基于深度学习的低剂量CT图像去噪

目的:提出一种基于深度学习的方法用于低剂量CT（LDCT）图像的噪声去除。方法:首先进行滤波反投影重建,然后利用多尺度并行残差U-net（MPR U-net）的深度学习模型对重建后的LDCT图像进行去噪。实验数据采用LoDoPaB-CT挑战赛的医学CT数据集,其中训练集35 820张图像,验证集3 522张图像,测试集3 553张图像,并采用峰值信噪比（PSNR）与结构相似性系数（SSIM）来评估模型的去噪效果。结果:LDCT图像处理前后PSNR分别为28.80、38.22 dB,SSIM分别为0.786、0.966,平均处理时间为0.03 s。结论:MPR U-net深度学习模型能较好地去除LDCT图像噪声,提升PSNR,保留更多图像细节。     

基于深度学习单投影的CT断层成像三维重建

【摘要】目的:提出一种基于深度学习的计算机断层扫描（CT）单视图断层成像三维（3D）重建方法,在减少数据采集量和降低成像剂量的情况下对不同患者进行CT图像的3D重建。方法:对不同患者的CT图像进行数据增强和模拟生成对应的数字重建放射影像（DRR）,并进行数据归一化操作。利用预处理后的数据通过卷积神经网络训练出一个普适于不同患者的神经网络模型。将训练好的神经网络模型部署在测试数据集上,使用平均绝对误差（MAE）、均方根误差（RMSE）、结构相似性（SSIM）和峰值信噪比（PSNR）对重建结果进行评估。结果:定性和定量分析的结果表明,该方法可以使用不同患者的单张2D图像分别重建出质量较高的3D CT图像,MAE、RMSE、SSIM和PSNR分别为0.006、0.079、0.982、38.424 dB。此外,相比特定于单个患者的情况,该方法可以大幅度提高重建速度并节省70%的模型训练时间。结论:构建的神经网络模型可通过不同患者的2D单视图重建出相应患者的3D CT图像。因此,本研究对简化临床成像设备和放射治疗当中的图像引导具有重要作用。     

Whole‐brain perfusion imaging with balanced steady‐state free precession arterial spin labeling

Recently, balanced steady‐state free precession (bSSFP) readout has been proposed for arterial spin labeling (ASL) perfusion imaging to reduce susceptibility artifacts at a relatively high spatial resolution and signal‐to‐noise ratio (SNR). However, the main limitation of bSSFP‐ASL is the low spatial coverage. In this work, methods to increase the spatial coverage of bSSFP‐ASL are proposed for distortion‐free, high‐resolution, whole‐brain perfusion imaging. Three strategies of (i) segmentation, (ii) compressed sensing (CS) and (iii) a hybrid approach combining the two methods were tested to increase the spatial coverage of pseudo‐continuous ASL (pCASL) with three‐dimensional bSSFP readout. The spatial coverage was increased by factors of two, four and six using each of the three approaches, whilst maintaining the same total scan time (5.3 min). The number of segments and/or CS acceleration rate (R) correspondingly increased to maintain the same bSSFP readout time (1.2 s). The segmentation approach allowed whole‐brain perfusion imaging for pCASL‐bSSFP with no penalty in SNR and/or total scan time. The CS approach increased the spatial coverage of pCASL‐bSSFP whilst maintaining the temporal resolution, with minimal impact on the image quality. The hybrid approach provided compromised effects between the two methods. Balanced SSFP‐based ASL allows the acquisition of perfusion images with wide spatial coverage, high spatial resolution and SNR, and reduced susceptibility artifacts, and thus may become a good choice for clinical and neurological studies. Copyright © 2015 John Wiley & Sons, Ltd.     

