Article: SC-PCA-Enr: multi-modal residual adaption enabled Shearlet-Contourlet autoencoder for fused multi-modal image enhancement Journal: International Journal of Signal and Imaging Systems Engineering (IJSISE) 2024 Vol.13 No.4 pp.203 - 217 Abstract: In modern clinical diagnosis, multi-modal image enhancement is essential. Every magnetic resonance imaging (MRI) modality has unique anatomical characteristics that complement other modalities and provide rich diagnostic data. However, the existing methods directly combine the different modalities such as texture details and object contrast but ignore the significant details resulting in inaccurate diagnosis. Consequently, this research proposes the fusion-based Shearlet Counterlet enabled principal component analysis-auto encoder image enhancer (SC-PCA-EnR) model to produce high quality images. The Non-Subsampled Shearlet Contourlet Transform serves as the backbone of the enhancement process, offering a multi-resolution and multi-directional analysis that captures intricate image details. Ultimately, the principal component analysis (PCA) is exploited for reducing dimensionality, and fusion mechanism-based encoder is applied to enhance the performance. The experimental results are reported as Structural Similarity index measure of 0.97, peak signal noise ratio of 51.98 dB, and mean square error of 0.57 for normal brain dataset. Inderscience Publishers - linking academia, business and industry through research

Title: SC-PCA-Enr: multi-modal residual adaption enabled Shearlet-Contourlet autoencoder for fused multi-modal image enhancement

Authors: Nitin Sudhakarrao Thakare; Mukesh Yadav

Addresses: Electronic and Communication Engineering, SAGE University Indore, Indore Campus, Indore, Madhya Pradesh, 452020, India ' Electronic and Communication Engineering, SAGE University Indore, Indore Campus, Indore, Madhya Pradesh, 452020, India

Abstract: In modern clinical diagnosis, multi-modal image enhancement is essential. Every magnetic resonance imaging (MRI) modality has unique anatomical characteristics that complement other modalities and provide rich diagnostic data. However, the existing methods directly combine the different modalities such as texture details and object contrast but ignore the significant details resulting in inaccurate diagnosis. Consequently, this research proposes the fusion-based Shearlet Counterlet enabled principal component analysis-auto encoder image enhancer (SC-PCA-EnR) model to produce high quality images. The Non-Subsampled Shearlet Contourlet Transform serves as the backbone of the enhancement process, offering a multi-resolution and multi-directional analysis that captures intricate image details. Ultimately, the principal component analysis (PCA) is exploited for reducing dimensionality, and fusion mechanism-based encoder is applied to enhance the performance. The experimental results are reported as Structural Similarity index measure of 0.97, peak signal noise ratio of 51.98 dB, and mean square error of 0.57 for normal brain dataset.

Keywords: non-sub sampled Shearlet Contourlet transforms; multimodal fusion; image enhancement; PCA; principal component analysis; autoencoder-based image enhancer.

DOI: 10.1504/IJSISE.2024.146215

International Journal of Signal and Imaging Systems Engineering, 2024 Vol.13 No.4, pp.203 - 217

Received: 22 Apr 2024
Accepted: 15 Oct 2024
Published online: 12 May 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article

Title: SC-PCA-Enr: multi-modal residual adaption enabled Shearlet-Contourlet autoencoder for fused multi-modal image enhancement

Keep up-to-date