Search

Colorectal cancer (CRC) is a common malignant tumor that seriously threatens human health. CRC presents a formidable challenge in terms of accurate identification due to its indistinct boundaries. With the widespread adoption of convolutional neural networks (CNNs) in image processing, leveraging CNNs for automatic classification and segmentation holds immense potential for enhancing the efficiency of colorectal cancer recognition and reducing treatment costs. This paper explores the imperative necessity for applying CNNs in clinical diagnosis of CRC. It provides an elaborate overview on research advancements pertaining to CNNs and their improved models in CRC classification and segmentation. Furthermore, this work summarizes the ideas and common methods for optimizing network performance and discusses the challenges faced by CNNs as well as future development trends in their application towards CRC classification and segmentation, thereby promoting their utilization within clinical diagnosis.

In response to the issues of single-scale information loss and large model parameter size during the sampling process in U-Net and its variants for medical image segmentation, this paper proposes a multi-scale medical image segmentation method based on pixel encoding and spatial attention. Firstly, by redesigning the input strategy of the Transformer structure, a pixel encoding module is introduced to enable the model to extract global semantic information from multi-scale image features, obtaining richer feature information. Additionally, deformable convolutions are incorporated into the Transformer module to accelerate convergence speed and improve module performance. Secondly, a spatial attention module with residual connections is introduced to allow the model to focus on the foreground information of the fused feature maps. Finally, through ablation experiments, the network is lightweighted to enhance segmentation accuracy and accelerate model convergence. The proposed algorithm achieves satisfactory results on the Synapse dataset, an official public dataset for multi-organ segmentation provided by the International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI), with Dice similarity coefficient (DSC) and 95% Hausdorff distance (HD95) scores of 77.65 and 18.34, respectively. The experimental results demonstrate that the proposed algorithm can enhance multi-organ segmentation performance, potentially filling the gap in multi-scale medical image segmentation algorithms, and providing assistance for professional physicians in diagnosis.

Magnetic resonance imaging (MRI) plays a crucial role in the diagnosis of ischemic stroke. Accurate segmentation of the infarct is of great significance for selecting intervention treatment methods and evaluating the prognosis of patients. To address the issue of poor segmentation accuracy of existing methods for multiscale stroke lesions, a novel encoder-decoder architecture network based on depthwise separable convolution is proposed. Firstly, this network replaces the convolutional layer modules of the U-Net with redesigned depthwise separable convolution modules. Secondly, an modified Atrous spatial pyramid pooling (MASPP) is introduced to enlarge the receptive field and enhance the extraction of multiscale features. Thirdly, an attention gate (AG) structure is incorporated at the skip connections of the network to further enhance the segmentation accuracy of multiscale targets. Finally, Experimental evaluations are conducted using the ischemic stroke lesion segmentation 2022 challenge (ISLES2022) dataset. The proposed algorithm in this paper achieves Dice similarity coefficient (DSC), Hausdorff distance (HD), sensitivity (SEN), and precision (PRE) scores of 0.816 5, 3.668 1, 0.889 2, and 0.894 6, respectively, outperforming other mainstream segmentation algorithms. The experimental results demonstrate that the method in this paper effectively improves the segmentation of infarct lesions, and is expected to provide a reliable support for clinical diagnosis and treatment.

In computer-aided medical diagnosis, obtaining labeled medical image data is expensive, while there is a high demand for model interpretability. However, most deep learning models currently require a large amount of data and lack interpretability. To address these challenges, this paper proposes a novel data augmentation method for medical image segmentation. The uniqueness and advantages of this method lie in the utilization of gradient-weighted class activation mapping to extract data efficient features, which are then fused with the original image. Subsequently, a new channel weight feature extractor is constructed to learn the weights between different channels. This approach achieves non-destructive data augmentation effects, enhancing the model's performance, data efficiency, and interpretability. Applying the method of this paper to the Hyper-Kvasir dataset, the intersection over union (IoU) and Dice of the U-net were improved, respectively; and on the ISIC-Archive dataset, the IoU and Dice of the DeepLabV3+ were also improved respectively. Furthermore, even when the training data is reduced to 70 %, the proposed method can still achieve performance that is 95 % of that achieved with the entire dataset, indicating its good data efficiency. Moreover, the data-efficient features used in the method have interpretable information built-in, which enhances the interpretability of the model. The method has excellent universality, is plug-and-play, applicable to various segmentation methods, and does not require modification of the network structure, thus it is easy to integrate into existing medical image segmentation method, enhancing the convenience of future research and applications.

