부산대학교 도서관

Spinal cord diseases pose a significant challenge in the medical field due to their complex nature and diverse manifestations. Early and accurate diagnosis is crucial for effective treatment and management. With the advent of deep learning techniques, the field of medical image analysis has witnessed remarkable advancements. Transfer learning, a subset of deep learning, has gained prominence for its ability to leverage pre-trained models on new tasks with limited labeled data, making it ideal for medical image classification tasks. In this study, we propose a novel approach for the identification of spinal cord diseases using transfer learning techniques, specifically employing a modified ResNet (Residual Neural Network) architecture. In order to train incredibly deep networks while avoiding the vanishing gradient issue, ResNet makes use of a deep structure and skip connections. This modification aims to extract subtle features that are crucial for accurate disease classification. A huge collection of spinal cord pictures, including both diseased and healthy instances, is used to train the suggested model. The model's capability to generalize and correctly diagnose spinal cord disorders is enhanced by the use of a pre-trained ResNet backbone, which draws on information gathered from a wide variety of datasets.