Average Peak Signal-to-noise Ratio Research Articles

Diffraction model-driven neural network trained using hybrid domain loss for real-time and high-quality computer-generated holography.

Learning-based computer-generated holography (CGH) has demonstrated great potential in enabling real-time, high-quality holographic displays. However, most existing learning-based algorithms still struggle to produce high-quality holograms, due to the difficulty of convolutional neural networks (CNNs) in learning cross-domain tasks. Here, we present a diffraction model-driven neural network (Res-Holo) using hybrid domain loss for phase-only hologram (POH) generation. Res-Holo utilizes the weights of the pretrained ResNet34 as the initialization during the encoder stage of the initial phase prediction network to extract more generic features and also to help prevent overfitting. Also, frequency domain loss is added to further constrain the information that the spatial domain loss is insensitive. The peak signal-to-noise ratio (PSNR) of the reconstructed image is improved by 6.05 dB using hybrid domain loss compared to using spatial domain loss alone. Simulation results show that the proposed Res-Holo can generate high-fidelity 2 K resolution POHs with an average PSNR of 32.88 dB at 0.014 seconds/frame on the DIV2K validation set. Both monochrome and full-color optical experiments show that the proposed method can effectively improve the quality of reproduced images and suppress image artifacts.

Adaptive fusion of multi-exposure images based on perceptron model

Abstract Multi-exposure image fusion as a technical means to bridge the dynamic range gap between real scenes and image acquisition devices, which makes the fused images better quality and more realistic and vivid simulation of real scenes, has been widely concerned by scholars from various countries. In order to improve the adaptive fusion effect of multi-exposure images, this paper proposes a fusion algorithm based on multilayer perceptron (MLP) based on the perceptron model and verifies the feasibility of the algorithm by the peak signal-to-noise ratio (PSNR), correlation coefficient (PCC), structural similarity (SSMI) and HDR-VDR-2, an evaluation index of HDR image quality. Comparison with other algorithms revealed that the average PSNR of the MLP algorithm improved by 4.43% over the Ma algorithm, 7.88% over the Vanmail algorithm, 10.30% over the FMMR algorithm, 11.19% over the PMF algorithm, and 11.19% over the PMF algorithm. For PCC, the MLP algorithm improves by 20.14%, 17.46%, 2.31%, 11.24%, and 15.36% over the other algorithms in that order. For SSMI, the MLP algorithm improved by 16.99%, 8.96%, 17.17%, 14.41%, and 4.85% over the other algorithms, in that order. For HDR-VDR-2, the MLP algorithm improved by 3.02%, 2.79%, 6.84%, 4.90%, and 6.55% over the other algorithms, in that order. The results show that the MLP algorithm can avoid image artifacts while retaining more details. The MLP-based adaptive fusion method is a step further in the theoretical study of multi-exposure image fusion, which is of great significance for subsequent research and practical application by related technology vendors.

An improved defocusing adaptive style transfer method based on a stroke pyramid.

Image style transfer aims to assign a specified artist's style to a real image. However, most existing methods cannot generate textures of various thicknesses due to the rich semantic information of the input image. The image loses some semantic information through style transfer with a uniform stroke size. To address the above problems, we propose an improved multi-stroke defocus adaptive style transfer framework based on a stroke pyramid, which mainly fuses various stroke sizes in the image spatial dimension to enhance the image content interpretability. We expand the receptive field of each branch and then fuse the features generated by the multiple branches based on defocus degree. Finally, we add an additional loss term to enhance the structural features of the generated image. The proposed model is trained using the Common Objects in Context (COCO) and Synthetic Depth of Field (SYNDOF) datasets, and the peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) are used to evaluate the overall quality of the output image and its structural similarity with the content image, respectively. To validate the feasibility of the proposed algorithm, we compare the average PSNR and SSIM values of the output of the modified model and those of the original model. The experimental results show that the modified model improves the PSNR and SSIM values of the outputs by 1.43 and 0.12 on average, respectively. Compared with the single-stroke style transfer method, the framework proposed in this study improves the readability of the output images with more abundant visual expression.