Medical image registration plays an important role in medical diagnosis and treatment planning. However, the current registration methods based on deep learning still face some challenges, such as insufficient ability to extract global information, large number of network model parameters, slow reasoning speed and so on. Therefore, this paper proposed a new model LCU-Net, which used parallel lightweight convolution to improve the ability of global information extraction. The problem of large number of network parameters and slow inference speed was solved by multi-scale fusion. The experimental results showed that the Dice coefficient of LCU-Net reached 0.823, the Hausdorff distance was 1.258, and the number of network parameters was reduced by about one quarter compared with that before multi-scale fusion. The proposed algorithm shows remarkable advantages in medical image registration tasks, and it not only surpasses the existing comparison algorithms in performance, but also has excellent generalization performance and wide application prospects.

Aiming at the problems of missing important features, inconspicuous details and unclear textures in the fusion of multimodal medical images, this paper proposes a method of computed tomography (CT) image and magnetic resonance imaging (MRI) image fusion using generative adversarial network (GAN) and convolutional neural network (CNN) under image enhancement. The generator aimed at high-frequency feature images and used double discriminators to target the fusion images after inverse transform; Then high-frequency feature images were fused by trained GAN model, and low-frequency feature images were fused by CNN pre-training model based on transfer learning. Experimental results showed that, compared with the current advanced fusion algorithm, the proposed method had more abundant texture details and clearer contour edge information in subjective representation. In the evaluation of objective indicators, Q AB/F , information entropy (IE), spatial frequency (SF), structural similarity (SSIM), mutual information (MI) and visual information fidelity for fusion (VIFF) were 2.0%, 6.3%, 7.0%, 5.5%, 9.0% and 3.3% higher than the best test results, respectively. The fused image can be effectively applied to medical diagnosis to further improve the diagnostic efficiency.

When applying deep learning algorithms to magnetic resonance (MR) image segmentation, a large number of annotated images are required as data support. However, the specificity of MR images makes it difficult and costly to acquire large amounts of annotated image data. To reduce the dependence of MR image segmentation on a large amount of annotated data, this paper proposes a meta-learning U-shaped network (Meta-UNet) for few-shot MR image segmentation. Meta-UNet can use a small amount of annotated image data to complete the task of MR image segmentation and obtain good segmentation results. Meta-UNet improves U-Net by introducing dilated convolution, which can increase the receptive field of the model to improve the sensitivity to targets of different scales. We introduce the attention mechanism to improve the adaptability of the model to different scales. We introduce the meta-learning mechanism, and employ a composite loss function for well-supervised and effective bootstrapping of model training. We use the proposed Meta-UNet model to train on different segmentation tasks, and then use the trained model to evaluate on a new segmentation task, where the Meta-UNet model achieves high-precision segmentation of target images. Meta-UNet has a certain improvement in mean Dice similarity coefficient (DSC) compared with voxel morph network (VoxelMorph), data augmentation using learned transformations (DataAug) and label transfer network (LT-Net). Experiments show that the proposed method can effectively perform MR image segmentation using a small number of samples. It provides a reliable aid for clinical diagnosis and treatment.

