Polycystic ovary syndrome (PCOS) exists as an extensive endocrine condition that generates substantial effects on the female body. Traditional diagnostic approaches often depend on clinical data and subjective interpretation of ultrasound images, which can benefit from advanced computational methods. This study introduces an innovative hybrid quantum-classical machine learning framework to significantly enhance PCOS classification from ultrasound imagery. We employ a quantum convolutional neural network (QCNN), utilizing a 16-qubit quantum circuit, to extract \( 7 \times 7 \times 16 \) feature maps from \( 4 \times 4 \) patches of resized to \( 28 \times 28 \) ultrasound images by leveraging quantum mechanical principles. These quantum-derived features subsequently serve as input for a suite of classical machine learning classifiers, which includes random forest (RF), support vector machine (SVM), gradient boosting (GB), extreme gradient boosting (XGBoost), light gradient boosting machine (LightGBM), category boosting (CatBoost), and a multilayer perceptron (MLP). The proposed hybrid methodology demonstrated exceptional performance on a publicly available dataset from kaggle, with the LightGBM and CatBoost classifiers achieving perfect 100% test accuracy, exceeding approximately 98% benchmarks. These results underscore the substantial potential of QCNNs as powerful feature extraction methods in medical imaging and highlight the synergistic capabilities of integrated quantum-classical systems for developing highly accurate, automated diagnostic technologies for PCOS. The code for this work is available at the following link: Colab Notebook.
