Detecting COVID-19 early may help in devising an appropriate treatment plan and disease containment decisions. In this study, we demonstrate how transfer learning from deep learning models can be used to perform COVID-19 detection using images from three most commonly used medical imaging modes X-Ray, Ultrasound, and CT scan. The aim is to provide over-stressed medical professionals a second pair of eyes through intelligent deep learning image classification models. We identify a suitable Convolutional Neural Network (CNN) model through initial comparative study of several popular CNN models. We then optimize the selected VGG19 model for the image modalities to show how the models can be used for the highly scarce and challenging COVID-19 datasets. We highlight the challenges (including dataset size and quality) in utilizing current publicly available COVID-19 datasets for developing useful deep learning models and how it adversely impacts the trainability of complex models. We also propose an image pre-processing stage to create a trustworthy image dataset for developing and testing the deep learning models. The new approach is aimed to reduce unwanted noise from the images so that deep learning models can focus on detecting diseases with specific features from them. Our results indicate that Ultrasound images provide superior detection accuracy compared to X-Ray and CT scans. The experimental results highlight that with limited data, most of the deeper networks struggle to train well and provides less consistency over the three imaging modes we are using. The selected VGG19 model, which is then extensively tuned with appropriate parameters, performs in considerable levels of COVID-19 detection against pneumonia or normal for all three lung image modes with the precision of up to 86% for X-Ray, 100% for Ultrasound and 84% for CT scans.

/pdf/covid-19-detection-through-transfer-learning-using-3jkti8jq1q.pdf

COVID-19 Detection through Transfer Learning Using Multimodal Imaging Data

Whole-field velocity measurement techniques based on ultrasound imaging (a.k.a. ‘ultrasound imaging velocimetry’ or ‘echo-PIV’) have received significant attention from the fluid mechanics community in the last decade, in particular because of their ability to obtain velocity fields in flows that elude characterisation by conventional optical methods. In this review, an overview is given of the history, typical components and challenges of these techniques. The basic principles of ultrasound image formation are summarised, as well as various techniques to estimate flow velocities; the emphasis is on correlation-based techniques. Examples are given for a wide range of applications, including in vivo cardiovascular flow measurements, the characterisation of sediment transport and the characterisation of complex non-Newtonian fluids. To conclude, future opportunities are identified. These encompass not just optimisation of the accuracy and dynamic range, but also extension to other application areas.

/pdf/ultrasound-imaging-velocimetry-a-review-1ycpegs6no.pdf

Ultrasound imaging velocimetry : A review

Optical coherence tomography (OCT) images of the retina are a powerful tool for diagnosing and monitoring eye disease. However, they are plagued by speckle noise, which reduces image quality and reliability of assessment. This paper introduces a novel speckle reduction method inspired by the recent successes of deep learning in medical imaging. We present two versions of the network to reflect the needs and preferences of different end-users. Specifically, we train a convolution neural network to denoise cross-sections from OCT volumes of healthy eyes using either (1) mean-squared error, or (2) a generative adversarial network (GAN) with Wasserstein distance and perceptual similarity. We then interrogate the success of both methods with extensive quantitative and qualitative metrics on cross-sections from both healthy and glaucomatous eyes. The results show that the former approach provides state-of-the-art improvement in quantitative metrics such as PSNR and SSIM, and aids layer segmentation. However, the latter approach, which puts more weight on visual perception, outperformed for qualitative comparisons based on accuracy, clarity, and personal preference. Overall, our results demonstrate the effectiveness and efficiency of a deep learning approach to denoising OCT images, while maintaining subtle details in the images.

Retinal optical coherence tomography image enhancement via deep learning.

Purpose
The aim of this paper is to provide access to a database consisting of the raw radio-frequency ultrasonic echoes acquired from malignant and benign breast lesions. The database is freely available for study and signal analysis.

Acquisition and Validation Methods
The ultrasonic radio-frequency echoes were recorded from breast focal lesions of patients of the Institute of Oncology in Warsaw. The data were collected between 11/2013 and 10/2015. Patients were examined by a radiologist with 18 years’ experience in the ultrasonic examination of breast lesions. The set of data includes scans from 52 malignant and 48 benign breast lesions recorded in a group of 78 women. For each lesion, two individual orthogonal scans from the pathological region were acquired with the Ultrasonix SonixTouch Research ultrasound scanner using the L14-5/38 linear array transducer. All malignant lesions were histologically assessed by core needle biopsy. In the case of benign lesions, part of them was histologically assessed and another part was observed over a two-year period.

Data Format and Usage Notes
The radio-frequency echoes were stored in Matlab file format. For each scan, the region of interest was provided to correctly indicate the lesion area. Moreover, for each lesion, the BI-RADS category and the lesion class were included. Two code examples of data manipulation are presented. The data can be downloaded via the Zenodo repository (DOI: 10.5281/zenodo.545928) or the website h t t p ://bluebox.ippt. gov. p l/~hpiotrzk.

Potential Applications
The database can be used to test quantitative ultrasound techniques and ultrasound image processing algorithms, or to develop computer-aided diagnosis systems.

This article is protected by copyright. All rights reserved.