Medical image segmentation based on deep learning has become a powerful tool in the field of medical image processing. Due to the special nature of medical images, image segmentation algorithms based on deep learning face problems such as sample imbalance, edge blur, false positive, false negative, etc. In view of these problems, researchers mostly improve the network structure, but rarely improve from the unstructured aspect. The loss function is an important part of the segmentation method based on deep learning. The improvement of the loss function can improve the segmentation effect of the network from the root, and the loss function is independent of the network structure, which can be used in various network models and segmentation tasks in plug and play. Starting from the difficulties in medical image segmentation, this paper first introduces the loss function and improvement strategies to solve the problems of sample imbalance, edge blur, false positive and false negative. Then the difficulties encountered in the improvement of the current loss function are analyzed. Finally, the future research directions are prospected. This paper provides a reference for the reasonable selection, improvement or innovation of loss function, and guides the direction for the follow-up research of loss function.

In recent years, the task of object detection and segmentation in medical image is the research hotspot and difficulty in the field of image processing. Instance segmentation provides instance-level labels for different objects belonging to the same class, so it is widely used in the field of medical image processing. In this paper, medical image instance segmentation was summarized from the following aspects: First, the basic principle of instance segmentation was described, the instance segmentation models were classified into three categories, the development context of the instance segmentation algorithm was displayed in two-dimensional space, and six classic model diagrams of instance segmentation were given. Second, from the perspective of the three models of two-stage instance segmentation, single-stage instance segmentation and three-dimensional (3D) instance segmentation, we summarized the ideas of the three types of models, discussed the advantages and disadvantages, and sorted out the latest developments. Third, the application status of instance segmentation in six medical images such as colon tissue image, cervical image, bone imaging image, pathological section image of gastric cancer, computed tomography (CT) image of lung nodule and X-ray image of breast was summarized. Fourth, the main challenges in the field of medical image instance segmentation were discussed and the future development direction was prospected. In this paper, the principle, models and characteristics of instance segmentation are systematically summarized, as well as the application of instance segmentation in the field of medical image processing, which is of positive guiding significance to the study of instance segmentation.

Intelligent medical image segmentation methods have been rapidly developed and applied, while a significant challenge is domain shift. That is, the segmentation performance degrades due to distribution differences between the source domain and the target domain. This paper proposed an unsupervised end-to-end domain adaptation medical image segmentation method based on the generative adversarial network (GAN). A network training and adjustment model was designed, including segmentation and discriminant networks. In the segmentation network, the residual module was used as the basic module to increase feature reusability and reduce model optimization difficulty. Further, it learned cross-domain features at the image feature level with the help of the discriminant network and a combination of segmentation loss with adversarial loss. The discriminant network took the convolutional neural network and used the labels from the source domain, to distinguish whether the segmentation result of the generated network is from the source domain or the target domain. The whole training process was unsupervised. The proposed method was tested with experiments on a public dataset of knee magnetic resonance (MR) images and the clinical dataset from our cooperative hospital. With our method, the mean Dice similarity coefficient (DSC) of segmentation results increased by 2.52% and 6.10% to the classical feature level and image level domain adaptive method. The proposed method effectively improves the domain adaptive ability of the segmentation method, significantly improves the segmentation accuracy of the tibia and femur, and can better solve the domain transfer problem in MR image segmentation.

High resolution (HR) magnetic resonance images (MRI) or computed tomography (CT) images can provide clearer anatomical details of human body, which facilitates early diagnosis of the diseases. However, due to the imaging system, imaging environment and human factors, it is difficult to obtain clear high-resolution images. In this paper, we proposed a novel medical image super resolution (SR) reconstruction method via multi-scale information distillation (MSID) network in the non-subsampled shearlet transform (NSST) domain, namely NSST-MSID network. We first proposed a MSID network that mainly consisted of a series of stacked MSID blocks to fully exploit features from images and effectively restore the low resolution (LR) images to HR images. In addition, most previous methods predict the HR images in the spatial domain, producing over-smoothed outputs while losing texture details. Thus, we viewed the medical image SR task as the prediction of NSST coefficients, which make further MSID network keep richer structure details than that in spatial domain. Finally, the experimental results on our constructed medical image datasets demonstrated that the proposed method was capable of obtaining better peak signal to noise ratio (PSNR), structural similarity (SSIM) and root mean square error (RMSE) values and keeping global topological structure and local texture detail better than other outstanding methods, which achieves good medical image reconstruction effect.

