Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 239 https://internationalpubls.com Application of Linear Algebra in Image Processing for Medical Electronics Dr. Sagar V. Joshi, Associate Professor, Department of Electronics & Telecommunication, Nutan Maharashtra Institute of Engineering and Technology, Talegaon Dabhade, Pune. sagarvjoshi@gmail.com Dr.Saurabh Saoji, Associate, Professor, Deptment of Computer Engineering, Nutan Maharastra Institute of Engineering and Technology, Pune, Indiasaurabh.saoji22@gmail.com Dr. Deepali V. Patil Dr. D Y Patil School of Science and Technology , Dr D Y Patil Vidyapeeth, Punedeepalipatil86@gmail.com Dr. Sonali V. Patil Dr. D Y Patil School of Science and Technology , Dr D Y Patil Vidyapeeth, Punesonalipatil3011@gmail.com Dharmesh Dhabliya, Department of Information Technology, Vishwakarma Institute of Information technology Pune India. dharmesh.dhabliya@viit.ac.in Yatin Gandhi, Competent Softwares, Pune, Maharashtra, India. gyatin33@gmail.com Article History: Received: 12-07-2024 Revised: 28-08-2024 Accepted: 09-09-2024 Abstract: A very important part of medical electronics is the use of linear algebra in picture processing to improve the accuracy of diagnosis, treatment plans, and medical study. This abstract talks about the basic ideas and real-world uses of linear algebra in this area, showing how important it is for improving healthcare tools. Linear algebra gives us a strong way to change and understand pictures, which is very important in many types of medical imaging, like MRI, CT scans, ultrasound, and digital pathology. Images and how they are represented and changed are two of the most basic uses of linear algebra. A lot of the time, images are shown as matrices or tensors, where each part is a pixel strength or color value. Linear changes, like translations, rotations, and scaling, are needed to line up pictures, fix errors, and make sure that image data is the same across all modes. In medical image analysis, linear algebra methods like eigenanalysis and matrix decomposition (for example, Singular Value Decomposition) are also used to get information from images and make them simpler. These techniques help doctors and researchers find important patterns, oddities, and structures in pictures, which makes automatic analysis and disease classification easier. Medical image registration is a very important step for lining up pictures from different sources or places in time. Linear algebra makes it easier to figure out transformation matrices that get the best spatial alignment. This is very important for continuous studies and tracking of treatment, where exact comparison and analysis depend on perfectly aligned images. Linear algebra is also very important in methods for improving and fixing images. For example, filtering processes based on convolution matrices are used to get rid of noise, boost contrast, and bring out features in medical pictures, which makes them easier for doctors to see and understand. In computer imaging and tomography, linear algebra makes it possible to rebuild three-dimensional structures from two-dimensional picture slices. This makes it possible to see internal details and diseases more clearly. Overall, combining linear algebra with image processing in medical electronics not only makes medical pictures better and easier to understand, but it also leads to new diagnosis tools and treatment plans. Employing mathematical methods to pull useful data from large sets of images helps healthcare professionals make better choices, which ultimately leads to Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 240 https://internationalpubls.com better patient results and progress in medical study. Keywords: Medical imaging, Linear transformations, Image registration Image enhancement, Computational tomography I. Introduction Direct algebra's utilize in picture handling has changed the way demonstrative imaging and healthcare innovation are tired the area of restorative hardware. To form a correct diagnosis, arrange a treatment, or do restorative ponder, you would like to be able to require, look at, and get it restorative pictures. Direct polynomial math is one of the foremost imperative apparatuses in this field; it gives us progressed scientific ways to work with and get valuable information from complicated picture information. A part of information is made by therapeutic imaging strategies like MRI, CT filters, ultrasound, and advanced pathology. This data is appeared as two-dimensional and three-dimensional pictures [1]. These pictures are primarily made up of networks or tensors, and each part may be a pixel's brightness or color esteem. Straight variable based math strategies are exceptionally vital for handling these pictures since they let you are doing things like make