An attention-based deep convolutional neural network for ultra-sparse-view CT reconstruction

X-ray Computed Tomography (CT) techniques play a vitally important role in clinical diagnosis, but radioactivity exposure can also induce the risk of cancer for patients. Sparse-view CT reduces the impact of radioactivity on the human body through sparsely sampled projections. However, images reconstructed from sparse-view sinograms often suffer from serious streaking artifacts. To overcome this issue, we propose an end-to-end attention-based mechanism deep network for image correction in this paper. Firstly, the process is to reconstruct the sparse projection by the filtered back-projection algorithm. Next, the reconstructed results are fed into the deep network for artifact correction. More specifically, we integrate the attention-gating module into U-Net pipelines, whose function is implicitly learning to emphasize relevant features beneficial for a given assignment while restraining background regions. Attention is used to combine the local feature vectors extracted at intermediate stages in the convolutional neural network and the global feature vector extracted from the coarse scale activation map. To improve the performance of our network, we fused a pre-trained ResNet50 model into our architecture. The model was trained and tested using the dataset from The Cancer Imaging Archive (TCIA), which consists of images of various human organs obtained from multiple views. This experience demonstrates that the developed functions are highly effective in removing streaking artifacts while preserving structural details. Additionally, quantitative evaluation of our proposed model shows significant improvement in peak signal-to-noise ratio (PSNR), structural similarity (SSIM), and root mean squared error (RMSE) metrics compared to other methods, with an average PSNR of 33.9538, SSIM of 0.9435, and RMSE of 45.1208 at 20 views. Finally, the transferability of the network was verified using the 2016 AAPM dataset. Therefore, this approach holds great promise in achieving high-quality sparse-view CT images.

A Hybrid Residual Attention Convolutional Neural Network for Compressed Sensing Magnetic Resonance Image Reconstruction.

We propose a dual-domain deep learning technique for accelerating compressed sensing magnetic resonance image reconstruction. An advanced convolutional neural network with residual connectivity and an attention mechanism was developed for frequency and image domains. First, the sensor domain subnetwork estimates the unmeasured frequencies of k-space to reduce aliasing artifacts. Second, the image domain subnetwork performs a pixel-wise operation to remove blur and noisy artifacts. The skip connections efficiently concatenate the feature maps to alleviate the vanishing gradient problem. An attention gate in each decoder layer enhances network generalizability and speeds up image reconstruction by eliminating irrelevant activations. The proposed technique reconstructs real-valued clinical images from sparsely sampled k-spaces that are identical to the reference images. The performance of this novel approach was compared with state-of-the-art direct mapping, single-domain, and multi-domain methods. With acceleration factors (AFs) of 4 and 5, our method improved the mean peak signal-to-noise ratio (PSNR) to 8.67 and 9.23, respectively, compared with the single-domain Unet model; similarly, our approach increased the average PSNR to 3.72 and 4.61, respectively, compared with the multi-domain W-net. Remarkably, using an AF of 6, it enhanced the PSNR by 9.87 ± 1.55 and 6.60 ± 0.38 compared with Unet and W-net, respectively.

Small training dataset convolutional neural networks for application-specific super-resolution microscopy.

Machine learning (ML) models based on deep convolutional neural networks have been used to significantly increase microscopy resolution, speed [signal-to-noise ratio (SNR)], and data interpretation. The bottleneck in developing effective ML systems is often the need to acquire large datasets to train the neural network. We demonstrate how adding a "dense encoder-decoder" (DenseED) block can be used to effectively train a neural network that produces super-resolution (SR) images from conventional microscopy diffraction-limited (DL) images trained using a small dataset [15 fields of view (FOVs)]. The ML helps to retrieve SR information from a DL image when trained with a massive training dataset. The aim of this work is to demonstrate a neural network that estimates SR images from DL images using modifications that enable training with a small dataset. We employ "DenseED" blocks in existing SR ML network architectures. DenseED blocks use a dense layer that concatenates features from the previous convolutional layer to the next convolutional layer. DenseED blocks in fully convolutional networks (FCNs) estimate the SR images when trained with a small training dataset (15 FOVs) of human cells from the Widefield2SIM dataset and in fluorescent-labeled fixed bovine pulmonary artery endothelial cells samples. Conventional ML models without DenseED blocks trained on small datasets fail to accurately estimate SR images while models including the DenseED blocks can. The average peak SNR (PSNR) and resolution improvements achieved by networks containing DenseED blocks are and , respectively. We evaluated various configurations of target image generation methods (e.g., experimentally captured a target and computationally generated target) that are used to train FCNs with and without DenseED blocks and showed that including DenseED blocks in simple FCNs outperforms compared to simple FCNs without DenseED blocks. DenseED blocks in neural networks show accurate extraction of SR images even if the ML model is trained with a small training dataset of 15 FOVs. This approach shows that microscopy applications can use DenseED blocks to train on smaller datasets that are application-specific imaging platforms and there is promise for applying this to other imaging modalities, such as MRI/x-ray, etc.