[ Abstract ]Automatic and accurate segmentation of lung parenchyma is essential for assisted diagnosis of lung cancer. In recent years, researchers in the field of deep learning have proposed a number of improved lung parenchyma segmentation methods based on U-Net. However, the existing segmentation methods ignore the complementary fusion of semantic information in the feature map between different layers and fail to distinguish the importance of different spaces and channels in the feature map. To solve this problem, this paper proposes the double scale parallel attention (DSPA) network (DSPA-Net) architecture, and introduces the DSPA module and the atrous spatial pyramid pooling (ASPP) module in the "encoder-decoder" structure. Among them, the DSPA module aggregates the semantic information of feature maps of different levels while obtaining accurate space and channel information of feature map with the help of cooperative attention (CA). The ASPP module uses multiple parallel convolution kernels with different void rates to obtain feature maps containing multi-scale information under different receptive fields. The two modules address multi-scale information processing in feature maps of different levels and in feature maps of the same level, respectively. We conducted experimental verification on the Kaggle competition dataset. The experimental results prove that the network architecture has obvious advantages compared with the current mainstream segmentation network. The values of dice similarity coefficient (DSC) and intersection on union (IoU) reached 0.972 ± 0.002 and 0.945 ± 0.004, respectively. This paper achieves automatic and accurate segmentation of lung parenchyma and provides a reference for the application of attentional mechanisms and multi-scale information in the field of lung parenchyma segmentation.

Remarkable results have been realized by the U-Net network in the task of medical image segmentation. In recent years, many scholars have been researching the network and expanding its structure, such as improvement of encoder and decoder and improvement of skip connection. Based on the optimization of U-Net structure and its medical image segmentation techniques, this paper elucidates in the following: First, the paper elaborates on the application of U-Net in the field of medical image segmentation; Then, the paper summarizes the seven improvement mechanism of U-Net: dense connection mechanism, residual connection mechanism, multi-scale mechanism, ensemble mechanism, dilated mechanism, attention mechanism, and transformer mechanism; Finally, the paper states the ideas and methods on the U-Net structure improvement in a bid to provide a reference for later researches, which plays a significant part in advancing U-Net.

In order to suppress the geometrical artifacts caused by random jitter in ray source scanning, and to achieve flexible ray source scanning trajectory and meet the requirements of task-driven scanning imaging, a method of free trajectory cone-beam computed tomography (CBCT) reconstruction is proposed in this paper. This method proposed a geometric calibration method of two-dimensional plane. Based on this method, the geometric calibration phantom and the imaging object could be simultaneously imaged. Then, the geometric parameters could be obtained by online calibration method, and then combined with the geometric parameters, the alternating direction multiplier method (ADMM) was used for image iterative reconstruction. Experimental results showed that this method obtained high quality reconstruction image with high contrast and clear feature edge. The root mean square errors (RMSE) of the simulation results were rather small, and the structural similarity (SSIM) values were all above 0.99. The experimental results showed that it had lower image information entropy (IE) and higher contrast noise ratio (CNR). This method provides some practical value for CBCT to realize trajectory freedom and obtain high quality reconstructed image.

The three-dimensional (3D) liver and tumor segmentation of liver computed tomography (CT) has very important clinical value for assisting doctors in diagnosis and prognosis. This paper proposes a tumor 3D conditional generation confrontation segmentation network (T3scGAN) based on conditional generation confrontation network (cGAN), and at the same time, a coarse-to-fine 3D automatic segmentation framework is used to accurately segment liver and tumor area. This paper uses 130 cases in the 2017 Liver and Tumor Segmentation Challenge (LiTS) public data set to train, verify and test the T3scGAN model. Finally, the average Dice coefficients of the validation set and test set segmented in the 3D liver regions were 0.963 and 0.961, respectively, while the average Dice coefficients of the validation set and test set segmented in the 3D tumor regions were 0.819 and 0.796, respectively. Experimental results show that the proposed T3scGAN model can effectively segment the 3D liver and its tumor regions, so it can better assist doctors in the accurate diagnosis and treatment of liver cancer.

