Journals / Akdeniz Tıp Dergisi / 2026 / Cilt: 12 Sayı: 1

Preoperative Risk Stratification of Adnexal Masses Using Artificial Intelligence: Comparison with Histopathological Outcomes

Adneksiyal Kitlelerin Preoperatif Risk Sınıflandırılmasında Yapay Zekâ Kullanımı: Histopatolojik Sonuçlarla Karşılaştırma

Mustafa Gökkaya(gaziantep şehir hastanesi)0000-0002-0477-157X
Erkan Şimşek(gaziantep şehir hastanesi)0000-0002-6723-1773
İbrahim Taşkum(gaziantep şehir hastanesi)0000-0001-5260-2087
Aysun Alcı(necip fazıl şehir hastanesi)0000-0002-7912-7375
Necim Yalçın(batman eğitim ve araştırma hastanesi)0000-0001-5980-3244
Pages
—
Publication Date
2026-06-08
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

The AI model achieved an overall accuracy of 70.0% (95% CI: 60.4-78.1) and a balanced accuracy of 59.1%.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Objective: To evaluate the performance of an artificial intelligence (AI)-based classification system using preoperative clinical data in distinguishing benign, borderline, and malignant adnexal masses, and to compare its predictions with final histopathological diagnoses.
📊
Key Findings & Data:Material and Methods: The present retrospective study comprised 100 patients who underwent surgery for adnexal masses.
💡
Conclusion & Impact:The AI model achieved an overall accuracy of 70.0% (95% CI: 60.4-78.1) and a balanced accuracy of 59.1%.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:yapay zeka · adneksiyal kitleler · over neoplazmları

Abstract

Objective: To evaluate the performance of an artificial intelligence (AI)-based classification system using preoperative clinical data in distinguishing benign, borderline, and malignant adnexal masses, and to compare its predictions with final histopathological diagnoses. Material and Methods: The present retrospective study comprised 100 patients who underwent surgery for adnexal masses. Preoperative data (age, serum CA-125 levels, and MRI findings) were analyzed by an AI model, which classified cases as benign, borderline, or malignant. The AI predictions were then compared with the postoperative histopathological results. Diagnostic performance was assessed using accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and agreement statistics including Cohen’s kappa and quadratic weighted kappa. Results: Histopathological analysis identified 53 benign, 16 borderline, and 31 malignant cases. The AI model achieved an overall accuracy of 70.0% (95% CI: 60.4-78.1) and a balanced accuracy of 59.1%. The degree of agreement with histopathology was moderate (Cohen’s kappa: 0.497) and substantial with quadratic weighting (0.750). The sensitivity levels were 81.1% for benign, 18.8% for borderline, and 77.4% for malignant lesions. Specificity for malignancy was found to be high (92.8%). The binary classification accuracy was found to be 88.0% for malignant vs. non-malignant and 79.0% for benign vs. non-benign cases. Conclusion: AI-based evaluation using routine preoperative data demonstrated moderate diagnostic performance, with good discrimination of benign and malignant lesions but limited detection of borderline tumors. Further refinement and validation are needed prior to clinical implementation.

Özet

Amaç: Preoperatif klinik veriler kullanarak benign, borderline ve malign adneksiyal kitleleri ayırt etmede yapay zekâ (YZ) tabanlı bir sistemin performansını değerlendirmek ve bu tahminleri nihai histopatolojik tanılarla karşılaştırmaktır. Gereç ve Yöntemler: Bu retrospektif çalışmaya adneksiyal kitle nedeniyle cerrahi uygulanan 100 hasta dahil edildi. Yaş, serum CA-125 düzeyi ve manyetik rezonans görüntüleme (MRG) rapor bulgularından oluşan preoperatif veriler toplandı. Bu parametreler bir YZ modeline aktarıldı ve her olgu benign, borderline veya malign olarak sınıflandırıldı. YZ tabanlı tahminler postoperatif histopatolojik sonuçlarla karşılaştırıldı. Tanısal performans; doğruluk (accuracy), duyarlılık (sensitivity), özgüllük (specificity), pozitif öngörü değeri (PPV), negatif öngörü değeri (NPV) ve Cohen’s kappa ile ağırlıklı (quadratic weighted) kappa dahil olmak üzere uyum istatistikleri kullanılarak değerlendirildi. Bulgular: Histopatolojik değerlendirmeye göre 100 hastanın 53’ü benign, 16’sı borderline ve 31’i malign olarak sınıflandırıldı. YZ modelinin genel doğruluğu %70,0 (95% GA: %60,4–%78,1), dengelenmiş doğruluğu ise %59,1 olarak bulundu. YZ tahminleri ile histopatoloji arasındaki uyum orta düzeydeydi (Cohen’s kappa: 0,497), ağırlıklı kappa ise belirgin uyum gösterdi (0,750). Benign, borderline ve malign lezyonlarda duyarlılık sırasıyla %81,1, %18,8 ve %77,4 olarak saptandı. Model, malign lezyonlar için yüksek özgüllük (%92,8) gösterdi. İkili analizlerde, malign ve non-malign ayrımında doğruluk %88,0; non-benign (borderline + malign) ve benign ayrımında ise %79,0 olarak bulundu. Sonuç: Rutin preoperatif klinik veriler kullanılarak yapılan YZ tabanlı değerlendirme, adneksiyal kitle sınıflandırmasında orta düzeyde tanısal performans göstermektedir. Model, benign ve malign lezyonları ayırt etmede tatmin edici, ancak borderline tümörlerde sınırlı kalmaktadır. YZ, klinik karar sürecinde destekleyici bir araç olarak kullanılabilir; ancak yaygın kullanım öncesinde daha ileri geliştirme ve doğrulama çalışmaları gerekmektedir.

