Search

Super-resolution is an ill-posed inverse problem. It needs a self-similarity regularization term in its formulation to solve the issue. Researchers have studied self-similarity in natural images thoroughly and extended the results to medical data. This paper investigates the self-similarity property using a general affine model and the SSIM, PSNR, matrix rank, and mutual information measures. We study variations of self-similarity with image modality, inter/intra-resolution, tissue types, and acquisition phase. Based on our experiments, super-resolution algorithms that use self-similarity property achieve better results in CT images than MR data. Some organs, including the liver, have a stronger self-similarity compared to the brain and lung. The extension of our study prepares a comprehensive guide map to employ self-similarity properties in medical image processing algorithms. [ABSTRACT FROM AUTHOR]

Enhancing the quality of diagnostic images and preserving their original features is crucial for early detection and further analysis. In non-contrast CT imaging, a noisy and low contrast CT image can give inadequate information for the visual analysis of affected regions. A new method for enhancing non-contrast CT images with dual-tree complex wavelet transform (DT-CWT) and adaptable morphology is presented in this paper. Input CT images are inserted into the DT-CWT system, resulting in low- and high-frequency subbands. On high-frequency subbands, denoising is performed using the wavelet-related shearlet transform method, which results in enhanced high-frequency sub-images. An adaptive morphology top-hat transform technique is used with DCT-based local enhancement to enhance the low-frequency sub-images. The improved low and high-frequency components are then recombined to form the enhanced CT image using inverse DT-CWT. In order to estimate the success of the proposed system, experiments and validations are carried out on a diverse collection of CT images taken from publicly accessible databases. An extensive quantitative analysis demonstrates that the proposed method outperforms existing image enhancement techniques in terms of peak signal-to-noise ratio, entropy, contrast ratio, and measure of enhancement. In the proposed algorithm, the contrast is enhanced while maintaining the brightness and natural characteristics of the CT image. The proposed approach produces CT images of higher quality, which can be useful for detecting and diagnosing illnesses. [ABSTRACT FROM AUTHOR]

In the past decades, CT images have been widely used and played a critical role in medical diagnosis. However, low-dose CT images are often contaminated by noise, this being the most important factor affecting the quality of a CT image. This paper proposes a novel integrated framework and a denoising method for low-dose medical CT images to obtain a better denoising effect whilst at the same time preserving an image's local structure information. First, an image moving decomposition is employed to decompose the CT image. The original CT noisy image is decomposed, and the components will be processed separately, so that the details and edges of the CT image can be better preserved. Next, the Shearlet Transformation-based denoising method is applied to the component which contains edges and detailed information of the CT image. The multi-directionality and multi-scale property of the Shearlet make it possible to obtain better effect in denoising the detail parts. BM3D filtering is used to remove noise in the component similar to the origin image, and obtain ideal denoising results in denoising the approximate components (mainly low frequency part) of the CT image. With the two processed components and the inverse decomposition, the denoised image is obtained. Finally, simulations and clinical experiments are conducted and comparisons made. The experimental results show the proposed denoising method can obtain better performances in terms of PSNR value, SSIM and FoM and thus have very competitive results compared with other existing CT denoising methods. [ABSTRACT FROM AUTHOR]

The coronavirus disease (COVID-19) is an infectious disease caused by the SARS-CoV-2 virus. COVID-19 is found to be the most infectious disease in last few decades. This disease has infected millions of people worldwide. The inadequate availability and the limited sensitivity of the testing kits have motivated the clinicians and the scientist to use Computer Tomography (CT) scans to screen COVID-19. Recent advances in technology and the availability of deep learning approaches have proved to be very promising in detecting COVID-19 with increased accuracy. However, deep learning approaches require a huge labeled training dataset, and the current availability of benchmark COVID-19 data is still small. For the limited training data scenario, the CNN usually overfits after several iterations. Hence, in this work, we have investigated different pre-trained network architectures with transfer learning for COVID-19 detection that can work even on a small medical imaging dataset. Various variants of the pre-trained ResNet model, namely ResNet18, ResNet50, and ResNet101, are investigated in the current paper for the detection of COVID-19. The experimental results reveal that transfer learned ResNet50 model outperformed other models by achieving a recall of 98.80% and an F1-score of 98.41%. To further improvise the results, the activations from different layers of best performing model are also explored for the detection using the support vector machine, logistic regression and K-nearest neighbor classifiers. Moreover, a classifier fusion strategy is also proposed that fuses the predictions from the different classifiers via majority voting. Experimental results reveal that via using learned image features and classification fusion strategy, the recall, and F1-score have improvised to 99.20% and 99.40%. [ABSTRACT FROM AUTHOR]

In many applications in computed tomography, practical limitations in data acquisition restrict the number of projections (views). The use of the standard convolution backprojection algorithm for reconstruction from an inadequate number of projections results in view aliasing artifacts. One approach to alleviating the effects of such artifacts consists of artificially increasing the number of views, by estimating a set of intermediate views. Two possible methods of estimating the intermediate views are interpolation and reprojection. In this paper, a study of the two is considered. Based on the merits and demerits of the two methods, a combination of the two methods is investigated. Specifically, a reconstruction from the available sinogram augmented by intermediate view reprojections, and the projections interpolated from the original views and the reprojections, provide an additional improvement with respect to view aliasing artifacts. The advantage of computing reprojections over smaller regions of interest is discussed. When the number of available projections is reasonably high but not adequate to produce an artifact-free reconstruction, estimating the intermediate views by interpolation provides an improvement without much additional degradation, at minimal computational cost. [ABSTRACT FROM AUTHOR]

In this article, an FPGA-based hardware prototype design has been proposed and implemented in real time for the security of the Electronic Patient Medical Record. This proposed hardware can be integrated into various medical image scanning instruments like X-ray, ultrasonography, magnetic resonance imaging, and computed tomography machines, which can embed the health record of a patient into his/her scanned medical images in real time. The FPGA prototype was implemented using Xilinx Spartan-3E FPGA. The hardware resource utilization results have been observed to be low and practically viable. High operating speed up to 245 MHz (for joint embedding-cum-encryption process) and up to 252 MHz (for joint decoding-cum-decryption process) is achieved, with a respective low power consumption of 95 mW and 82 mW. Extracting the medical data is possible using a key, generated within the proposed algorithm using the patient's fingerprint. The proposed hardware architecture is the first architecture for the algorithm proposed in this article to the best of our knowledge. [ABSTRACT FROM AUTHOR]