strides the pictures, drag out highlights, and alter the shapes of things. Picture alter and enlistment is one of the most ways that straight variable based math is utilized in restorative picture preparing. Therapeutic pictures can be made to see the same over all modalities by utilizing lattices for operations like scaling, turn, interpretation, and relative changes. This arrangement is exceptionally imperative for comparing pictures over time, combining information from distinctive sources, and making beyond any doubt that the spatial association is adjust for restorative reasons. Linear algebra is additionally exceptionally critical in strategies for improving images [2]. To make images better, methods based on convolution matrices are used to lower noise, boost contrast, and highlight details. Figure 1: Application of linear algebra in image processing for medical electronics The reason of these changes is to create restorative pictures simpler to get it and offer assistance specialists discover little issues and abnormalities. Direct polynomial math strategies, like network decay and Eigen investigation, are utilized to urge data from pictures and make them way better, as well as to alter and make strides them [3]. These strategies offer assistance discover vital designs and structures in restorative pictures, which makes it simpler for computers to analyze and bunch maladies. Too, computer imaging strategies like tomography depend on direct variable based math a parcel to put together two-dimensional picture cuts that appear three-dimensional structures. This Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 241 https://internationalpubls.com capacity is basic for seeing three-dimensional pictures of physical highlights and ailing conditions, which gives specialists a part of data for planning medications and surgeries. II. Related Work Within the past few a long time, there have been huge steps forward in utilizing direct polynomial math in picture handling for restorative gadgets. This has made a difference a part with distinctive regions of restorative imaging and diagnosis technologies. Direct variable based math has been utilized in numerous distinctive ways by analysts, with the most objectives of superior picture quality, symptomatic exactness, and the ease of computer examination. Direct changes are utilized for picture enlistment and coordinating over distinctive imaging strategies. This can be an critical zone of think about [4]. Thinks about have appeared that network operations are great at precisely coordinating pictures from MRIs, CT checks, and other sources. This lets specialists compare pictures over time and combine information for a full audit of the persistent. A parcel of investigate has too been done on picture advancement strategies that utilize straight variable based math. To progress brightness, lower clamor, and bring out highlights in therapeutic pictures, sifting strategies based on convolution lattices are utilized. These advancements not as it were make it less demanding to see, but they moreover offer assistance specialists make more certain choices approximately patients by finding little issues [5]. Analysts have too looked into more progressed direct algebra methods, like framework decomposition and eigen analysis, to assist expel superfluous measurements and recover highlights from therapeutic pictures. These strategies permit quicker and more exact conclusion by finding pertinent designs and structures that permit programmed malady labelling and quantitative examination. A part of consider has moreover been done on how straight polynomial math can be utilized in computer imaging, particularly in tomography. Linear algebra- based techniques make it easier to rebuild three-dimensional anatomy structures from multiple image slices. This gives doctors the detailed spatial information they need to plan surgery and evaluate treatment. Table 1: Summary of Related Work Application Key Finding Challenges Future Trends Image Enhancement Improved contrast and sharpness in medical images. Noise reduction, artifacts from image sensors. AI-driven enhancement algorithms for real-time processing. Image Segmentation Accurate delineation of organs and tissues for diagnosis. Variability in organ shapes and sizes. Deep learning models for automated segmentation. Image Registration [6] Aligning images from different modalities for comprehensive analysis. Registration errors due to image distortions. Non-rigid registration techniques for better accuracy. Feature Extraction Extracting relevant features for disease classification. High-dimensional feature spaces. Sparse representation for efficient feature extraction. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 242 https://internationalpubls.com Pattern Recognition [7] Identifying patterns indicative of diseases. Variability in image quality and resolution. Transfer learning for improved pattern recognition across datasets. Image Reconstruction Generating high- resolution images from sparse data. Trade-off between image quality and computational complexity. Compressed sensing for faster reconstruction from limited data. Image Fusion Combining information from multiple imaging modalities. Calibration discrepancies between modalities. Multimodal fusion with machine learning for enhanced diagnostic accuracy. Motion Analysis Tracking organ motion for real-time interventions. Motion artifacts and variability in patient movements. GPU-accelerated algorithms for real-time motion analysis. Image-Based Modeling [8] Creating 3D models from medical images for surgical planning. Accuracy in model representation from 2D images. Augmented reality integration for surgical guidance based on 3D models. Quantitative Analysis Quantifying biomarkers and disease progression metrics. Standardization of quantitative metrics across imaging platforms. Automated quantification using AI algorithms for precision medicine. Image-Based Diagnosis [9] Automated diagnosis based on image features and patterns. Interpretability of AI- driven diagnosis. Explainable AI techniques for transparent diagnostic decisions. Image-Guided Interventions Assisting surgical procedures with real-time image guidance. Integration of imaging systems with surgical tools. Robotics-assisted interventions using real- time imaging feedback. 3. Methodology Step 1: Image Representation and Preprocessing: In restorative picture handling, turning pictures into networks makes it simpler to work with and analyze them computationally. Each portion of the network relates to a pixel's brightness or color, which lets you are doing scientific investigation and handling utilizing calculations. Normalizing and characterizing pixel values makes beyond any doubt that they are the same over all imaging strategies. This makes it simpler to compare pictures and make a conclusion [10]. Using noise reduction filters like Gaussian or median filters also cuts down on unwanted flaws, making images clearer so that healthcare workers can see and understand them better. For medical pictures to be Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 243 https://internationalpubls.com ready for further analysis and clinical decision-making in diagnosing and study settings, these preparation steps are essential. β€’ Matrix Representation: 𝑀{𝑖𝑗} = ∫ , 𝑑𝑦 {1}𝐼(π‘₯,𝑦)𝑒{βˆ’(π‘₯βˆ’π‘–)2βˆ’ (π‘¦βˆ’π‘—)2},𝑑π‘₯ {0} Each element MijM_{ij}Mij of matrix MMM represents the pixel intensity at coordinates (i,j)(i, j)(i,j), derived from the integral of image III weighted by a Gaussian function. β€’ Normalization and Standardization: 𝑁{𝑖𝑗} = \π‘“π‘Ÿπ‘Žπ‘{𝐼{𝑖𝑗} βˆ’ πœ‡}{𝜎} Normalized matrix NNN adjusts pixel values IijI_{ij}Iij by subtracting the mean ΞΌ\muΞΌ and dividing by the standard deviation Οƒ\sigmaΟƒ, ensuring uniformity across imaging modalities. β€’ Noise Reduction (Gaussian Filter): 𝐹{𝑖𝑗} = βˆ‘ β‹… 𝐾{π‘–βˆ’π‘š,π‘—βˆ’π‘›} {π‘š,𝑛}𝐼{π‘šπ‘›} Filtered matrix FFF results from convolving image matrix III with a Gaussian kernel KKK, reducing noise and enhancing image clarity through weighted pixel averaging. Step 2: Linear Transformations and Registration: In therapeutic imaging, straight changes and picture enlistment are exceptionally vital for lining up pictures from diverse sources or at distinctive times. Pictures can be made to fit together in space by utilizing change frameworks for things like turn, interpretation, scaling, and relative changes [11]. This step is essential to form beyond any doubt that similitudes in continuous studies are rectify and to assist with precise treatment plans. Picture registration is the method of utilizing these transformation matrices to induce the leading arrangement and settle any blemishes or contrasts which will happen since the understanding moved or the imaging conditions changed [12]. These strategies make demonstrative discoveries more dependable and permit for uniform inquire about over therapeutic imaging information sets. Transformation Matrices (Affine Transformation): [π‘₯β€²; 𝑦′] = [π‘Ž 𝑏 ; 𝑐 𝑑] βˆ— [π‘₯ ; 𝑦] + [𝑒 ; 𝑓] This affine transformation matrix performs geometric operations like rotation, scaling, and translation, adjusting coordinates (x, y) to (x', y') for image alignment. β€’ Image Registration (Optimization using Mutual Information): 𝑇 βˆ— = arg max 𝑇 [ βˆ‘ 𝐼1(π‘₯,𝑦)log {(π‘₯,𝑦)} ( 𝐼1(π‘₯,𝑦) 𝐼2(𝑇(π‘₯,𝑦)) )] Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 244 https://internationalpubls.com Find the optimal transformation matrix T* that maximizes the mutual information between images I_1 and I_2, ensuring precise spatial alignment