review of the literature[J]. BMC Pediatr, 2024, 24(1): 525.
[2] BHANSAKARYA R, SUBEDI S.Laparoscopic Management of Large Right Paratubal Cyst: A Case Report[J]. JNMA J Nepal Med Assoc, 2020, 58(227): 501-504.
[3] ALPENDRE F, PEDROSA I, SILVA R, et al.Giant paratubal cyst presenting as adnexal torsion: A case report[J]. Case Rep Womens Health, 2020, 27: e222.
[4] 顾颖超. PET-CT和PET-MRI在常见妇科恶性肿瘤诊治中的应用价值[J]. 中国实用妇科与产科杂志, 2019, 35(7): 839-843.
[5] YANG M, CHEN Y, ZHOU X, et al.Machine learning models for prediction of NPVR>/=80% with HIFU ablation for uterine fibroids[J]. Int J Hyperthermia, 2025, 42(1): 2473754.
[6] 陈丽妃, 杨柳, 陈西燕, 等. 基于超声影像组学和深度学习特征的机器学习模型诊断囊性交界性卵巢肿瘤的临床价值[J]. 临床超声医学杂志, 2025, 27(07): 560-565.
[7] TIBSHIRANI R.Regression shrinkage and selection via the lasso[J]. Journal of the Royal Statistical Society, 1996, 58(1): 267-288.
[8] ZHANG Y, ZHU Q, WU P, et al.Thirty-eight cases of paraovarian cysts in children and adolescents: a retrospective study[J]. Pediatr Surg Int, 2024, 40(1): 62.
[9] MARRA D D C, RODRIGUES B S, MIURA G, et al. Paraovarian cyst with associated ovarian torsion[J]. Rev Fac Cien Med Univ Nac Cordoba, 2023, 80(4): 559-567.
[10] 祝承宇, 项燕妮. 超声下女性输卵管系膜囊肿的影像学特征及临床表现分析[J]. 中国妇幼保健, 2025, 40(19): 3683-3686.
[11] QIAN L, WANG X, LI D, et al.Isolated fallopian tube torsion with paraovarian cysts: a case report and literature review[J]. BMC Womens Health, 2021, 21(1): 345.
[12] ADJEI N N, YUNG N, TOWERS G, et al.Establishing an Association between Polycystic Ovarian Syndrome and Pilonidal Disease in Adolescent Females[J]. J Pediatr Adolesc Gynecol, 2023, 36(1): 39-44.
[13] KEYHANIAN K, MACK T, FORGO E, et al.Female Adnexal Tumor of Probable Wolffian Origin (Wolffian Tumor): A Potential Mimic of Peritoneal Mesothelioma[J]. Am J Surg Pathol, 2024, 48(8): 1041-1051.
[14] STEFANOPOL I A, BAROIU L, NEAGU A I, et al.Clinical, Imaging, Histological and Surgical Aspects Regarding Giant Paraovarian Cysts: A Systematic Review[J]. Ther Clin Risk Manag, 2022, 18: 513-522.
[15] 周玉敏, 查正霞, 方金枝. 子宫内膜异位囊肿患者输卵管系膜状态与卵巢囊肿剥除术后血清CA125、性激素水平的关系[J]. 中国性科学, 2024, 33(2): 122-125.
[16] 张斯淼, 赵倩, 海盼盼, 等. 经阴道 (直肠) 三维超声在输卵管系膜囊肿中的诊断价值[J]. 肿瘤基础与临床, 2017, 30(1): 39-41.
[17] DAWOOD M T, NAIK M, BHARWANI N, et al.Adnexal Torsion: Review of Radiologic Appearances[J]. Radiographics, 2021, 41(2): 609-624.
[18] STEFANOPOL I A, BAROIU L, NEAGU A, et al.Clinical, Imaging, Histological and Surgical Aspects Regarding Giant Paraovarian Cysts: A Systematic Review[J]. Ther Clin Risk Manag, 2022, 18(1): 513-522.
[19] 马凤华, 强金伟. 卵巢肿瘤的MRI表现[J]. 中华放射学杂志, 2024, 58(2): 238-242.
[20] 蔡晓燕. 彩色多普勒超声诊断输卵管系膜囊肿的漏误诊分析[J]. 实用医技杂志, 2013, 20(10): 1138-1139.
基金
甘肃省卫生健康行业科技创新重大科研项目“基于超声影像组学的CHD-DPBF胎儿肺成熟度预测模型建立及临床应用研究”（GSWSQNPY2024-05）
PDF(2042 KB)
