Journal: Sensors (Basel, Switzerland)

Article Title: Towards to Optimal Wavelet Denoising Scheme—A Novel Spatial and Volumetric Mapping of Wavelet-Based Biomedical Data Smoothing

doi: 10.3390/s20185301

Figure Lengend Snippet: Upper row is showing a CT scan of blood vessel calcification with the native level of noise ( A ) used as an input for algorithm testing. Images ( B – D ) are showing the original image with added artificial noise for 0.1, 0.2 and 0.5 level (in fact the noise consists of a sum of three noise types, in which the number means a mean value for additive white Gaussian noise and density for salt and pepper noise, speckle noise is kept on a constant default value used in MATLAB), respectively. Lower row shows the example of MRI scan of knee cartilage ( E used as an input), again with different consecutive noise levels of corresponding magnitude ( F – H ).

Article Snippet: Step 7 : Calculation of MSE—comparison of input and denoised signal Step 8 : Calculation of correlation level—comparison of input and denoised signal Step 9 : Calculation of Euclidean distance between the input and denoised signal Step 10 : Iterate to next wavelet type Step 11 : End for Step 12 : Iterate to next decomposition level Step 13 : End for Step 14 : Iterate to noise level Step 15 : End for Step 16 : Iterate to next wavelet family Step 17 : End for Step 18 : Storing of evaluation matrices for MSE, correlation level and Euclidean distance Algorithm 2 2D algorithm natural language description Step 1 :Scan folder for available image files and create multiarray structure of images Step 2 : For wavelet family Symlet, Dabeuchies and Coiflet test the image denoising Step 3 : For var values 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1; generate artificial noise as a sum of additive white Gaussian noise (imnoise(Input,’gaussian’,var)), speckle noise (imnoise(Input,’speckle’); and salt and pepper noise (imnoise(Input,’salt & pepper’,var); with AWGN mean value equal to var, salt and paper noise density equal to var and speckle noise equal to the MATLAB‘s constant default and add it to all loaded input images Step 4 : For decomposition level 1:1:5 Step 5 : For wavelet type 1:1:5 (e.g., Sym1, ..., Sym5) Step 6 : Denoise the image with use of wavelet (noise reduction of the signal based on thresholding of wavelet coefficients using a global threshold proposed by Donoho et al. [ ].)

Techniques: Computed Tomography

Journal: Sensors (Basel, Switzerland)

Article Title: Towards to Optimal Wavelet Denoising Scheme—A Novel Spatial and Volumetric Mapping of Wavelet-Based Biomedical Data Smoothing

doi: 10.3390/s20185301

Figure Lengend Snippet: Distribution of p-values of Chi-squared test normality for the Gaussian noise with dynamical noise intensity: μ = 0 , σ = 0.01–0.6.

Article Snippet: Step 7 : Calculation of MSE—comparison of input and denoised signal Step 8 : Calculation of correlation level—comparison of input and denoised signal Step 9 : Calculation of Euclidean distance between the input and denoised signal Step 10 : Iterate to next wavelet type Step 11 : End for Step 12 : Iterate to next decomposition level Step 13 : End for Step 14 : Iterate to noise level Step 15 : End for Step 16 : Iterate to next wavelet family Step 17 : End for Step 18 : Storing of evaluation matrices for MSE, correlation level and Euclidean distance Algorithm 2 2D algorithm natural language description Step 1 :Scan folder for available image files and create multiarray structure of images Step 2 : For wavelet family Symlet, Dabeuchies and Coiflet test the image denoising Step 3 : For var values 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1; generate artificial noise as a sum of additive white Gaussian noise (imnoise(Input,’gaussian’,var)), speckle noise (imnoise(Input,’speckle’); and salt and pepper noise (imnoise(Input,’salt & pepper’,var); with AWGN mean value equal to var, salt and paper noise density equal to var and speckle noise equal to the MATLAB‘s constant default and add it to all loaded input images Step 4 : For decomposition level 1:1:5 Step 5 : For wavelet type 1:1:5 (e.g., Sym1, ..., Sym5) Step 6 : Denoise the image with use of wavelet (noise reduction of the signal based on thresholding of wavelet coefficients using a global threshold proposed by Donoho et al. [ ].)

Techniques:

Journal: Sensors (Basel, Switzerland)

Article Title: Towards to Optimal Wavelet Denoising Scheme—A Novel Spatial and Volumetric Mapping of Wavelet-Based Biomedical Data Smoothing

doi: 10.3390/s20185301

Figure Lengend Snippet: Distribution of p-values of Mann-Whitney test for the Gaussian and Salt and Pepper noise.

Article Snippet: Step 7 : Calculation of MSE—comparison of input and denoised signal Step 8 : Calculation of correlation level—comparison of input and denoised signal Step 9 : Calculation of Euclidean distance between the input and denoised signal Step 10 : Iterate to next wavelet type Step 11 : End for Step 12 : Iterate to next decomposition level Step 13 : End for Step 14 : Iterate to noise level Step 15 : End for Step 16 : Iterate to next wavelet family Step 17 : End for Step 18 : Storing of evaluation matrices for MSE, correlation level and Euclidean distance Algorithm 2 2D algorithm natural language description Step 1 :Scan folder for available image files and create multiarray structure of images Step 2 : For wavelet family Symlet, Dabeuchies and Coiflet test the image denoising Step 3 : For var values 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1; generate artificial noise as a sum of additive white Gaussian noise (imnoise(Input,’gaussian’,var)), speckle noise (imnoise(Input,’speckle’); and salt and pepper noise (imnoise(Input,’salt & pepper’,var); with AWGN mean value equal to var, salt and paper noise density equal to var and speckle noise equal to the MATLAB‘s constant default and add it to all loaded input images Step 4 : For decomposition level 1:1:5 Step 5 : For wavelet type 1:1:5 (e.g., Sym1, ..., Sym5) Step 6 : Denoise the image with use of wavelet (noise reduction of the signal based on thresholding of wavelet coefficients using a global threshold proposed by Donoho et al. [ ].)

Techniques:

Journal: Sensors (Basel, Switzerland)

Article Title: Towards to Optimal Wavelet Denoising Scheme—A Novel Spatial and Volumetric Mapping of Wavelet-Based Biomedical Data Smoothing

doi: 10.3390/s20185301

Figure Lengend Snippet: A comparison of time consumption (calculated in seconds) for Gaussian noise for CT and MR images and individual Wavelet families.

Article Snippet: Step 7 : Calculation of MSE—comparison of input and denoised signal Step 8 : Calculation of correlation level—comparison of input and denoised signal Step 9 : Calculation of Euclidean distance between the input and denoised signal Step 10 : Iterate to next wavelet type Step 11 : End for Step 12 : Iterate to next decomposition level Step 13 : End for Step 14 : Iterate to noise level Step 15 : End for Step 16 : Iterate to next wavelet family Step 17 : End for Step 18 : Storing of evaluation matrices for MSE, correlation level and Euclidean distance Algorithm 2 2D algorithm natural language description Step 1 :Scan folder for available image files and create multiarray structure of images Step 2 : For wavelet family Symlet, Dabeuchies and Coiflet test the image denoising Step 3 : For var values 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1; generate artificial noise as a sum of additive white Gaussian noise (imnoise(Input,’gaussian’,var)), speckle noise (imnoise(Input,’speckle’); and salt and pepper noise (imnoise(Input,’salt & pepper’,var); with AWGN mean value equal to var, salt and paper noise density equal to var and speckle noise equal to the MATLAB‘s constant default and add it to all loaded input images Step 4 : For decomposition level 1:1:5 Step 5 : For wavelet type 1:1:5 (e.g., Sym1, ..., Sym5) Step 6 : Denoise the image with use of wavelet (noise reduction of the signal based on thresholding of wavelet coefficients using a global threshold proposed by Donoho et al. [ ].)

Techniques: Comparison