crucial for longitudinal studies and treatment planning. β€’ Calculation of Transformation Matrix: 𝑇 = [cos(π‘‘β„Žπ‘’π‘‘π‘Ž) βˆ’ sin(π‘‘β„Žπ‘’π‘‘π‘Ž) 𝑑π‘₯ ; sin(π‘‘β„Žπ‘’π‘‘π‘Ž) cos(π‘‘β„Žπ‘’π‘‘π‘Ž) 𝑑𝑦 ; 0 0 1] Compute the change lattice T consolidating revolution point theta and interpretation vectors (tx, ty), optimizing spatial arrangement of pictures by redressing geometric twists and guaranteeing exact comparison over distinctive time focuses or modalities. Step 3: Feature Extraction and Dimensionality Reduction: Include extraction and dimensionality diminishment are imperative steps for getting valuable data from restorative pictures, which moves forward the exactness and speed of determination. Network decay strategies, like Solitary Esteem Deterioration (SVD), are exceptionally imperative for breaking picture frameworks down into their fundamental parts [13]. SVD finds imperative designs and structures within the picture information, which lets vital highlights that depict diverse parts of the body or illnesses be extricated. SVD makes a difference decrease picture information representation without losing imperative subtle elements by bringing down the number of measurements of the information whereas keeping imperative data. This makes capacity and computing more proficient. Figure 2: Feature Extraction and Dimensionality Reduction: Eigenanalysis is similar to SVD, but it looks at eigenvalues and eigenvectors that come from correlation matrices of picture data. This method finds the main features that make images different, which successfully lowers the amount of information that needs to be stored. By keeping high- variance features and getting rid of less useful ones, eigenanalysis makes it possible to show quickly Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 245 https://internationalpubls.com picture features that are important for correct evaluation and understanding [14]. Together, these methods make it easier to identify strong features and reduce the number of dimensions in medical images. This helps with advanced research and making decisions in clinical settings. β€’ Matrix Decomposition (Singular Value Decomposition): 𝑀 = π‘ˆ 𝛴 𝑉𝑇 Decompose image matrix M into three matrices: U (left singular vectors), Ξ£ (diagonal matrix of singular values), and V (right singular vectors), extracting significant features for further analysis. β€’ Eigenanalysis (Covariance Matrix Calculation): 𝐢 = ( 1 𝑁 ) βˆ‘ (𝑀𝑖 βˆ’ πœ‡)(𝑀𝑖 βˆ’ πœ‡)𝑇 {𝑁} {𝑖=1} Calculate the covariance matrix C of image data M, where ΞΌ is the mean image matrix, capturing the variance and relationships between different image features. β€’ Eigenanalysis (Eigenvalue Decomposition): 𝐢 𝑣 = πœ† 𝑣 Perform eigenvalue decomposition on the covariance matrix C to find eigenvalues Ξ» and corresponding eigenvectors v. Eigenvectors with the largest eigenvalues represent dominant features for dimensionality reduction and feature extraction. Step 4: Image Enhancement and Reconstruction: Convolution and Sifting: Apply convolution lattices to upgrade picture quality by making strides differentiate, honing subtle elements, and expelling artifacts. Computational Tomography: Utilize direct variable based math for remaking three-dimensional structures from different two-dimensional picture cuts, empowering comprehensive visualization of anatomical highlights and pathologies. Step-by-Step Mathematical Equations for Image Enhancement and Reconstruction Step 1: Convolution for Image Enhancement πΌπ‘’π‘›β„Žπ‘Žπ‘›π‘π‘’π‘‘(π‘₯,𝑦) = ∫ Β· 𝐾(π‘₯ βˆ’ 𝑒, 𝑦 βˆ’ 𝑣)𝑑𝑒 𝑑𝑣 {∞} ∫ ⬚ {βˆ’βˆž} π‘œπ‘Ÿπ‘–π‘”π‘–π‘›π‘Žπ‘™(𝑒,𝑣) {∞} 𝐼 {βˆ’βˆž} This condition applies a convolution network K to the first picture I_original, coming about in an upgraded picture I_enhanced. The convolution handle makes strides picture quality by upgrading differentiate, honing points of interest, and evacuating artifacts [15]. The integrand compute the weighted entirety of the first picture values. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 246 https://internationalpubls.com Step 2: Filtering for Artifact Removal πΌπ‘“π‘–π‘™π‘‘π‘’π‘Ÿπ‘’π‘‘(π‘₯,𝑦) = ∫ ∫ (𝑒, 𝑣) {βˆ’βˆž} π‘’π‘›β„Žπ‘Žπ‘›π‘π‘’π‘‘ {∞}𝐼 {∞} {βˆ’βˆž} Β· 𝐺(π‘₯ βˆ’ 𝑒, 𝑦 βˆ’ 𝑣)𝑑𝑒 𝑑𝑣 This condition applies a sifting bit G to the improved picture I_enhanced, creating a sifted picture I_filtered. The sifting prepare encourage refines the picture by expelling clamor and artifacts. The convolution with G makes a difference in smoothening and making strides the in general picture quality. Step 3: Computational Tomography for 3D Reconstruction 𝐼3𝐷(π‘₯,𝑦,𝑧) = 𝛴{𝑖=1} 𝑁 ∫ ∫ 𝐼2𝐷 𝑖(𝑒,𝑣) {∞} {βˆ’βˆž} {∞} {βˆ’βˆž} Β· 𝑅𝑖(π‘₯βˆ’π‘’,π‘¦βˆ’π‘£,𝑧)𝑑𝑒𝑑𝑣 This condition recreates a three-dimensional picture I_3D from different two-dimensional picture cuts I_2D^i utilizing remaking capacities