Compensation cycle consistent generative adversarial networks (Comp-GAN) for synthetic CT generation from MR scans with truncated anatomy.

MR scans used in radiotherapy can be partially truncated due to the limited field of view (FOV), affecting dose calculation accuracy in MR-based radiation treatment planning. We proposed a novel Compensation-cycleGAN (Comp-cycleGAN) by modifying the cycle-consistent generative adversarial network (cycleGAN), to simultaneously create synthetic CT (sCT) images and compensate the missing anatomy from the truncated MR images. Computed tomography (CT) and T1 MR images with complete anatomy of 79 head-and-neck patients were used for this study. The original MR images were manually cropped 10-25mm off at the posterior head to simulate clinically truncated MR images. Fifteen patients were randomly chosen for testing and the rest of the patients were used for model training and validation. Both the truncated and original MR images were used in the Comp-cycleGAN training stage, which enables the model to compensate for the missing anatomy by learning the relationship between the truncation and known structures. After the model was trained, sCT images with complete anatomy can be generated by feeding only the truncated MR images into the model. In addition, the external body contours acquired from the CT images with full anatomy could be an optional input for the proposed method to leverage the additional information of the actual body shape for each test patient. The mean absolute error (MAE) of Hounsfield units (HU), peak signal-to-noise ratio (PSNR), and structural similarity index (SSIM) were calculated between sCT and real CT images to quantify the overall sCT performance. To further evaluate the shape accuracy, we generated the external body contours for sCT and original MR images with full anatomy. The Dice similarity coefficient (DSC) and mean surface distance (MSD) were calculated between the body contours of sCT and original MR images for the truncation region to assess the anatomy compensation accuracy. The average MAE, PSNR, and SSIM calculated over test patients were 93.1 HU/91.3 HU, 26.5dB/27.4dB, and 0.94/0.94 for the proposed Comp-cycleGAN models trained without/with body-contour information, respectively. These results were comparable with those obtained from the cycleGAN model which is trained and tested on full-anatomy MR images, indicating the high quality of the sCT generated from truncated MR images by the proposed method. Within the truncated region, the mean DSC and MSD were 0.85/0.89 and 1.3/0.7mm for the proposed Comp-cycleGAN models trained without/with body contour information, demonstrating good performance in compensating the truncated anatomy. We developed a novel Comp-cycleGAN model that can effectively create sCT with complete anatomy compensation from truncated MR images, which could potentially benefit the MRI-based treatment planning.

CG-DRR：digital reconstructed radiograph generation algorithm based on Cycle-GAN

CG-DRR：digital reconstructed radiograph generation algorithm based on Cycle-GAN

A Method Noise-Based Convolutional Neural Network Technique for CT Image Denoising

Medical imaging is a complex process that capitulates images created by X-rays, ultrasound imaging, angiography, etc. During the imaging process, it also captures image noise during image acquisition, some of which are extremely corrosive, creating a disturbance that results in image degradation. The proposed work addresses the challenge to eliminate the corrosive Gaussian additive white noise from computed tomography (CT) images while preserving the fine details. The proposed approach is synthesized by amalgamating the concept of method noise with a deep learning-based framework of a convolutional neural network (CNN). The corrupted images are obtained by explicit addition of Gaussian additive white noise at multiple noise variance levels (σ = 10, 15, 20, 25). The denoised images obtained are then evaluated according to their visual quality and quantitative metrics, such as peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM). These metrics for denoised CT images are then compared with their respective values for the reference CT image. The average PSNR value of the proposed method is 25.82, the average SSIM value is 0.85, and the average computational time is 2.8760. To better understand the proposed approach’s effectiveness, an intensity profile of denoised and original medical images is plotted and compared. To further test the performance of the proposed methodology, the results obtained are also compared with that of other non-traditional methods. The critical analysis of the results shows the commendable efficiency of the proposed methodology in denoising the medical CT images corrupted by Gaussian noise. This approach can be utilized in multiple pragmatic areas of application in the field of medical image processing.

Development of an unsupervised cycle contrastive unpaired translation network for MRI-to-CT synthesis.