Compressed sensing MRI with singular value decomposition-based sparsity basis

Compressed sensing MRI (CS-MRI) aims to significantly reduce the measurements required for image reconstruction in order to accelerate the overall imaging speed. The sparsity of the MR images in transformation bases is one of the fundamental criteria for CS-MRI performance. Sparser representations can require fewer samples necessary for a successful reconstruction or achieve better reconstruction quality with a given number of samples. Generally, there are two kinds of 'sparsifying' transforms: predefined transforms and data-adaptive transforms. The predefined transforms, such as the discrete cosine transform, discrete wavelet transform and identity transform have usually been used to provide sufficiently sparse representations for limited types of MR images, in view of their isolation to the object images. In this paper, we present singular value decomposition (SVD) as the data-adaptive 'sparsity' basis, which can sparsify a broader range of MR images and perform effective image reconstruction. The performance of this method was evaluated for MR images with varying content (for example, brain images, angiograms, etc), in terms of image quality, reconstruction time, sparsity and data fidelity. Comparison with other commonly used sparsifying transforms shows that the proposed method can significantly accelerate the reconstruction process and still achieve better image quality, providing a simple and effective alternative solution in the CS-MRI framework.     

Evaluation of segmented 3D acquisition schemes for whole‐brain high‐resolution arterial spin labeling at 3 T

Recent technical developments have significantly increased the signal‐to‐noise ratio (SNR) of arterial spin labeled (ASL) perfusion MRI. Despite this, typical ASL acquisitions still employ large voxel sizes. The purpose of this work was to implement and evaluate two ASL sequences optimized for whole‐brain high‐resolution perfusion imaging, combining pseudo‐continuous ASL (pCASL), background suppression (BS) and 3D segmented readouts, with different in‐plane k‐space trajectories. Identical labeling and BS pulses were implemented for both sequences. Two segmented 3D readout schemes with different in‐plane trajectories were compared: Cartesian (3D GRASE) and spiral (3D RARE Stack‐Of‐Spirals). High‐resolution perfusion images (2 × 2 × 4 mm³) were acquired in 15 young healthy volunteers with the two ASL sequences at 3 T. The quality of the perfusion maps was evaluated in terms of SNR and gray‐to‐white matter contrast. Point‐spread‐function simulations were carried out to assess the impact of readout differences on the effective resolution. The combination of pCASL, in‐plane segmented 3D readouts and BS provided high‐SNR high‐resolution ASL perfusion images of the whole brain. Although both sequences produced excellent image quality, the 3D RARE Stack‐Of‐Spirals readout yielded higher temporal and spatial SNR than 3D GRASE (spatial SNR = 8.5 ± 2.8 and 3.7 ± 1.4; temporal SNR = 27.4 ± 12.5 and 15.6 ± 7.6, respectively) and decreased through‐plane blurring due to its inherent oversampling of the central k‐space region, its reduced effective T_E and shorter total readout time, at the expense of a slight increase in the effective in‐plane voxel size. Copyright © 2014 John Wiley & Sons, Ltd.     

Patch-based local learning method for cerebral blood flow quantification with arterial spin-labeling MRI

Arterial spin-labeling (ASL) perfusion MRI is a non-invasive method for quantifying cerebral blood flow (CBF). Standard ASL CBF calibration mainly relies on pair-wise subtraction of the spin-labeled images and controls images at each voxel separately, ignoring the abundant spatial correlations in ASL data. To address this issue, we previously proposed a multivariate support vector machine (SVM) learning-based algorithm for ASL CBF quantification (SVMASLQ). But the original SVMASLQ was designed to do CBF quantification for all image voxels simultaneously, which is not ideal for considering local signal and noise variations. To fix this problem, we here in this paper extended SVMASLQ into a patch-wise method by using a patch-wise classification kernel. At each voxel, an image patch centered at that voxel was extracted from both the control images and labeled images, which was then input into SVMASLQ to find the corresponding patch of the surrogate perfusion map using a non-linear SVM classifier. Those patches were eventually combined into the final perfusion map. Method evaluations were performed using ASL data from 30 young healthy subjects. The results showed that the patch-wise SVMASLQ increased perfusion map SNR by 6.6% compared to the non-patch-wise SVMASLQ.     