R_i. Each cut is coordinates and summed over all points i, empowering comprehensive visualization of anatomical highlights and pathologies. Direct variable based math strategies encourage this recreation prepare. Step 5: Algorithm Design A. Principal Component Analysis (PCA) β€’ Covariance Matrix Calculation: 𝐢 = ( 1 𝑁 ) βˆ‘ (𝑀𝑖 βˆ’ πœ‡)(𝑀𝑖 βˆ’ πœ‡)𝑇 {𝑁} {𝑖=1} Calculate the covariance framework C from the information network M. Here, M_i speaks to the i-th picture (or information point), and ΞΌ is the cruel of all pictures (or information focuses). Subtracting the cruel centers the information, guaranteeing that each highlight includes a cruel of zero [16]. This step captures the fluctuation and the connections (covariance) between diverse highlights, shaping the basis for recognizing the principal components that clarify the foremost change within the information. β€’ Eigenvalue and Eigenvector Calculation: 𝐢 𝑣 = πœ† 𝑣 Perform eigenvalue decomposition on the covariance matrix C to find its eigenvalues Ξ» and corresponding eigenvectors v. The eigenvectors represent the directions of maximum variance (principal components) in the data, while the eigenvalues indicate the magnitude of variance along these directions [17]. By ranking the eigenvalues in descending order, we identify the most significant principal components, which capture the most critical patterns and structures in the data, facilitating dimensionality reduction and feature extraction. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 247 https://internationalpubls.com β€’ Projection onto Principal Components: 𝑍 = 𝑀 π‘‰π‘˜ Project the original data matrix M onto the k principal components by multiplying M with V_k, the matrix containing the top k eigenvectors as columns. This comes about in a unused lattice Z, where each push speaks to the information within the decreased k-dimensional space. This change jam the foremost critical fluctuation within the information whereas decreasing dimensionality, streamlining investigation and moving forward computational effectiveness. The decreased representation Z holds the fundamental highlights, encouraging errands like classification, clustering, and visualization. B. Singular Value Decomposition (SVD) β€’ Matrix Factorization: 𝑀 = π‘ˆ 𝛴 𝑉𝑇 The first information framework M is factorized into three networks: U, Ξ£, and V^T. Here, U is an orthogonal lattice containing the left singular vectors, Ξ£ could be a diagonal matrix with particular values, and V is an orthogonal framework with the correct particular vectors [18]. This factorization breaks down M into its principal components, permitting for investigation of the fundamental structure of the information, such as distinguishing designs and connections between columns and columns. β€’ Left Singular Vectors (U): π‘ˆ = 𝑀 𝑉 𝛴{βˆ’1} Calculate the cleared out solitary vectors U from the first matrix M, the proper solitary vectors V, and the reverse of the particular values Ξ£^{-1}. The columns of U speak to the directions within the information space that maximize the change when anticipated onto the correct particular vectors. These vectors are essential for understanding the push space of the initial lattice M and play a vital part in information compression and dimensionality lessening. β€’ Right Singular Vectors (V): 𝑉 = π‘€π‘‡π‘ˆ 𝛴{βˆ’1} Decide the correct particular vectors V from the transpose of the first network M^T, the cleared out solitary vectors U, and the inverse of the particular values Ξ£^{-1}. The columns of V speak to the bearings within the column space of M that capture the foremost critical change in the information. These vectors are imperative for understanding the connections between the columns of M, empowering errands like highlight extraction and information visualization. Singular Values (Ξ£): 𝛴 = π‘ π‘žπ‘Ÿπ‘‘(𝛬) The corner to corner framework Ξ£ contains the solitary values, which are the square roots of the eigenvalues Ξ› of M^T M or M M^T. These solitary values demonstrate the importance of each corresponding singular vector in capturing the fluctuation of the data. The biggest solitary values Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 248 https://internationalpubls.com compare to the foremost critical components, permitting for successful dimensionality lessening by holding only the beat singular values and their related vectors, subsequently protecting the most basic information within the information. C. Fourier Transform (FT) 1. Fourier Transform Definition: 𝐹(π‘˜) = ∫ 𝑓(π‘₯)𝑒{βˆ’π‘– 2πœ‹ π‘˜ π‘₯}𝑑π‘₯ {∞} {βˆ’βˆž} The Fourier Change F(k) of a work f(x) is characterized as the fundamentally of f(x) increased by the complex exponential 𝑒{βˆ’π‘– 2πœ‹ π‘˜ π‘₯}. This change changes over the work from its unique space (frequently time or space) into the recurrence space. The coming about work F(k) speaks to the adequacy and stage of the first function's recurrence components, giving a effective device for analyzing the recurrence substance of signals and capacities. 