Enhanced Medical Image Reconstruction Using Deep Learning Classification: A High-Resolution, Noise- Resilient Approach

Shivganga Patil; Lakshmi Patil

Author(s)

Shivganga Patil Department of Electronics and Communication Engineering , Sharnbasva UniversityKalaburagi, India

Lakshmi Patil Department of Electronics and Communication Engineering , Sharnbasva University Kalaburagi, India

Abstract

This study introduces a novel deep learning classification (DLC) method for reconstructing high-resolution medical images, showcasing its superiority over traditional techniques such as zero-filled (ZF) and compressed sensing (CS), as well as deep learning regression methods with L1 (DLR-L1) and L2 (DLR-L2) loss functions. Unlike conventional methods, DLC generates a probability distribution for each pixel, allowing for the approximation of continuous pixel values, which effectively reduces quantization errors and preserves fine image details. To address the computational challenges of high bit-depth imaging, a divide-and-conquer strategy was implemented, enabling the DLC network to handle 16-bit images without a significant increase in network parameters. The DLC method's performance was compared to ZF, CS, DLR-L1, and DLR-L2 across scenarios with high acceleration factors, low signal-to-noise ratios (SNR), and high bit-depths. The results demonstrated that DLC consistently produced images with superior resolution, sharper edges, and better preservation of low-contrast features. Quantitative metrics, including structural similarity index (SSI), peak signal-to-noise ratio (PSNR), relative error, and mean squared error (MSE), further validated the DLC method's superiority, particularly in noisy environments. These findings highlight the potential of DLC as a powerful tool in medical imaging, offering significant improvements in image quality and diagnostic accuracy across various applications.

Publication Details

Page(s) 822-829

DOI DOI not available

Published Journal: International Journal of Communication Networks and Information Security, Volume: 16, Issue: S1, Year: 2024

Request the full text from PASTIC

Keywords

probability distribution, Image quality improvement, High bit-depth imaging, High-resolution medical images, Deep learning classification

References

Mans P.,Phys J. .1977 ._eld, “Multi-planar image formation using NMR spin echoes,”. , 10(3) : 55.

Frahm J.,Haase A.,Matthaei D. .1986 .Rapid NMR imaging of dynamic processes using the FLASH technique. ” Magn. Reson. Med, 3(2) : 321-327.

Pruessmann K. P.,Weiger M.,Scheidegger M. B.,Boesiger P. .1999 .Sensitivity encoding for fast MRI,” Magn. , 42(5) : 952-962.

Sodickson D. K. .2000 .Tailored SMASH image reconstructions for robust in vivo parallel MR imaging. ''Magn. Reson. Med, 44(2) : 243-251.

Sodickson D. K.,Manning W. J. .1997 .Simultaneous acquisition of spatial harmonics (SMASH): Fast imaging with radiofrequency coil arrays. ” Magn. Reson. Med, 38 : 591-603.

Lustig M.,Donoho D. L.,Santos J. M.,Pauly J. M. .2008 .Compressed sensing MRI,” IEEE Signal Process. , 25(2) : 72-82.

Lustig M.,Donoho D.,Pauly J. M.,Sparse MRI .2007 .The application of compressed sensing for rapid MR imaging. ” Magn. Reson. Med, 58(6) : 1182-1195.

Quan T. M.,H T. M.,Computer-Assisted T. M.,Navab N.,Hornegger J.,Wells W.,A W. .2015 .and. MICCAI (Lecture Notes in Computer Science), 9351 : .

Zimmermann M.,Abbas Z.,Dzieciol K.,Shah N. J. .2018 .Accelerated parameter mapping of multiple-echo gradient-echo data using model-based iterative reconstruction. ''IEEE Trans. Med, 37(2) : 626-637.

Baron C. A.,Dwork N.,Pauly J. M.,Nishimura D. G.,Cartesian MRI .2018 .Rapid compressed sensing reconstruction of 3D non-. , 79(5) : 2685-2692.

LeCun Y. A.,Bengio Y.,Hinton G. E. .2015 .. ” Nature, 521(7553) : 436-444.

Litjens G.,Kooi T.,Bejnordi B. E.,Setio A. A. A.,Ciompi F.,Ghafoorian M.,W. M. van der Laak J. A.,B. van Ginneken J. A.,C. I. Sánchez J. A. .2017 .A survey on deep learning in medical image analysis. ” Med, 42 : 60-88.

Pawar K.,Chen Z.,Shah N. J. .2018 .Residual encoder and convolutional decoder neural network for glioma segmentation. ” Proc. Int. MICCAI Brainlesion Workshop, : 263-273.

Ronneberger O.,Fischer P.,Brox T. .2015 .Convolutional networks for biomedical image segmentation. in Proc. MICCAI, : 234-241.