The purpose of this work is to develop and evaluate a novel cycle-contrastive unpaired translation network (cycleCUT) for synthetic computed tomography (sCT) generation from T1-weighted magnetic resonance images (MRI). The cycleCUT proposed in this work integrated the contrastive learning module from contrastive unpaired translation network (CUT) into the cycle-consistent generative adversarial network (cycleGAN) framework to effectively achieve unsupervised CT synthesis from MRI. The diagnostic MRI and radiotherapy planning CT images of 24 brain cancer patients were obtained and reshuffled to train the network. For comparison, the traditional cycleGAN and CUT were also implemented. The sCT images were then imported into a treatment planning system to verify their feasibility for radiotherapy planning. The mean absolute error (MAE), peak signal-to-noise ratio (PSNR), and structural similarity index (SSIM) between the sCT and the corresponding real CT images were calculated. Gamma analysis between sCT- and CT-based dose distributions was also conducted. Quantitative evaluation of an independent test set of six patients showed that the average MAE was 69.62 ± 5.68 Hounsfield Units (HU) for the proposed cycleCUT, significantly (p-value<0.05) lower than that for cycleGAN (77.02 ± 6.00 HU) and CUT (78.05 ± 8.29). The average PSNR was 28.73 ± 0.46 decibels (dB) for cycleCUT, significantly larger than that for cycleGAN (27.96 ± 0.49dB) and CUT (27.95 ± 0.69dB). The average SSIM for cycleCUT (0.918 ± 0.012) was also significantly higher than that for cycleGAN (0.906 ± 0.012) and CUT (0.903 ± 0.015). Regarding gamma analysis, cycleCUT achieved the highest passing rate (97.95 ± 1.24% at the 2%/2mm criteria and 10% dose threshold) but was not significantly different from the others. The proposed cycleCUT could be effectively trained using unaligned image data, and could generate better sCT images than cycleGAN and CUT in terms of HU number accuracy and fine structural details.

Prediction of corneal astigmatism based on corneal tomography after femtosecond laser arcuate keratotomy using a pix2pix conditional generative adversarial network.

PurposeThis study aimed to develop a deep learning model to generate a postoperative corneal axial curvature map of femtosecond laser arcuate keratotomy (FLAK) based on corneal tomography using a pix2pix conditional generative adversarial network (pix2pix cGAN) for surgical planning.MethodsA total of 451 eyes of 318 nonconsecutive patients were subjected to FLAK for corneal astigmatism correction during cataract surgery. Paired or single anterior penetrating FLAKs were performed at an 8.0-mm optical zone with a depth of 90% using a femtosecond laser (LenSx laser, Alcon Laboratories, Inc.). Corneal tomography images were acquired from Oculus Pentacam HR (Optikgeräte GmbH, Wetzlar, Germany) before and 3 months after the surgery. The raw data required for analysis consisted of the anterior corneal curvature for a range of ± 3.5 mm around the corneal apex in 0.1-mm steps, which the pseudo-color corneal curvature map synthesized was based on. The deep learning model used was a pix2pix conditional generative adversarial network. The prediction accuracy of synthetic postoperative corneal astigmatism in zones of different diameters centered on the corneal apex was assessed using vector analysis. The synthetic postoperative corneal axial curvature maps were compared with the real postoperative corneal axial curvature maps using the structural similarity index measure (SSIM) and peak signal-to-noise ratio (PSNR).ResultsA total of 386 pairs of preoperative and postoperative corneal tomography data were included in the training set, whereas 65 preoperative data were retrospectively included in the test set. The correlation coefficient between synthetic and real postoperative astigmatism (difference vector) in the 3-mm zone was 0.89, and that between surgically induced astigmatism (SIA) was 0.93. The mean absolute errors of SIA for real and synthetic postoperative corneal axial curvature maps in the 1-, 3-, and 5-mm zone were 0.20 ± 0.25, 0.12 ± 0.17, and 0.09 ± 0.13 diopters, respectively. The average SSIM and PSNR of the 3-mm zone were 0.86 ± 0.04 and 18.24 ± 5.78, respectively.ConclusionOur results showed that the application of pix2pix cGAN can synthesize plausible postoperative corneal tomography for FLAK, showing the possibility of using GAN to predict corneal tomography, with the potential of applying artificial intelligence to construct surgical planning models.