磁共振部分K空间重建算法

磁共振成像（MRI）是必要的获取临床图像的影像学方法之一,但是它获取数据过程缓慢使得成像时间过长。目前提出了许多高效的成像算法来降低磁共振的成像时间,如半傅里叶成像和压缩感知MRI等。半傅里叶成像仅采用多于一半的K空间数据进行图像重建,不仅提高了MRI的成像速度,而且降低了运动伪影,是有效的部分K空间重建技术之一。基于压缩感知理论的MRI仅采用25%~30%的K空间数据就能重建出MRI图像,与其它成像技术相比,可在相同的扫描时间内获得更高质量的MRI图像,也可在相同的空间分辨率下加速成像。本文综述几种半傅里叶成像算法的原理,也阐述了压缩感知理论与MRI相结合的原理,包括MR图像的稀疏表示、K空间的采样轨迹设计、重建算法的选择等。     

基于3DV-Net的肺结节检测分割算法

目的：提出基于深度学习的肺结节识别与分割算法,以辅助医生进行肺部疾病检测。方法：针对LUNA16数据集数据量大以及肺结节种类大小多样性等特征,采用基于改进的深度神经网络3DV-Net实现多种肺结节的检测分割,然后使用ResNet对结节图像和非结节图像进行分类。对LUNA16数据集中的肺部CT图像进行图像去噪、插值采样等预处理,然后生成粗分割图像和Mask图像,再使用改进的3DV-Net模型对数据进行多次训练预测。网络层级越深,出现梯度消散、梯度爆炸等问题的概率越大,改进的3DV-Net使用残差连接来改善这一问题。结果：改进的3DV-Net的Dice相似系数和IoU分别达到88.29%和88.25%。结论：本文方法有助于肺结节的检测分割,在肺结节的辅助诊断方面有重要意义。     

Deep learning‐based fully automatic segmentation of wrist cartilage in MR images

The study objective was to investigate the performance of a dedicated convolutional neural network (CNN) optimized for wrist cartilage segmentation from 2D MR images. CNN utilized a planar architecture and patch‐based (PB) training approach that ensured optimal performance in the presence of a limited amount of training data. The CNN was trained and validated in 20 multi‐slice MRI datasets acquired with two different coils in 11 subjects (healthy volunteers and patients). The validation included a comparison with the alternative state‐of‐the‐art CNN methods for the segmentation of joints from MR images and the ground‐truth manual segmentation. When trained on the limited training data, the CNN outperformed significantly image‐based and PB‐U‐Net networks. Our PB‐CNN also demonstrated a good agreement with manual segmentation (Sørensen–Dice similarity coefficient [DSC] = 0.81) in the representative (central coronal) slices with a large amount of cartilage tissue. Reduced performance of the network for slices with a very limited amount of cartilage tissue suggests the need for fully 3D convolutional networks to provide uniform performance across the joint. The study also assessed inter‐ and intra‐observer variability of the manual wrist cartilage segmentation (DSC = 0.78‐0.88 and 0.9, respectively). The proposed deep learning‐based segmentation of the wrist cartilage from MRI could facilitate research of novel imaging markers of wrist osteoarthritis to characterize its progression and response to therapy.     

Sharpness-Aware Low-Dose CT Denoising Using Conditional Generative Adversarial Network

Low-dose computed tomography (LDCT) has offered tremendous benefits in radiation-restricted applications, but the quantum noise as resulted by the insufficient number of photons could potentially harm the diagnostic performance. Current image-based denoising methods tend to produce a blur effect on the final reconstructed results especially in high noise levels. In this paper, a deep learning-based approach was proposed to mitigate this problem. An adversarially trained network and a sharpness detection network were trained to guide the training process. Experiments on both simulated and real dataset show that the results of the proposed method have very small resolution loss and achieves better performance relative to state-of-the-art methods both quantitatively and visually.     