Open access database of raw ultrasonic signals acquired from malignant and benign breast lesions

Purpose
Statistical modeling of an ultrasound backscattered echo envelope is used for tissue characterization. However, in the presence of complex structures within the analyzed area, estimation of parameters is disturbed and unreliable, e.g., in the case of breast tumor classification. In order to improve the differentiation of breast lesions, the authors proposed a method based on the segmentation of homodyned K distribution parameter maps. Regions within lesions of different scattering properties were extracted and analyzed. In order to improve the classification, the best-performing features were selected from various regions and then combined.

Methods
A radio-frequency data set consisting of 103 breast lesions was used in the authors’ analysis. Maps of homodyned K distribution parameters were created using an algorithm based on signal-to-noise ratio, kurtosis, and skewness of fractional-order envelope moments. A Markov random field model was used to segment parametric maps. Features of different segments were extracted and evaluated based on bootstrapping and the receiver operating characteristic curve. To determine the best-performing feature subset, the authors applied the joint mutual information criterion.

Results
It was found that there were individual features which performed better than the ones commonly used for lesion characterization, like the parameter obtained through averaging of values over the whole lesion. The authors selected and discussed the best-performing features. Properties of different extracted regions were important and improved the distinction between benign and malignant tumors. The best performance was obtained by combining four features with the area under the receiver operating curve of 0.84.

Conclusions
The study showed that the analysis of internal changes in lesion parametric maps leads to a better classification of breast tumors. The authors recommend combining multiple features for characterization, instead of using only one parameter, especially in the case of heterogeneous lesions.

Classification of breast lesions using segmented quantitative ultrasound maps of homodyned K distribution parameters.

Ultrasound image enhancement: A review

This paper describes an approach to improve the contrast and signal to noise ratio on ultrasound images. Images with sub-pixel lateral displacements were re-sampled using a hexagonal grid, registered and compounded. The resultant image was filtered using a hexagonal adaptive masking filter. This approach was evaluated with simulated images and real images from a breast phantom. The results show total improvements in signal to noise ratio of up to 313% in simulated images, and 182% in phantom images. Contrast to noise ratio was improved by 286% in simulated images and 56% in phantom images.

Hexagonal adaptive filtering on compound ultrasound images

The resolution of ultrasound imaging is restricted primarily by the blurring process of the imaging system embedded in
the point spread function (PSF). Supercompounding has been found to be a highly effective way to enhance the
resolution of ultrasound imaging. Here we utilize a spatial ultrasound compounding technique using a B-mode array
rotated around a target in a range encompassing 180° or greater which we term "supercompounding." For some
ultrasound imaging modalities, the PSF is unknown and is space variant, caused by a mono-focus imaging device. To
use linear algorithms to enhance the resolution, images must be assumed to have a uniform PSF which is space invariant;
otherwise, it is necessary to use complicated non-linear algorithms. Under the above circumstances, an image with a
uniform PSF is the key element to more effective resolution enhancement. The supercompounding technique as here can
create an image with a uniform PSF from 214 B-scan images thus allowing the use of linear algorithm enhancement.
Once a supercompound image with a uniform PSF is constructed, the resolution of the image was further improved with
Weiner deconvolution. The processing technique was demonstrated on imaging a dissected porcine aortic root at 5
different critical height levels both with and without inflated pressure into to the sinus. The results can be analyzed to
acquire the mechanical properties and geometry of the aortic valve for future use. The resolution as measured by -6dB
width of the sinus wall shows a 14% improvement.

Supercompound imaging with Weiner deconvolution

Ultrasound is a widely used medical imaging methodology because it is safe and relatively inexpensive. However, the
quality of the images is affected by the point spread function of the system and coherent wave interference or speckle.
The present research studies the averaging of images that have been displaced laterally and displays them using an
interlaced grid. The main goals are to reduce speckle and improve contrast and resolution. The point spread function of
the ultrasound scanner was estimated using a thin nylon thread within a water bath. Then, a set of eight images of a
breast phantom (having lateral displacements smaller than the width of the point spread function) were averaged and
interlaced. The results show a total improvement of 4% in signal to noise ratio and 7% in contrast to noise ratio.

Super-resolution of ultrasound images by displacement, averaging, and interlacing

Spatial compounding has recently been introduced into modern commercial ultrasound systems. These systems are able to reduce angular dependent artifacts by averaging multiple images taken from different angles. In order to further reduce angular dependent artifacts, paired angle multiplicative compounding (PAMC) was recently introduced. The multiplication of pairs of images acquired from different angles reduces angular dependent artifacts while amplifying areas of signal. The quantification of improvement in PAMC images has proved difficult since images can become corrupt with high standard deviation background noise, limiting statistical evaluations from showing the true merit. In order to present a more theoretical understanding of PAMC, as well a way to gauge theoretical improvements, the point spread function (PSF) is used for analysis. This paper presents a simple geometrical model of the PAMC PSF based on the combination of two component B-scan PSFs. In order to validate the model, we compare it to a set of simulated PAMC PSFs created by varying the angle of paired multiplication. Both the theory and geometrical model are in agreement over the range [10°, 170°], allowing the optimal angle of paired multiplication to be determined to be 90°, yielding a PSF approximately 5× smaller than without pairing.

Tsuicheng Chiu

Papers

Ultrasound image enhancement: A review

Hexagonal adaptive filtering on compound ultrasound images

Supercompound imaging with Weiner deconvolution

Super-resolution of ultrasound images by displacement, averaging, and interlacing

Modeling the point spread function of paired angle multiplicative compounding in ultrasound