Diagnosis of Osteoporosis by Bone X-Ray Based on Non-Destructive Compression

Abstract

X-ray is a commonly used imaging method for the diagnosis and evaluation of osteoporosis, which has crucial clinical diagnostic significance. To ensure the high fidelity requirements of medical image diagnosis and achieve high-quality transmission and storage of image data, this study uses deep learning to design a dual stream lossless compression network suitable for X-ray images. The results demonstrated that the designed network performed well in compression and bit rates on different datasets, with a minimum bit rate of 0.204 and a maximum compression rate of 0.946. Compared to other advanced models, this network had the highest peak signal-to-noise ratio and lower distortion of compressed images. In the compression process of X-ray images of osteoporosis, this network outperformed other models in different structural similarity indices, with values above 0.90, showing significant advantages. The equivalent number of views of the compressed image reached 0.93, and the visual quality of the lossless compressed image was high, ensuring the efficiency and accuracy of diagnosis. The research method can significantly improve the theoretical research level of lossless compression technology and enhance its practical value in remote medical diagnosis.

Keywords

• Lossless compression
• medical imaging
• neural network
• osteoporosis
• X-ray