Akademik Veri Yönetim Sistemi
European Physical Journal Plus, cilt.140, sa.12, 2025 (SCI-Expanded, Scopus)
In recent years, the use of artificial intelligence (AI) technologies in healthcare has increased significantly. In the face of increasing data complexity, deep neural networks (DNN) and quantum machine learning (QML) have become prominent solutions in the field of healthcare applications. In particular, quantum machine learning algorithms have been shown to facilitate the efficient analysis of large and complex data sets, enabling the discovery of hidden patterns and important diagnostic information. This study aims to improve the interpretability of medical image data. To this end, a hybrid classification model is proposed that combines DenseNet and quantum neural network (QNN) architectures with MRI images of the lumbar spine acquired between 2020 and 2023 in a private clinic. The model was trained and tested on a real data set containing four severity classes (Normal, Mild, Moderate, and Severe). The DenseNet and QNN models were integrated in the feature extraction and classification phases, using techniques such as early stopping to reduce the time required for training and prevent overlearning. In this study, the proposed DenseNet-based hybrid quantum neural network (DenseNet-QNN) showed the optimal performance. It achieved an accuracy rate of 83.94% in classifying lumbar spinal stenosis (LSS) into four classes. The model showed a significant improvement in accuracy, with an increase of 58% compared to the classical QNN architecture. It also showed high F1 values in the ‘Normal’ and ‘Severe’ classes. The developed approach combines the advantages of classical deep learning and quantum learning architectures and offers a statistically significant improvement over previous methods for LSS diagnosis. The objective of this study is twofold: firstly, to enhance the diagnostic process, and secondly, to reduce the workload of clinical experts in classifying lumbar spine MRI images. The proposed approach is a hybrid quantum-classical method.