Enhancement of Medical Images through an Iterative McCann Retinex Algorithm: A Case of Detecting Brain Tumor and Retinal Vessel Segmentation

Analyzing medical images has always been a challenging task because these images are used to observe complex internal structures of the human body. This research work is based on the study of the retinal fundus and magnetic resonance images (MRI) for the analysis of ocular and cerebral abnormalities. Typically, clinical quality images of the eyes and brain have low-varying contrast, making it challenge to diagnose a specific disease. These issues can be overcome, and preprocessing or an image enhancement technique is required to properly enhance images to facilitate postprocessing. In this paper, we propose an iterative algorithm based on the McCann Retinex algorithm for retinal and brain MRI. The foveal avascular zone (FAZ) region of retinal images and the coronal, axial, and sagittal brain images are enhanced during the preprocessing step. The High-Resolution Fundus (HRF) and MR brain Oasis images databases are used, and image contrast and peak signal-to-noise ratio (PSNR) are used to assess the enhancement step parameters. The average PSNR enhancement on images from the Oasis brain MRI database was about 3 dB with an average contrast of 57.4. The average PSNR enhancement of the HRF database images was approximately 2.5 dB with a contrast average of 40 over the database. The proposed method was also validated in the postprocessing steps to observe its impact. A well-segmented image was obtained with an accuracy of 0.953 and 0.0949 on the DRIVE and STARE databases. Brain tumors were detected from the Oasis brain MRI database with an accuracy of 0.97. This method can play an important role in helping medical experts diagnose eye diseases and brain tumors from retinal images and Oasis brain images.

High capacity video hiding based on multi-resolution stationary wavelet transform and hybrid-matrix decomposition techniqueset Transform and hybrid-matrix decomposition techniques

Internet simplified digital data transferring. This data needs to be secured; so securing digital data becomes an important concern. Steganography provides security for data by inserting it into a cover and concealing it. In this paper, a steganography algorithm was introduced. This algorithm used the stationary wavelet transform (SWT) and اhybrid-matrix decomposition techniques, singular value decomposition (SVD) and QR factorization to conceal a video in another video. The algorithm performance was measured with reference to the peak signal to noise ratio (PSNR) and structural similarity index measure (SSIM) for the cover and the secret videos. The algorithm successfully hid a video in another cover video and both are of the same size; the hiding capacity was 100%. The algorithm achieved a SSIM that reached 0.97 and a high PSNR value that reached 68.8 proving that the imperceptibility of the proposed algorithm is very high. The comparative analysis shows that the suggested algorithm achieved higher imperceptibility than the other state-of-the-art algorithms regarding the average PSNR. The enhanced version of the proposed method was more robust against different types of attacks.

Adaptive compressed sensing algorithm for terahertz spectral image reconstruction based on residual learning

Terahertz time-domain spectroscopy (THz-TDS) is widely applied in the field of rapid nondestructive detection of grain owing to its low photon energy and high penetrating power. Nevertheless, terahertz imaging systems suffer from the problems of long image acquisition time and massive data processing. To mitigate these issues, this work presents an adaptive compressed sensing reconstruction algorithm for terahertz spectral images based on residual learning (ATResCS). The algorithm compresses the number of data samples, reducing the amount of data required for imaging and improving the imaging speed. Further, ATResCS reduces the time complexity by employing a convolutional neural network. The algorithm is validated by acquiring terahertz spectral image data via a THz-TDS system. ATResCS outperforms conventional algorithms regarding peak signal-to-noise ratio (PSNR) and structural similarity, significantly reducing the reconstruction time and, thus, enabling real-time reconstruction. Specifically, at low sampling rates (0.1), ATResCS retains key spectral image information. The average PSNR is 0.96 – 1.015 dB higher than that of DR2-Net, reducing the average reconstruction time by 0.1 – 0.2 s. Experiments demonstrate that ATResCS has better reconfiguration capability and lower algorithm complexity, enabling high-quality and fast reconstruction of terahertz spectral images.

