Scholarlink

Polycystic ovary syndrome (PCOS) is a hormonal problem that affects the ovaries of women in their adulthood. It is a complex endocrinological and metabolic disease that causes abnormal menstrual periods, increased testosterone levels, and the development of polycystic ovaries. It often culminates into infertility, insulin resistance, obesity, diabetes (type-2), and cancer. Early identification of PCOS is necessary for immediate treatment, improving reproduction, minimizing chronic issues, and enhancing overall quality of life. This study uses deep learning (DL) models and explainable artificial intelligence (XAI) to produce a powerful, real-time diagnostic method for PCOS using ultrasound images. The research used three advanced DL models (DenseNet201, InceptionV3, and EfficientNetV2) to classify ovaries as infected or non-infected with PCOS, employing 3874 ultrasound pictures obtained from Kaggle. These models attained a classification accuracy of 100%, verified by precision, recall, and F1-score. Seven XAI techniques were utilized to compare and visually explain the decision-making process. This research advances the accuracy and efficacy of PCOS identification while encouraging faith in AI-driven medical services through transparency and a real-time web application.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Leveraging Explainable Artificial Intelligence for Real-Time Diagnosis of Polycystic Ovary Syndrome: A Transparent Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Kanij Fatema Arpa,
  • Md. Tahsin,
  • Maksura Binte Rabbani Nuha,
  • Ahmed Wasif Reza

摘要

Polycystic ovary syndrome (PCOS) is a hormonal problem that affects the ovaries of women in their adulthood. It is a complex endocrinological and metabolic disease that causes abnormal menstrual periods, increased testosterone levels, and the development of polycystic ovaries. It often culminates into infertility, insulin resistance, obesity, diabetes (type-2), and cancer. Early identification of PCOS is necessary for immediate treatment, improving reproduction, minimizing chronic issues, and enhancing overall quality of life. This study uses deep learning (DL) models and explainable artificial intelligence (XAI) to produce a powerful, real-time diagnostic method for PCOS using ultrasound images. The research used three advanced DL models (DenseNet201, InceptionV3, and EfficientNetV2) to classify ovaries as infected or non-infected with PCOS, employing 3874 ultrasound pictures obtained from Kaggle. These models attained a classification accuracy of 100%, verified by precision, recall, and F1-score. Seven XAI techniques were utilized to compare and visually explain the decision-making process. This research advances the accuracy and efficacy of PCOS identification while encouraging faith in AI-driven medical services through transparency and a real-time web application.