The construction of brain functional network based on resting-state functional magnetic resonance imaging (fMRI) is an effective method to reveal the mechanism of human brain operation, but the common brain functional network generally contains a lot of noise, which leads to wrong analysis results. In this paper, the least absolute shrinkage and selection operator (LASSO) model in compressed sensing is used to reconstruct the brain functional network. This model uses the sparsity of L 1 -norm penalty term to avoid over fitting problem. Then, it is solved by the fast iterative shrinkage-thresholding algorithm (FISTA), which updates the variables through a shrinkage threshold operation in each iteration to converge to the global optimal solution. The experimental results show that compared with other methods, this method can improve the accuracy of noise reduction and reconstruction of brain functional network to more than 98%, effectively suppress the noise, and help to better explore the function of human brain in noisy environment.

With the rapid development of network structure, convolutional neural networks (CNN) consolidated its position as a leading machine learning tool in the field of image analysis. Therefore, semantic segmentation based on CNN has also become a key high-level task in medical image understanding. This paper reviews the research progress on CNN-based semantic segmentation in the field of medical image. A variety of classical semantic segmentation methods are reviewed, whose contributions and significance are highlighted. On this basis, their applications in the segmentation of some major physiological and pathological anatomical structures are further summarized and discussed. Finally, the open challenges and potential development direction of semantic segmentation based on CNN in the area of medical image are discussed.

With the change of medical diagnosis and treatment mode, the quality of medical image directly affects the diagnosis and treatment of the disease for doctors. Therefore, realization of intelligent image quality control by computer will have a greater auxiliary effect on the radiographer's filming work. In this paper, the research methods and applications of image segmentation model and image classification model in the field of deep learning and traditional image processing algorithm applied to medical image quality evaluation are described. The results demonstrate that deep learning algorithm is more accurate and efficient than the traditional image processing algorithm in the effective training of medical image big data, which explains the broad application prospect of deep learning in the medical field. This paper developed a set of intelligent quality control system for auxiliary filming, and successfully applied it to the Radiology Department of West China Hospital and other city and county hospitals, which effectively verified the feasibility and stability of the quality control system.

To assess the background field removal method usually used in quantitative susceptibility mapping (QSM), and to analyze the cause of serious artifacts generated in the truncated k -space division (TKD) method, this paper discusses a variety of background field removal methods and proposes an improved method to suppress the artifacts of susceptibility inversion. Firstly, we scanned phase images with the gradient echo sequence and then compared the quality and the speed of reconstructed images of sophisticated harmonic artifact reduction for phase data (SHARP), regularization enable of SHARP (RESHARP) and laplacian boundary value (LBV) methods. Secondly, we analyzed the reasons for reconstruction artifacts caused by the multiple truncations and discontinuity of the TKD method, and an improved TKD method was proposed by increasing threshold truncation range and improving data continuity. Finally, the result of susceptibility inversion from the improved and original TKD method was compared. The results show that the reconstruction of SHARP and RESHARP are very fast, but SHARP reconstruction artifacts are serious and the reconstruction precision is not high and implementation of RESHARP is complicated. The reconstruction speed of LBV method is slow, but the detail of the reconstructed image is prominent and the precision is high. In the QSM inversion methods, the reconstruction artifact of the original TKD method is serious, while the improved method obtains good artifact suppression image and good inversion result of artifact regions.