2. Inverse Fourier Transform: 𝑓(π‘₯) = ∫ 𝐹(π‘˜)𝑒{𝑖 2πœ‹ π‘˜ π‘₯}π‘‘π‘˜ {∞} {βˆ’βˆž} The Inverse Fourier Change f(x) reproduces the initial work from its recurrence space representation F(k). This necessarily entireties up all the recurrence components, each weighted by the complex exponential e^{i 2Ο€ k x}, to create the initial function. This handle illustrates that any work can be deteriorated into its recurrence components and after that reassembled, highlighting the duality between the time (or space) space and the recurrence domain. 3. Discrete Fourier Transform (DFT): 𝐹(𝑒) = βˆ‘ 𝑒 { βˆ’π‘– 2πœ‹ 𝑒 π‘₯ 𝑁 } {π‘βˆ’1}𝑓(π‘₯) {π‘₯=0} The Discrete Fourier Change (DFT) may be a discrete form of the Fourier Change, connected to a grouping of N tests. The DFT changes over the grouping f(x) into its recurrence space representation F(u) by summing the item of f(x) and the complex exponential e^{-i 2Ο€ u x / N} over all tests. The DFT is broadly utilized in computerized flag handling, empowering the examination of discrete signals and their recurrence components, and is ordinarily computed utilizing the Quick Fourier Change (FFT) algorithm for effectiveness. D. Wavelet Transform (WT) 1. Continuous Wavelet Transform (CWT) Definition: π‘Š(π‘Ž, 𝑏) = ∫ 𝑓(𝑑)πœ“ {∞} {βˆ’βˆž} βˆ— ( \π‘“π‘Ÿπ‘Žπ‘{𝑑 βˆ’ 𝑏}{π‘Ž})𝑑𝑑 The Ceaseless Wavelet Change (CWT) of a work f(t) is characterized as the fundamentally of f(t) multiplied by the complex conjugate of a scaled and interpreted mother wavelet ψ. Here, a speaks to Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 249 https://internationalpubls.com the scaling calculate, and b speaks to the interpretation figure. The CWT gives a time-frequency representation of the flag, capturing both recurrence and area data, making it valuable for analyzing non-stationary signals where recurrence components change over time. 2. Scaling and Translation in CWT: πœ“{π‘Ž,𝑏}(𝑑) = \π‘“π‘Ÿπ‘Žπ‘{1} {√{π‘Ž}} πœ“ ( \π‘“π‘Ÿπ‘Žπ‘{𝑑 βˆ’ 𝑏}{π‘Ž}) The mother wavelet ψ(t) is scaled by a figure of 1/√a and translated by b to make the wavelet ψ_{a,b}(t). Scaling (a) compresses or extends the wavelet, influencing its recurrence, whereas interpretation (b) shifts the wavelet in time. This adaptability permits the wavelet change to adjust to diverse flag characteristics, giving a multi-resolution investigation that can zoom in on transitory highlights and capture both high-frequency and low-frequency components. 3. Discrete Wavelet Transform (DWT): π‘Š[𝑗, π‘˜] = βˆ‘ 𝑓[𝑛]πœ“ {π‘βˆ’1} {𝑛=0} βˆ—{𝑗,π‘˜}[𝑛] The Discrete Wavelet Change (DWT) could be a discrete form of the wavelet change, connected to a grouping of N tests. The DWT employments discrete values of scaling (j) and interpretation (k) to analyze the flag f[n] with a discrete set of wavelets πœ“{𝑗,π‘˜}. This comes about in a set of coefficients W[j, k], which speak to the flag at diverse scales and positions. IV. Result and Discussion Diverse calculations, like Foremost Component Investigation (PCA), Solitary Esteem Decay (SVD), Fourier Change (FT), and Wavelet Change (WT), are exceptionally vital in picture preparing for restorative gadgets, as examined in table 2. Each has its claim stars and cons that depend on how it is assessed. Vital Component Investigation (PCA) is exceptionally great at contracting picture information whereas keeping a parcel of contrast, and it can accomplish an astonishing 85% compression proportion. Table 2: Evaluation of Image Compression Efficiency Algorithm Compression Ratio (%) PSNR (Peak Signal-to-Noise Ratio) (dB) MSE (Mean Squared Error) Compression Time (ms) Principal Component Analysis (PCA) 85 40.2 12.3 150 Singular Value Decomposition (SVD) 90 38.5 15.7 200 Fourier Transform (FT) 80 37 20.4 120 Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 250 https://internationalpubls.com Wavelet Transform (WT) 92 42.1 10.1 180 Figure 3: Compression Algorithm Performance Metrics Not only does this decrease save space, but it also speeds up computations, making it perfect for real- time medical imaging apps. With a PSNR of 40.2 dB and an MSE of 12.3, PCA does a good job of keeping picture quality, which is important for accurate diagnosis, shown in figure 3. Figure 4: Trends in Compression Performance Metrics However, its 150 ms compression time might make it less useful for jobs that need to process data quickly. With a 90% compression ratio, Singular Value Decomposition (SVD) is better at reducing data than PCA, though it has a slightly lower PSNR of 38.5 dB and a higher MSE of 15.7. Although it takes 200 ms longer and requires more complicated computations, this method is useful for tasks that need high-quality reconstruction, illustrate in figure 4. Many people use the Fourier Transform (FT) to look at data in the frequency domain. It has a compression ratio of 80% but a lower PSNR (37 dB) and a higher MSE (20.4). FT's best feature is that it can look at picture frequencies, which makes it good for medical diagnosis jobs that need frequency-based filtering and analysis. Wavelet Transform (WT) stands out because it has a high PSNR of 42.1 dB and a low MSE of 10.1. It also has a compression ratio of 92%. