Automatic context-based health informatics system for diagnosing spinal deformity using deep learning

Abstract

Abnormal spinal alignment, which caused low back pain, also it’s one of the most common musculoskeletal ailments. As a result, it is a source of not just productivity loss but also personal anguish. By utilizing Convolutional Neural Network, the prediction of spinal deformity emerged to be a reliable model in medical imaging applications. One of the most difficult aspects of disease segmentation and categorization is detecting spinal deformities. However, the deep learning method aids in accurate and quick diagnosis. To classify normal as well as abnormal spinal X-ray images automatically, the deep learning approach is utilized. This research is utilized to detect X-ray image datasets, and a unique hybrid Convolution Neural Network-based Long Short-Term Memory classifier is applied. This proposed method extracts an automatic set of features for classifying spinal deformity abnormality. When compared to existing deep learning-based classifiers, a high computational spinal deformity prediction is proposed i.e., a hybrid Convolution Neural Network-based Long Short-Term Memory segmentation-based classifier. The analysis proved that the presented model is superior to existing spinal deformity prediction representations like accuracy 97.462%, recall 93.627%, precision 99.721%, specificity 93.289%, F1_score 97.182%, training time 102.50s, and Area Under ROC Curve 95.25% are concerned.