Talathi Manju A, Dabhadkar Suchita, Doke Prakash P, Singh Varsha
Obstetrics and Gynecology, Bharati Vidyapeeth (Deemed to be University) Medical College, Pune, IND.
Community Medicine, Bharati Vidyapeeth (Deemed to be University) Medical College, Pune, IND.
Cureus. 2025 Mar 26;17(3):e81212. doi: 10.7759/cureus.81212. eCollection 2025 Mar.
Objectives The primary objective of this study was to compare the screening accuracy of AI assessment with colposcopy. Secondary objectives included comparing the triaging accuracy of AI and colposcopy assessments against histopathology. Methodology This prospective, single-arm screening test assessment study was conducted at the obstetrics and gynecology department of Bharati Vidyapeeth (Deemed to be University) Medical College in Pune, India. The study included sexually active, nonpregnant women aged 25-65 years visiting the OPD for per-speculum examination. Women with a clinically unhealthy cervix detected during the examination were counseled, and those who provided consent were enrolled. Patients with a history of prior cervical cancer treatment or hysterectomy were excluded. A total of 130 women were enrolled. Each participant underwent colposcopy, Smart Scope-AI (SS-AI) assisted visual inspection with acetic acid (VIA), and visual inspection with Lugol's iodine during the same visit. Positive findings from any test led to a biopsy, with samples sent for histopathological analysis. Results Of the 130 women enrolled, 30 were referred for biopsy. Histopathology results were obtained for 18 consenting women. Using colposcopy as the reference standard (N = 130), the accuracy of SS-AI was 76.53%. When compared to histopathology (N = 18) as the gold standard, the accuracy of colposcopy and SS-AI was 63.67% and 83.33%, respectively. The sensitivity and specificity of SS-AI were both 83.33%, while colposcopy had a sensitivity of 83.33% and a specificity of 50%. Likelihood ratios for SS-AI were superior to those of colposcopy. These findings suggest that the SS-AI-assisted test, a digital VIA test, accurately detects positive and negative cervical lesions. Conclusions The SS-AI system demonstrated comparable effectiveness to colposcopy and has the potential to be used as a point-of-care screening and triaging tool in primary healthcare centers lacking colposcopy equipment for triaging purposes.
目的 本研究的主要目的是比较人工智能(AI)评估与阴道镜检查的筛查准确性。次要目的包括比较AI和阴道镜检查评估相对于组织病理学的分流准确性。方法 这项前瞻性单臂筛查试验评估研究在印度浦那巴拉蒂维迪佩特(被视为大学)医学院妇产科进行。该研究纳入了年龄在25至65岁、有性生活、非妊娠且前来门诊进行窥器检查的女性。检查中发现临床宫颈不健康的女性接受了咨询,同意者被纳入研究。既往有宫颈癌治疗史或子宫切除术史的患者被排除。共纳入130名女性。每位参与者在同一次就诊时接受了阴道镜检查、智能Scope-AI(SS-AI)辅助醋酸染色肉眼观察(VIA)以及卢戈氏碘染色肉眼观察。任何一项检查的阳性结果均导致活检,样本送去进行组织病理学分析。结果 在纳入的130名女性中,30名被转诊进行活检。18名同意的女性获得了组织病理学结果。以阴道镜检查作为参考标准(N = 130),SS-AI的准确性为76.53%。与作为金标准的组织病理学(N = 18)相比,阴道镜检查和SS-AI的准确性分别为63.67%和83.33%。SS-AI的敏感性和特异性均为83.33%,而阴道镜检查的敏感性为83.33%,特异性为50%。SS-AI的似然比优于阴道镜检查。这些发现表明,SS-AI辅助检查(一种数字VIA检查)能准确检测出宫颈病变的阳性和阴性情况。结论 SS-AI系统显示出与阴道镜检查相当的有效性,并且有潜力在缺乏用于分流目的阴道镜设备的基层医疗中心用作即时筛查和分流工具。