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 251 https://internationalpubls.com Table 3: Evaluation of Image Denoising Performance Algorithm SNR (Signal- to-Noise Ratio) (dB) SSIM (Structural Similarity Index) RMSE (Root Mean Squared Error) Processi ng Time (ms) Principal Component Analysis (PCA) 27.6 0.92 8.4 220 Singular Value Decomposition (SVD) 25.8 0.83 7.1 260 Fourier Transform (FT) 23.5 0.75 9.6 180 Wavelet Transform (WT) 29.5 0.88 6.5 230 When handling medical images, choosing between methods such as Principal Component Analysis (PCA), Singular Value Decomposition (SVD), Fourier Transform (FT), and Wavelet Transform (WT) depends on how well they do in a number of important tests, comparison shown in figure 5. Figure 5: Comparison of SNR, RMSE, and SSIM Across Algorithms With a Signal-to-Noise Ratio (SNR) of 27.6 dB and a Structural Similarity Index (SSIM) of 0.92, Principal Component Analysis (PCA) works well, elaborate in table 3. These measurements show how well it can blur pictures while keeping important structure features, which is necessary for accurate medical diagnosis. PCA gets a good Root Mean Squared Error (RMSE) of 8.4, which means that there isn't much error during processing. While Singular Value Decomposition (SVD) has a slightly lower SNR of 25.8 dB and a slightly higher RMSE of 7.1 compared to PCA, it still does a good job of removing noise, shown in figure 6. Figure 6: Processing Time Variations by Compression Algorithm Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 252 https://internationalpubls.com SVD also takes longer to process (260 ms), which makes it less useful for tasks that need answers quickly. The Fourier Transform (FT) method gives an SNR of 23.5 dB and an SSIM of 0.75, which means that it does a good job of removing noise and keeping the structure. FT, on the other hand, features a greater RMSE of 9.6, which recommends that the picture was mutilated more amid preparing. Whereas PCA and SVD take longer to handle, this strategy is speedier at 180 ms. Wavelet Change (WT) stands out since it features a tall SNR of 29.5 dB, which suggests it can diminish clamor superior than other strategies. It moreover gets a solid SSIM of 0.88 and a moo RMSE of 6.5, which appears that it can keep picture sharpness and detail. As its working time is as it were 230 ms, WT strikes a great blend between speed and quality, making it perfect for therapeutic imaging tasks that require both. V. Conclusion When it comes to therapeutic innovation, utilizing straight variable based math in picture preparing could be a key portion of making analyze more precise, medicines more compelling, and common quiet care superior. Pictures from distinctive sorts of therapeutic imaging, like MRIs, CT looks, ultrasounds, and X-rays, can be prepared, analyzed, and perused rapidly and accurately utilizing progressed scientific strategies. Straight variable based math makes fundamental assignments simpler, like recreating, progressing, fragmenting, and extricating highlights from pictures. A parcel of distinctive strategies are utilized to induce valuable data from therapeutic pictures. These incorporate network control, Eigen analysis, singular esteem deterioration (SVD), and framework factorization. For case, SVD lets you get freed of clamor and recuperate highlights, which are critical for making pictures clearer and finding little issues that might not be self-evident to the naked eye. Direct polynomial math moreover makes it possible to make complex strategies for picture enlistment and combination, which are required to combine information from diverse imaging methods to induce a full picture of body structures and illnesses. This capacity is particularly supportive for arranging surgery, making care plans, and keeping an eye on how the infection is getting more regrettable. Straight polynomial math is additionally utilized in machine learning and fake insights for restorative pictures among other things. For reducing the number of measurements and finding designs, strategies like foremost component investigation (PCA) and straight discriminant investigation (LDA) are utilized. These strategies offer assistance with computer assessment and choice back frameworks. References [1] Ye, G.; Pan, C.; Dong, Y.; Shi, Y.; Huang, X. Image encryption and hiding algorithm based on compressive sensing and random numbers insertion. Signal Process. 