DR-only Carbon-ion radiotherapy treatment planning via deep learning

PurposeTo evaluate the feasibility of patient-specific digital radiography (DR)-only treatment planning for carbon ion radiotherapy in anthropomorphic thorax-and-abdomen phantom and head-and-neck patients. MethodsThe study was conducted on the anthropomorphic phantom and head-and-neck patients. We collected computed tomography (CT) and DR images of the phantom and cone beam CT (CBCT) and DR images of the patients, respectively. Two different deep neural networks were established to correlate the relationships between DR and digitally reconstructed radiograph (DRR) images, as well as DRR and CT images. The similarity between CT and predicted CT images was evaluated by computing the mean absolute error (MAE), root mean square error (RMSE), peak signal-to-noise ratio (PSNR) and structural similarity (SSIM), respectively. Dose calculations on the predicted CT images were compared against the true CT-based dose distributions for carbon-ion radiotherapy treatment planning with intensity-modulated pencil-beam spot scanning. Relative dose differences in the target volumes and organ-at-risks were computed and three-dimensional gamma analyses (3 mm, 3%) were performed. ResultsThe average MAE, RMSE, PSNR and SSIM of the framework were 0.007, 0.144, 37.496 and 0.973, respectively. The average relative dose differences between the predicted CT- and CT-based dose distributions at the same carbon-ion irradiation settings for the phantom and the patients were <2% and ≤4%, respectively. The average gamma pass-rates were >98% for the predicted CT-based versus CT-based carbon ion plans of the phantom and the patients. ConclusionWe have demonstrated the feasibility of a patient-specific DR-only treatment planning workflow for heavy ion radiotherapy by using deep learning approach.

Walsh–Hadamard Kernel Feature-Based Image Compression Using DCT with Bi-Level Quantization

To meet the high bit rate requirements in many multimedia applications, a lossy image compression algorithm based on Walsh–Hadamard kernel-based feature extraction, discrete cosine transform (DCT), and bi-level quantization is proposed in this paper. The selection of the quantization matrix of the block is made based on a weighted combination of the block feature strength (BFS) of the block extracted by projecting the selected Walsh–Hadamard basis kernels on an image block. The BFS is compared with an automatically generated threshold for applying the specific quantization matrix for compression. In this paper, higher BFS blocks are processed via DCT and high Q matrix, and blocks with lower feature strength are processed via DCT and low Q matrix. So, blocks with higher feature strength are less compressed and vice versa. The proposed algorithm is compared to different DCT and block truncation coding (BTC)-based approaches based on the quality parameters, such as peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) at constant bits per pixel (bpp). The proposed method shows significant improvements in performance over standard JPEG and recent approaches at lower bpp. It achieved an average PSNR of 35.61 dB and an average SSIM of 0.90 at a bpp of 0.5 and better perceptual quality with lower visual artifacts.

Deep Learning-Based Ultrasound Combined with Gastroscope for the Diagnosis and Nursing of Upper Gastrointestinal Submucous Lesions.

The study focused on the diagnostic value of deep learning-based ultrasound combined with gastroscope examination for upper gastrointestinal submucous lesions and nursing. A total of 104 patients with upper gastrointestinal submucous lesions diagnosed in hospital were selected as the research subjects. In this study, the feed forward denoising convulsive neural network (DnCNN) was improved, and the n-DnCNN model was designed and applied to ultrasonic image processing. The peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) of Gaussian filtering, NL-means, and DnCNN were then compared with n-DnCNN. Subsequently, the distribution and types of submucosal lesions in different parts of the upper digestive tract were analyzed by ultrasound combined with gastroscope and gastroscope examination alone, and the diagnostic performance of this method was evaluated. The results showed that the average PSNR and SSIM of the n-DnCNN model were 33.01 dB and 0.87, respectively, which were significantly higher than GF, NL-means, and DnCNN algorithms, and the difference was statistically significant (P < 0.05). Of the 116 lesions detected, 49 were located in the esophagus (42.24%), 52 in the stomach (44.83%), and 15 in the duodenum (12.93%). Of the 49 esophageal submucosal lesions, 6.12% were located in the upper esophagus, 55.1% in the middle esophagus, and 38.79% in the lower esophagus, and the difference was statistically significant (P < 0.05). Of the gastric submucosal lesions, the lesions in the gastric cardia were significantly less than in other parts, and the difference was statistically significant (P < 0.05). The accuracy of ultrasound combined with gastroscope in the diagnosis of upper gastrointestinal submucous episodes was 82.32%, higher than that of gastroscope examination, and the difference was statistically significant (P < 0.05). In conclusion, the n-DnCNN model has a good noise reduction effect, and the obtained image is of high quality. Ultrasound combined with gastroscope examination can effectively improve the accuracy of diagnosis of upper gastrointestinal submucous lesions.

Delayed PET imaging using image synthesis network and nonrigid registration without additional CT scan.