🔖

Cite This Article

Gökkaya, M., Şimşek, E., Taşkum, İ., Alcı, A., Yalçın, N. (2026). Preoperative Risk Stratification of Adnexal Masses Using Artificial Intelligence: Comparison with Histopathological Outcomes. Akdeniz Tıp Dergisi, Cilt: 12 Sayı: 1. https://doi.org/10.53394/akd.1927030

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (9)

  1. Timmerman D, Van Calster B, Testa A, Savelli L, Fischerova D, Froyman W, Wynants L, Van Holsbeke C, Epstein E, Franchi D, Kaijser J, Czekierdowski A, Guerriero S, Fruscio R, Leone FPG, Rossi A, Landolfo C, Vergote I, Bourne T, Valentin L. Predicting the risk of malignancy in adnexal masses based on the simple rules from the international ovarian tumor analysis group. Am J Obstet Gynecol 2016;214:424–37.
  2. Viora E, Piovano E, Baima Poma C, Cotrino I, Castiglione A, Cavallero C, Sciarrone A, Bastonero S, Iskra L, Zola P. The ADNEX model to triage adnexal masses: an external validation study and comparison with the IOTA two-step strategy and subjective assessment by an experienced ultrasound operator. Eur J Obstet Gynecol Reprod Biol 2020;247:207–11.
  3. Thomassin-Naggara I, Poncelet E, Jalaguier-Coudray A, Guerra A, Fournier LS, Stojanovic S, Millet I, Bharwani N, Juhan V, Cunha TM, Masselli G, Balleyguier C, Malhaire C, Perrot NF, Sadowski EA, Bazot M, Taourel P, Porcher R, Darai E, Reinhold C, Rockall AG. Ovarian-adnexal reporting data system magnetic resonance imaging (O-RADS MRI) score for risk stratification of sonographically indeterminate adnexal masses. JAMA Netw Open 2020;3:e1919896.
  4. Andreotti RF, Timmerman D, Strachowski LM, Froyman W, Benacerraf BR, Bennett GL, Bourne T, Brown DL, Coleman BG, Frates MC, Goldstein SR, Hamper UM, Horrow MM, Hernanz-Schulman M, Reinhold C, Rose SL, Whitcomb BP, Wolfman WL, Glanc P. O-RADS US risk stratification and management system: a consensus guideline from the ACR ovarian-adnexal reporting and data system committee. Radiology 2020;294:168–85.
  5. Cui L, Xu H, Zhang Y. Diagnostic accuracies of the ultrasound and magnetic resonance imaging ADNEX scoring systems for ovarian adnexal mass: systematic review and meta-analysis. Acad Radiol 2022;29:897–908.
  6. Mitchell S, Nikolopoulos M, El-Zarka A, Al-Karawi D, Al-Zaidi S, Ghai A, Gaughran JE, Sayasneh A. Artificial intelligence in ultrasound diagnoses of ovarian cancer: a systematic review and meta-analysis. Cancers (Basel) 2024;16:422.
  7. Qi L, Li X, Yang Y, Zhao M, Lin A, Ma L. Accuracy of machine learning in the preoperative identification of ovarian borderline tumors: a meta-analysis. Clin Radiol 2024;79:501–14.
  8. Grigore M, Popovici RM, Gafitanu D, Himiniuc L, Murarasu M, Micu R. Logistic models and artificial intelligence in the sonographic assessment of adnexal masses: a systematic review of the literature. Med Ultrason 2020;22:469–475.
  9. Dai WL, Wu YN, Ling YT, Zhao J, Zhang S, Gu ZW, Gong LP, Zhu MN, Dong S, Xu SC, Wu L, Sun LT, Kong DX. Development and validation of a deep learning pipeline to diagnose ovarian masses using ultrasound screening: a retrospective multicenter study. EClinicalMedicine 2024;78:102923.