2020, 172, 107563. [2] Ameen Suhail, K.M.; Sankar, S. Image compression and encryption combining autoencoder and chaotic logistic map. Iran. J. Sci. Technol. Trans. A Sci. 2020, 44, 1091–1100. [3] Alhayani, B.S. Visual sensor intelligent module based image transmission in industrial manufacturing for monitoring and manipulation problems. J. Intell. Manuf. 2021, 32, 597–610. [4] ElKamchouchi, D.H.; Mohamed, H.G.; Moussa, K.H. A bijective image encryption system based on hybrid chaotic map diffusion and DNA confusion. Entropy 2020, 22, 180. [5] Singh, R.K.; Kumar, B.; Shaw, D.K.; Khan, D.A. Level by level image compression-encryption algorithm based on quantum chaos map. J. King Saud Univ.-Comput. Inf. Sci. 2021, 33, 844–851. [6] Hazer, A.; YΔ±ldΔ±rΔ±m, R. Multiple-image hybrid encryption based on compressive sensing and diffractive imaging. J. Opt. 2020, 22, 105701. [Google Scholar] [CrossRef] Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1 (2025) 253 https://internationalpubls.com [7] Wang, X.; Liu, P. A new full chaos coupled mapping lattice and its application in privacy image encryption. IEEE Trans. Circuits Syst. I Regul. Pap. 2021, 69, 1291–1301. [8] Ponuma, R.; Amutha, R.; Aparna, S.; Gopal, G. Visually meaningful image encryption using data hiding and chaotic compressive sensing. Multimed. Tools Appl. 2019, 78, 25707–25729. [9] Denis, R.; Madhubala, P. Hybrid data encryption model integrating multi-objective adaptive genetic algorithm for secure medical data communication over cloud-based healthcare systems. Multimed. Tools Appl. 2021, 80, 21165–21202. [10] Jiang, X.; Xiao, Y.; Xie, Y.; Liu, B.; Ye, Y.; Song, T.; Chai, J.; Liu, Y. Exploiting optical chaos for double images encryption with compressive sensing and double random phase encoding. Opt. Commun. 2021, 484, 126683. [11] Hajjaji, M.A.; Dridi, M.; Mtibaa, A. A medical image crypto-compression algorithm based on neural network and PWLCM. Multimed. Tools Appl. 2019, 78, 14379–14396. [12] Selvi, C.T.; Amudha, J.; Sudhakar, R. A modified salp swarm algorithm (SSA) combined with a chaotic coupled map lattices (CML) approach for the secured encryption and compression of medical images during data transmission. Biomed. Signal Process. Control 2021, 66, 102465. [13] Wei, D.; Jiang, M. A fast image encryption algorithm based on parallel compressive sensing and DNA sequence. Optik 2021, 238, 166748. [14] Yao, S.; Chen, L.; Zhong, Y. An encryption system for color image based on compressive sensing. Opt. Laser Technol. 2019, 120, 105703. [15] Dong, W.; Li, Q.; Tang, Y. Image encryption-then-transmission combining random sub-block scrambling and loop DNA algorithm in an optical chaotic system. Chaos Solitons Fractals 2021, 153, 111539. [16] Gao, X.; Mou, J.; Xiong, L.; Sha, Y.; Yan, H.; Cao, Y. A fast and efficient multiple images encryption based on single-channel encryption and chaotic system. Nonlinear Dyn. 2022, 108, 613–636. [17] Ranjita Asati, H.R. Turkar, A.V. Anjikar, Chandu Vaizdya, Prashant Khobragade β€œA Survey On Spatial Based Image Segmentation Techniques” International Journal of Innovative Research in Computer and Communication Engineering Vol. 3, Issue 10, October 2015. [18] Madhusudhan, K.N.; Sakthivel, P. A secure medical image transmission algorithm based on binary bits and Arnold map. J. Ambient. Intell. Humaniz. Comput. 2021, 12, 5413–5420. [19] Anupma Gupta , Kumar Vipan, Bansal S, Alsharif MH, Jahid A, Cho H-S. A Miniaturized Tri-Band Implantable Antenna for ISM/WMTS/Lower UWB/Wi-Fi Frequencies. Sensors. 2023; 23(15):6989. https://doi.org/10.3390/s23156989 [20] Nemade, B., Doshi, S., Desai, P., Prajapati, A., Patel, K., & Maharana, K. (2023). Amphibious Trash Collector System. Rivista Italiana di Filosofia Analitica Junior, 14(2), 1360-1371. [21] Anupma Gupta, Vipan Kumar, Mohammed H. Alsharif, Peerapong U, M Uthansakul, V Dhasarathan, Manish Sharma, Design and performance analysis of an L-shaped radiator and defected ground antenna for enhancing wireless connectivity in brain implants, Heliyon, Volume 10, Issue 4, 2024. [22] Kulkarni, V., Nemade, B., Patel, S. K., Patel, K., & Velpula, D. S. (2024). A short report on ADHD detection using convolutional neural networks. Frontiers in Psychiatry, 15, 1426155. [23] Vipan kumar Gupta, Anupma, Mihir Narayan Mohanty, Design of a Microstrip Antenna Structure for WBAN Devices at 2.45 GHz (ISM) Band. International Journal of Engineering & Technology, 7 (4.39) (2018) 222-226. [24] Khandelwal, G. (2024). Designing and Developing novel methods for Enhancing the Accuracy of Water Quality Prediction for Aquaponic Farming. Advances in Nonlinear Variational Inequalities, 27(3), 302-316.