Attenuation correction is critical for positron emission tomography (PET) image reconstruction. The standard protocol for obtaining attenuation information in a clinical PET scanner is via coregistered computed tomography (CT) images. Therefore, for delayed PET imaging, the CT scan is repeated twice, which increases the radiation dose for the patient. In this paper, we propose a zero-extradose delayed PET imaging method that requires no additional CT scans. A deep learning-based synthesis network is designed to convert PET data into pseudo-CT images for delayed scans. Then, nonrigid registration is performed between this pseudo CT image and the CT image of the first scan, warping the CT image of the first scan to an estimated CT image for the delayed scan. Finally, the PET image attenuation correction in the delayed scan is obtained from this estimated CT image. Experiments with clinical datasets are implemented to assess the effectiveness of the proposed method with the well-recognized Generative Adversarial Networks (GAN) method. The average peak signal-to-noise ratio (PSNR) and the mean absolute percent error (MAPE) are used for comparison. We also use scoring from three experienced radiologists as subjective measurement means based on the diagnostic consistency of the PET images reconstructed from GAN and the proposed method with respect to the ground truth images. The experiments show that the average PSNR is 47.04dB (the proposed method) vs. 44.41dB (the traditional GAN method) for the reconstructed delayed PET images in our evaluation dataset. The average MAPEs are 1.59% for the proposed method and 3.32% for the traditional GAN method across five organ regions of interest (ROIs). The scores for the GAN and the proposed method rated by three experienced radiologists are 8.08±0.60 and 9.02±0.52, indicating that the proposed method yields more consistent PET images with the ground truth. This work proposes a novel method for CT-less delayed PET imaging based on image synthesis network and nonrigid image registration. The PET image reconstructed using the proposed method yields delayed PET images with high image quality without artifacts and is quantitatively more accurate than the traditional GAN method.

Multiscale deep network for compressive sensing image reconstruction

Deep learning-based image compressive sensing methods have received extensive attention in recent years due to their superior learning ability and fast processing speed. The majority of existing image compressive sensing neural networks use single-scale sampling, whereas multiscale sampling has demonstrated excellent performance compared to single-scale. We propose a multiscale deep network for compressive sensing image reconstruction that consists of a multiscale sampling network and a reconstruction network. First, we use convolution to mimic the linear decomposition of images, and the convolution is learned during the training process. Then a sampling network captures compressive measurements across multiple decomposed scales. The reconstruction network, which includes both the initial and enhanced reconstruction networks, learns an end-to-end mapping between the compressed sensing (CS) measurements and the recovered images of the network. Experimental results indicate that the proposed network framework outperforms the existing CS methods in terms of objective metrics, peak signal to noise ratio (PSNR), structural similarity index, and subjective visual quality. Specifically, at a 0.1 sampling rate, using 10 images for testing, and the average PSNR maximum (minimum) gain is 5.95 dB (0.25 dB).

An Energy-Efficient Wearable Functional Near-infrared Spectroscopy System Employing Dual-level Adaptive Sampling Technique.

Functional near-infrared spectroscopy (fNIRS) is a powerful medical imaging tool in brain science and psychology, it can also be employed in brain-computer interface (BCI) due to its noninvasive and artifact-less-sensitive characteristics. Conventional ways to detect large-area brain activity using near-infrared (NIR) technology are based on Time-division or Frequency-division modulation technique, which traverses all physical sensory channels in a specific period. To achieve higher imaging resolution or brain-tasks classification accuracy, the NIRS system require higher density and more channels, which conflict with the limited battery capacity. Inspired by the functional atlas of the human brain, this paper proposes a spatial adaptive sampling (SAS) method. It can change the active channel pattern of the fNIRS system to match with the real-time brain activity, to increase the energy efficiency without significant reduction on the brain imaging quality or the accuracy of brain activity classification. Therefore, the number of the averaging enabled channels will be dramatically reduced in practice. To verify the proposed SAS technique, a wearable and flexible NIRS system has been implemented, in which each channel of light-emitting diode (LED) drive circuits and photodiode (PD) detection circuits can be power gated independently. Brain task experiments have been conducted to validate the proposed method, the power consumption of the LED drive module is reduced by 46.58% compared to that without SAS technology while maintaining an average brain imaging PSNR (Peak Signal to Noise Ratio) of 35 dB. The brain-task classification accuracy is 80.47%, which has a 2.67% reduction compared to that without the SAS technique.