融合感知损失的深度学习在常规MR图像转换的研究

目的:研究在完全无监督的条件下深度神经网络实现常规磁共振图像间相互转换的可行性。方法:在循环生成式对抗网络(CycleGAN)中引入感知损失,使网络利用对抗损失学习图像结构信息的同时,结合循环一致性损失和感知损失生成高质量的磁共振图像,并将生成图像与CycleGAN模型以及有监督的CycleGAN模型(S_CycleGAN)生成的图像进行定量比较。结果:引入感知损失后的网络生成的图像定量评估值均高于CycleGAN模型生成的图像,生成的T1加权图像(T1WI)的定量评估值也均高于S_CycleGAN模型生成的T1WI,生成的T2加权图像(T2WI)与S_CycleGAN模型生成的T2WI的定量评估值相似。结论:在CycleGAN中引入感知损失,可以在完全无监督的条件下生成高质量的磁共振图像,进而实现高质量的常规磁共振图像的相互转换。     

A Bayesian-MAP Method Based on TV for CT Image Reconstruction from Sparse and Limited Data

Computed tomography(CT) plays an important role in the field of modern medical imaging. Reducing radiation exposure dose without significantly decreasing image's quality is always a crucial issue. Inspired by the outstanding performance of total variation(TV) technique in CT image reconstruction, a TV regularization based Bayesian-MAP(MAP-TV) is proposed to reconstruct the case of sparse view projection and limited angle range imaging. This method can suppress the streak artifacts and geometrical deformation while preserving image edges. We used ordered subset(OS) technique to accelerate the reconstruction speed. Numerical results show that MAP-TV is able to reconstruct a phantom with better visual performance and quantitative evaluation than classical FBP,MLEM and quadrate prior to MAP algorithms. The proposed algorithm can be generalized to cone-beam CT image reconstruction.     

基于剂量预测和自动勾画技术的PET/CT器官内照射剂量率快速评估方法

目的：实现一种基于深度学习的剂量预测和自动勾画技术的正电子发射断层成像（PET）/CT检查器官内照射剂量率的快速评估方法。方法：首先基于患者特定时刻的PET/CT图像,使用蒙特卡罗程序GATE进行内照射剂量率计算,获得每个患者的剂量率分布图。随后,基于U-Net构建深度神经网络,将患者的CT和PET图像作为输入,GATE计算的剂量率图作为金标准进行训练。训练后的深度学习模型能够根据患者的CT和PET图像预测对应的剂量率分布。同时,使用勾画软件DeepViewer对患者CT图像中的器官和组织进行自动勾画,结合预测得到的剂量率分布结果计算相应器官和组织的吸收剂量率。使用50名患者的PET/CT数据,其中10份用于测试,其余40份进行4折交叉训练,每次使用30份用于训练,10份用于验证。将测试集结果与GATE和GPU蒙特卡罗工具ARCHER-NM进行对比。结果：在自动勾画软件DeepViewer勾画的24个器官中,绝大部分器官的深度学习预测剂量率与GATE计算结果偏差在±10%以内。其中大脑、心脏、肝脏、左肺、右肺的平均偏差分别为3.3%、1.1%、1.0%、-1.1%、0.0%,与GATE...     

基于卷积神经网络的超声图像散斑去噪算法

为了提高超声图像质量,解决传统去噪算法在抑制散斑噪声和保留超声图像纹理特征方面的难题,提出一种基于卷积神经网络的超声图像散斑去噪算法DSCNN（De-speckling CNN）。本文提出的算法利用卷积神经网络强大的拟合能力来学习从超声图像到其相应的高质量图像的复杂映射,同时,通过改进损失函数的方式来减少去噪过程中纹理信息的损失和细节的模糊。不同于以往简单地假设超声散斑噪声为乘性噪声,本文利用基于超声图像采集模型和散斑噪声形成模型的模拟超声成像技术为去噪模型生成更贴合真实超声图像的训练数据,解决深度学习方法训练数据匮乏以及在临床上无法获得与超声图像空间配准作为标签的无噪声图像的难题。通过与其他具有代表性的超声图像去噪算法比较,经DSCNN去噪后的超声图像无论在视觉效果还是图像质量评价指标上都取得了更好的结果,其中SSIM达到0.856 9,在文中所有方法中最高。     

DIMENSION: Dynamic MR imaging with both k-space and spatial prior knowledge obtained via multi-supervised network training

Dynamic MR image reconstruction from incomplete k-space data has generated great research interest due to its capability in reducing scan time. Nevertheless, the reconstruction problem is still challenging due to its ill-posed nature. Most existing methods either suffer from long iterative reconstruction time or explore limited prior knowledge. This paper proposes a dynamic MR imaging method with both k-space and spatial prior knowledge integrated via multi-supervised network training, dubbed as DIMENSION. Specifically, the DIMENSION architecture consists of a frequential prior network for updating the k-space with its network prediction and a spatial prior network for capturing image structures and details. Furthermore, a multi-supervised network training technique is developed to constrain the frequency domain information and the spatial domain information. The comparisons with classical k-t FOCUSS, k-t SLR, L+S and the state-of-the-art CNN-based method on in vivo datasets show our method can achieve improved reconstruction results in shorter time.     

DEep learning-based rapid Spiral Image REconstruction (DESIRE) for high-resolution spiral first-pass myocardial perfusion imaging

The objective of the current study was to develop and evaluate a DEep learning-based rapid Spiral Image REconstruction (DESIRE) technique for high-resolution spiral first-pass myocardial perfusion imaging with whole-heart coverage, to provide fast and accurate image reconstruction for both single-slice (SS) and simultaneous multislice (SMS) acquisitions. Three-dimensional U-Net–based image enhancement architectures were evaluated for high-resolution spiral perfusion imaging at 3 T. The SS and SMS MB = 2 networks were trained on SS perfusion images from 156 slices from 20 subjects. Structural similarity index (SSIM), peak signal-to-noise ratio (PSNR), and normalized root mean square error (NRMSE) were assessed, and prospective images were blindly graded by two experienced cardiologists (5: excellent; 1: poor). Excellent performance was demonstrated for the proposed technique. For SS, SSIM, PSNR, and NRMSE were 0.977 [0.972, 0.982], 42.113 [40.174, 43.493] dB, and 0.102 [0.080, 0.125], respectively, for the best network. For SMS MB = 2 retrospective data, SSIM, PSNR, and NRMSE were 0.961 [0.950, 0.969], 40.834 [39.619, 42.004] dB, and 0.107 [0.086, 0.133], respectively, for the best network. The image quality scores were 4.5 [4.1, 4.8], 4.5 [4.3, 4.6], 3.5 [3.3, 4], and 3.5 [3.3, 3.8] for SS DESIRE, SS L1-SPIRiT, MB = 2 DESIRE, and MB = 2 SMS-slice-L1-SPIRiT, respectively, showing no statistically significant difference (p = 1 and p = 1 for SS and SMS, respectively) between L1-SPIRiT and the proposed DESIRE technique. The network inference time was ~100 ms per dynamic perfusion series with DESIRE, while the reconstruction time of L1-SPIRiT with GPU acceleration was ~ 30 min. It was concluded that DESIRE enabled fast and high-quality image reconstruction for both SS and SMS MB = 2 whole-heart high-resolution spiral perfusion imaging.     