Acoustic properties of biological tissues usually vary inhomogeneously in space. Tissues with different chemical composition often have different acoustic properties. The assumption of acoustic homogeneity may lead to blurred details, misalignment of targets and artifacts in the reconstructed photoacoustic tomography (PAT) images. This paper summarizes the main solutions to PAT imaging of acoustically heterogeneous tissues, including the variable sound speed and acoustic attenuation. The advantages and limits of the methods are discussed and the possible future development is prospected.

To locate the nuclei in hematoxylin-eosin (HE) stained section images more simply, efficiently and accurately, a new method based on distance estimation is proposed in this paper, which shows a new mind on locating the nuclei from a clump image. Different from the mainstream methods, proposed method avoids the operations of searching the combined singles. It can directly locate the nuclei in a full image. Furthermore, when the distance estimation built on the matrix sequence of distance rough estimating (MSDRE) is combined with the fact that a center of a convex region must have the farthest distance to the boundary, it can fix the positions of nuclei quickly and precisely. In addition, a high accuracy and efficiency are achieved by this method in experiments, with the precision of 95.26% and efficiency of 1.54 second per thousand nuclei, which are better than the mainstream methods in recognizing nucleus clump samples. Proposed method increases the efficiency of nuclear location while maintaining the location's accuracy. This can be helpful for the automatic analysis system of HE images by improving the real-time performance and promoting the application of related researches.

Dual-energy computed tomography (CT) reconstruction imaging technology is an important development direction in the field of CT imaging. The mainstream model of dual-energy CT reconstruction algorithm is the basis material decomposition model, and the projection decomposition is the crucial technique. The projection decomposition algorithm based on projection matching was a general method. With establishing the energy spectrum lookup table, we can obtain the stable solution by the least squares matching method. But the computation cost will increase dramatically when size of lookup table enlarges and it will slow down the computer. In this paper, an acceleration algorithm based on projection matching is proposed. The proposed algorithm makes use of linear equations and plane equations to fit the lookup table data, so that the projection value of the decomposition coefficients can be calculated quickly. As the result of simulation experiment, the acceleration algorithm can greatly shorten the running time of the program to get the stable and correct solution.

In order to get high-resolution glomerulus image with large field of view(FOV),stitching multiple small FOV images with high-resolution is necessary.Directly stitching images without properly correction is not acceptable and cannot afford any significant assistance in pathological diagnosis for intensity inhomogeneity and geometric distortion.Therefore we proposed a method of distortion correction and intensity inhomogeneity correction of glomerulus transmission electron microscope(TEM)image.In this paper,we firstly describe how these two distortions degrade images.Secondly,based on the TEM imaging system,image acquisition model and distortion correction model were proposed.Then according to these two models,distortions were greatly degraded and stitching results were improved by respectively applying two corrections,intensity inhomogeneity correction and geometric distortion correction.With the method proposed here,the result was improved significantly and stripes,fuzzy and artifacts were decreased dramatically.Our method has been proved to be valid to solve the problems of TEM glomerulus image distortion and at the same time to improve the result of multiple TEM glomerulus image stitching.

For tooth segmentation problem on the three-dimensional computed tomography(CT)volume data,this paper proposes a regional adaptive deformable model for tooth structure measurement of CT images.The proposed method combines the automatic thresholding segmentation,CV active contour model,and graph-cut.Firstly,we achieved the segmentation and location of dental crowns by automatic thresholding segmentation.And then by using the above segmentation result as the initial contour,we utilized active contour method to slice gradually the segment of remaining tooth.By incorporating active contour and graph-cut then,we realized the accurate segmentation for tooth root,which is the most difficult to be segmented.The experimental results showed that the proposed tooth structure measurement accurately and automatically segmented dental crowns from CT data,and then rapidly and accurately segmented the tooth neck and tooth root.The structure of tooth could be effectively segmented from CT data by using the proposed method.Experimental results indicated that the proposed method was rather robust and accurate,and could effectively assist the doctor for diagnosis in clinical treatment.

This paper presents a kind of automatic segmentation method for white blood cell based on HSI corrected space information fusion.Firstly,the original cell image is transformed to HSI colour space conversion.Because the transformation formulas of H component piecewise function was discontinuous,the uniformity of uniform visual cytoplasm area in the original image was lead to become lower in this channel.We then modified formulas,and then fetched information of nucleus,cytoplasm,red blood cells and background region according to distribution characteristics of the H,S and I-channel,using the theory and method of information fusion to build fusion imageⅠand fusion imageⅡ,which only contained cytoplasm and a small amount of interference,and fetched nucleus and cytoplasm respectively.Finally,we marked the nucleus and cytoplasm region and obtained the final result of segmentation.The simulation results showed that the new algorithm of image segmentation for white blood cell had high accuracy,robustness and universality.

Considering the survival rate of small animals and the continuity of the experiments,high-dose X-ray shooting process is not suitable for the small animals in computed tomography(CT)experiments.But the low-dose process results with images might be polluted by noises which are not conducive for the experiments.In order to solve this problem,we in this paper introduce a global dictionary learning based denoising method to apply the promotion of the low dose CT image.We at first adopted the K-means singular value decomposition(K-SVD)algorithm to train a global dictionary based on the high dose CT image.Then,the noise image could be decomposed into sparse component which was free from noise through the orthogonal matching pursuit(OMP)algorithm.Finally,the noisefree image could be achieved by reconstructing the image only with its sparse components.The experiments results showed that the method we proposed here could decrease the noise efficiently and remain the details,and it would help promote the low dose image quality and increase the survival rate of the small animals.

Based on repeated experiments as well as continuous researching and improving, an efficient scheme to measure velocity and displacement of the coxa and knee movements based on video image processing technique is presented in this paper. The scheme performed precise and real-time quantitative measurements of 2D velocity or displacement of the coxa and knee using a video camera mounted on one side of the healing and training beds. The beds were based on simplified pinhole projection model. In addition, we used a special-designed auxiliary calibration target, composed by 24 circle points uniformly located on two concentric circles and two straight rods which can rotate freely along the concentric center within the vertical plane, to do the measurements. Experiments carried out in our laboratory showed that the proposed scheme could basically satisfy the requirements about precision and processing speed of such kind of system, and would be very suitable to be applied to smart evaluation/training and healing system for muscles/balance function disability as an advanced and intuitional helping method.

This paper is to report our study in which the differences between prosthetic restoration and surgical reconstruction using traditional clasp retention technology were analyzed based on three-dimensional finite element methods in our laboratory. Firstly, the maxillary unilateral defect model was developed using medical image processing software MIMICS. Secondly, the prosthesis was generated by mirroring technology. The clasp was designed according to the methods raised by Aramany. Then, the stress distribution of maxilla was calculated by simulating occlusion. According to the results, after osseointegration of surgical reconstruction, stresses of unaffected abutments were reduced significantly, and less stress of junction occurred near zygoma of affected side, which were all less than stresses of prosthesis restoration. Thus, removing the clasp of surgical reconstruction increased the stresses of unaffected abutments. The stress trends of maxillary components were different between prosthetic restoration and surgical reconstruction. Surgical reconstruction is better than prosthesis restoration in protection of the abutments. Clasp can alleviate the occlusal burden of maxilla. Varieties of retentive technologies can be considered in prosthesis restoration. The surgical reconstruction is more conducive to rehabilitate unilateral maxilla biomechanically in clinic.

Aiming at motion artifacts and off-resonance artifacts in multi-shot diffusion magnetic resonance imaging (MRI), we proposed a joint correction method in this paper to correct the two kinds of artifacts simultaneously without additional acquisition of navigation data and field map. We utilized the proposed method using multi-shot variable density spiral sequence to acquire MRI data and used auto-focusing technique for image deblurring. We also used direct method or iterative method to correct motion induced phase errors in the process of deblurring. In vivo MRI experiments demonstrated that the proposed method could effectively suppress motion artifacts and off-resonance artifacts and achieve images with fine structures. In addition, the scan time was not increased in applying the proposed method.

