Multicenter evaluation of concordance between AI-assisted and hybrid AI-assisted cervical imaging systems and visual inspection with acetic acid in community-based cervical precancer screening

Abstract Background Visual Inspection With Acetic Acid (VIA) remains widely used for cervical precancer screening in low- and middle-income countries but is subject to operator-dependent variability. AI-assisted cervical imaging may improve consistency, although its performance across heterogeneous real-world settings remains insufficiently characterized. This study evaluated the concordance of AI-assisted and hybrid AI-assisted cervical imaging systems with routine VIA across multiple screening facilities. Methods This multicenter cross-sectional study included 253 women undergoing cervical screening at four facilities in West Java, Indonesia, between October 2025 and January 2026. VIA was used as the operational comparator rather than a histopathological reference standard. Cervical images obtained before and after acetic acid application were assessed using an automated AI-assisted system and a hybrid workflow involving expert review. Results were dichotomized as positive or negative. Concordance was assessed using Cohen’s kappa, with sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall agreement calculated relative to VIA. Results Twenty-three participants (9.1%) were VIA-positive. The AI-assisted system showed 56.5% sensitivity (95% CI, 34.5–76.8), 92.2% specificity (95% CI, 87.9–95.3), 41.9% PPV, and 95.5% NPV, with 88.9% overall agreement and moderate concordance (κ = 0.421; 95% CI, 0.266–0.622). The hybrid system showed 21.7% sensitivity (95% CI, 7.5–43.7), 95.7% specificity (95% CI, 92.1–97.9), 33.3% PPV, and 92.4% NPV, with 88.9% agreement and fair concordance (κ = 0.206; 95% CI, 0.014–0.398). Site-specific agreement varied but was limited by sparse VIA-positive cases. Conclusion Both AI-assisted workflows demonstrated high specificity and NPV but limited sensitivity relative to routine VIA. Findings highlight the variability of AI-assisted screening across real-world settings and the importance of standardized image acquisition. Histopathological validation is required to establish diagnostic accuracy and clinical utility.

Authors

Institutions

Publication Details

Journal
BMC Medical Imaging
Published
2026-11-14
DOI
https://doi.org/10.1186/s12880-026-03002-x
Primary Topic
Cervical Cancer and HPV Research
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Multicenter evaluation of concordance between AI-assisted and hybrid AI-assisted cervical imaging systems and visual inspection with acetic acid in community-based cervical precancer screening

Dwi Agustian, Dodi Suardi, Kemala Isnainiasih Mantilidewi, Qonita Adzkia et al.
BMC Medical Imaging
Cervical Cancer and HPV Research
article

Multicenter evaluation of concordance between AI-assisted and hybrid AI-assisted cervical imaging systems and visual inspection with acetic acid in community-based cervical precancer screening

Dwi Agustian, Dodi Suardi, Kemala Isnainiasih Mantilidewi, Qonita Adzkia, Muhammad Riza Vikar Akbar Alhayat, Rizky Ramadhan Sayyid Zenyda, Hilmi Mukhayadi, Muhammad Zahran, Rajendra Maheswara Wisnu Wardhana, Nadiyah An Nabiylah, Yudi Mulyana Hidayat
article en

Abstract

Abstract Background Visual Inspection With Acetic Acid (VIA) remains widely used for cervical precancer screening in low- and middle-income countries but is subject to operator-dependent variability. AI-assisted cervical imaging may improve consistency, although its performance across heterogeneous real-world settings remains insufficiently characterized. This study evaluated the concordance of AI-assisted and hybrid AI-assisted cervical imaging systems with routine VIA across multiple screening facilities. Methods This multicenter cross-sectional study included 253 women undergoing cervical screening at four facilities in West Java, Indonesia, between October 2025 and January 2026. VIA was used as the operational comparator rather than a histopathological reference standard. Cervical images obtained before and after acetic acid application were assessed using an automated AI-assisted system and a hybrid workflow involving expert review. Results were dichotomized as positive or negative. Concordance was assessed using Cohen’s kappa, with sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall agreement calculated relative to VIA. Results Twenty-three participants (9.1%) were VIA-positive. The AI-assisted system showed 56.5% sensitivity (95% CI, 34.5–76.8), 92.2% specificity (95% CI, 87.9–95.3), 41.9% PPV, and 95.5% NPV, with 88.9% overall agreement and moderate concordance (κ = 0.421; 95% CI, 0.266–0.622). The hybrid system showed 21.7% sensitivity (95% CI, 7.5–43.7), 95.7% specificity (95% CI, 92.1–97.9), 33.3% PPV, and 92.4% NPV, with 88.9% agreement and fair concordance (κ = 0.206; 95% CI, 0.014–0.398). Site-specific agreement varied but was limited by sparse VIA-positive cases. Conclusion Both AI-assisted workflows demonstrated high specificity and NPV but limited sensitivity relative to routine VIA. Findings highlight the variability of AI-assisted screening across real-world settings and the importance of standardized image acquisition. Histopathological validation is required to establish diagnostic accuracy and clinical utility.

BMC Medical Imaging
Padjadjaran University (ID)
Cervical Cancer and HPV Research
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon