Denoising of computed tomography using bilateral median based autoencoder network

Abstract

Denoising of Computed tomography (CT) images is a critical aspect of image processing that is expected to improve the performance of Computer‐aided diagnosis (CAD) systems. However, the use of complex imaging modalities such as CT imaging to ascertain pancreatic cancer is vulnerable to gaussian and poisson noises, making image denoising an imperative step for the accurate performance of CAD systems. This paper presents a Bilateral median based autoencoder network (BMAuto‐Net) constructed with intermediate batch normalization layers and dropout factors to eliminate gaussian noise from the CT images. The skip connections adjoining the network, prevent performance degradation that generally occurs in most autoencoder architectures. Based on the presented study, BMAuto‐Net is reckoned to outperform other traditional filters and autoencoders. The performance measurement of the proposed architecture is performed using the peak signal‐to‐noise ratio (PSNR), mean squared error (MSE), and structured similarity index (SSIM) metric values. The Cancer imaging archive (TCIA) dataset consisting of 19 000 CT images is used to validate the performance of the architecture with average PSNR values of 30.01, 30.53, and 30.52, MSE values of 98.23, 98.87, and 98.94, and SSIM of values of 0.67, 0.60, and 0.57 for noise factors (NFs) of 0.1, 0.3, and 0.5 respectively.

An efficient deep multi‐task learning structure for covid‐19 disease

“…In the Juneja et al. [4] study introduced BMAuto‐Net, a Bilateral Median‐based Autoencoder Network, for denoising CT images. Their proposed architecture incorporated intermediate batch normalization layers and dropout factors to reduce Gaussian noise from the images.…”

An Image Denoising Technique Using Wavelet-Anisotropic Gaussian Filter-Based Denoising Convolutional Neural Network for CT Images

“…The performance of the median filter for Gaussian noise removal could have been more effective due to its discrete nature, acceptable detail loss, and edge preservation. Authors in [23][24][25][26][27][41][42][43][44][45][46][47][48][49][50][51][52][53] discussed the challenges of image denoising based on the state-of-the-art medical image denoising techniques, such as bioinspired optimization-based filters and spatial filters using CNN, which included the preservation of image details, trade-off between noise removal and detail preservation, noise characteristics, computational complexity, and spatial and temporal coherence. However, these conventional image denoising techniques do not remove additive Gaussian noise from CT scan images because these spatial filters and denoising techniques [23][24][25][26][27][41][42][43][44][45][46][47][48][49][50][51][52][53]] may excel at reducing noise but need help to maintain the integrity of intricate anatomical information.…”

“…In [25,26], median filter and wavelet transform were applied to denoise CT scan images, and better results were achieved. However, there were challenges of blurring and detail preservation, high computational complexity, and edge smudging, which affect the accurate diagnosis and distortion of critical anatomical structures.…”

“…It is demonstrated that the significant challenges faced by filters in [17][18][19][20][21][22][23][24][25][26][27][28][29][30][31] were edge preservation, image restoration, computational intensity, and the blurring effect problem, which decreases image sharpness, obstructs the view of the underlying anatomy, and renders the CT scan images unsuitable. In recent years, several convolutional neural network (CNN)-based methods have been proposed for natural image denoising, and the application of a three-layer CNN for low-dose CT has shown promising results.…”