基于深度残差网络的皮肤镜图像黑色素瘤的识别

恶性黑色素瘤是最常见和最致命的皮肤癌之一。临床上,皮肤镜检查是恶性黑色素瘤早期诊断的常规手段。但是人工检查费力、费时,并且高度依赖于皮肤科医生的临床经验。因此,研究出自动识别皮肤镜图像中的黑色素瘤算法显得尤为重要。提出一种皮肤镜图像自动评估的新框架,利用深度学习方法,使其在有限的训练数据下产生更具区分性的特征。具体来说,首先在大规模自然图像数据集上预训练一个深度为152层的残差神经网络(Res-152),用来提取皮肤病变图像的深度卷积层特征,并对其使用均值池化得到特征向量,然后利用支持向量机(SVM)对提取的黑色素瘤特征进行分类。在公开的皮肤病变图像ISBI 2016挑战数据集中,用所提出的方法对248幅黑色素瘤图像和1 031幅非黑色素瘤图像进行评估,达到86.28%的准确率及84.18%的AUC值。同时,为论证神经网络深度对分类结果的影响,比较不同深度的模型框架。与现有使用传统手工特征的研究(如基于密集采样SIFT描述符的词袋模型)相比,或仅从深层神经网络的全连接层提取特征进行分类的方法相比,新方法能够产生区分性能更强的特征表达,可以在有限的训练数据下解决黑色素瘤的类内差异大、黑色素瘤与非黑素瘤之间的类间差异小的问题。     

EIT image reconstruction with four dimensional regularization

Electrical impedance tomography (EIT) reconstructs internal impedance images of the body from electrical measurements on body surface. The temporal resolution of EIT data can be very high, although the spatial resolution of the images is relatively low. Most EIT reconstruction algorithms calculate images from data frames independently, although data are actually highly correlated especially in high speed EIT systems. This paper proposes a 4-D EIT image reconstruction for functional EIT. The new approach is developed to directly use prior models of the temporal correlations among images and 3-D spatial correlations among image elements. A fast algorithm is also developed to reconstruct the regularized images. Image reconstruction is posed in terms of an augmented image and measurement vector which are concatenated from a specific number of previous and future frames. The reconstruction is then based on an augmented regularization matrix which reflects the a priori constraints on temporal and 3-D spatial correlations of image elements. A temporal factor reflecting the relative strength of the image correlation is objectively calculated from measurement data. Results show that image reconstruction models which account for inter-element correlations, in both space and time, show improved resolution and noise performance, in comparison to simpler image models.     

Alternative reconstruction algorithm for NMR imaging

A new algorithm for image reconstruction is proposed. Because a varying sampling increment is introduced, the interpolation during reconstruction is eliminated and the Fourier transform from modified polar to Cartesian sample grid is applied. The comparison of the back-projection of filtered projections method with the proposed PCFT algorithm is performed on computer-simulated data. Good resolution of the reconstructed images and a relatively short reconstruction time are the merits of the newly developed technique NMR imaging is cited as a possible field of application of the algorithm.     

Deep Transfer Learning for Automated Intestinal Bleeding Detection in Capsule Endoscopy Imaging

Purpose: The objective of this paper was to develop a computer-aided diagnostic (CAD) tools for automated analysis of capsule endoscopic (CE) images, more precisely, detect small intestinal abnormalities like bleeding. Methods: In particular, we explore a convolutional neural network (CNN)-based deep learning framework to identify bleeding and non-bleeding CE images, where a pre-trained AlexNet neural network is used to train a transfer learning CNN that carries out the identification. Moreover, bleeding zones in a bleeding-identified image are also delineated using deep learning-based semantic segmentation that leverages a SegNet deep neural network. Results: To evaluate the performance of the proposed framework, we carry out experiments on two publicly available clinical datasets and achieve a 98.49% and 88.39% F1 score, respectively, on the capsule endoscopy.org and KID datasets. For bleeding zone identification, 94.42% global accuracy and 90.69% weighted intersection over union (IoU) are achieved. Conclusion: Finally, our performance results are compared to other recently developed state-of-the-art methods, and consistent performance advances are demonstrated in terms of performance measures for bleeding image and bleeding zone detection. Relative to the present and established practice of manual inspection and annotation of CE images by a physician, our framework enables considerable annotation time and human labor savings in bleeding detection in CE images, while providing the additional benefits of bleeding zone delineation and increased detection accuracy. Moreover, the overall cost of CE enabled by our framework will also be much lower due to the reduction of manual labor, which can make CE affordable for a larger population.