It is difficult to distinguish the inferior alveolar nerve (IAN) from other tissues inside the IAN canal due to their similar CT values in the X image which are smaller than that of the bones. The direct reconstruction, therefore, is difficult to achieve the effects. The traditional clinical treatments mainly rely on doctors' manually drawing the X images so that some subjective results could not be avoided. This paper proposes the partition reconstruction of IAN canal based on shape features. According to the anatomical features of the IAN canal, we divided the image into three parts and treated the three parts differently. For the first, the directly part of the mandibular, we used Shape-driven Level-set Algorithm Restrained by Local Information (BSLARLI) segment IAN canal. For the second part, the mandibular body, we used Space B-spline curve fitting IAN canal's center, then along the center curve established the cross section. And for the third part, the mental foramen, we used an adaptive threshold Canny algorithm to extract IAN canal's edge to find center curve, and then along it established the cross section similarly. Finally we used the Visualization Toolkit (VTK) to reconstruct the CT data as mentioned above. The VTK reconstruction result by setting a different opacity and color values of tissues CT data can perspectively display the INA canal clearly. The reconstruction result by using this method is smoother than that using the segmentation results and the anatomical structure of mental foramen position is similar to the real tissues, so it provides an effective method for locating the spatial position of the IAN canal for implant surgeries.

Image interpolation is often required during medical image processing and analysis. Although interpolation method based on Gaussian radial basis function (GRBF) has high precision, the long calculation time still limits its application in field of image interpolation. To overcome this problem, a method of two-dimensional and three-dimensional medical image GRBF interpolation based on computing unified device architecture (CUDA) is proposed in this paper. According to single instruction multiple threads (SIMT) executive model of CUDA, various optimizing measures such as coalesced access and shared memory are adopted in this study. To eliminate the edge distortion of image interpolation, natural suture algorithm is utilized in overlapping regions while adopting data space strategy of separating 2D images into blocks or dividing 3D images into sub-volumes. Keeping a high interpolation precision, the 2D and 3D medical image GRBF interpolation achieved great acceleration in each basic computing step. The experiments showed that the operative efficiency of image GRBF interpolation based on CUDA platform was obviously improved compared with CPU calculation. The present method is of a considerable reference value in the application field of image interpolation.

We searched and retrieved literature on the topic of medical image processing published on SCI journals in the past 10 years. We then imported the retrieved literature into TDA for data cleanup before data analysis and pro cessing by EXCLE and UCINET to generate tables and figures that could indicate disciplinary correlation and research hotspots from the perspective of bibliometrics. The results indicated that people in Europe and USA were leading researchers on medical image processing with close international cooperation. Many disciplines contributed to the fast development of medical image processing with intense interdisciplinary researches. The papers that we found show recent research hotspots of the algorithm, system, model, image and segmentation in the field of medical image processing. Cluster analysis on key words of high frequency demonstrated complicated clustering relationship.

In this paper, we present a 3D virtual phantom design software, which was developed based on object-oriented programming methodology and dedicated to medical physics research. This software was named Magical Phan tom (MPhantom), which is composed of 3D visual builder module and virtual CT scanner. The users can conveniently construct any complex 3D phantom, and then export the phantom as DICOM 3.0 CT images. MPhantom is a user-friendly and powerful software for 3D phantom configuration, and has passed the real scene's application test. MPhantom will accelerate the Monte Carlo simulation for dose calculation in radiation therapy and X ray imaging reconstruction algorithm research.

Medical image fusion and recognition has a wide range of applications, such as focal location, cancer staging and treatment effect assessment. Multi-model medical image fusion and recognition are analyzed and summarized in this paper. Firstly, the question of multi-model medical image fusion and recognition is discussed, and its advantage and key steps are discussed. Secondly, three fusion strategies are reviewed from the point of algorithm, and four fusion recognition structures are discussed. Thirdly, difficulties, challenges and possible future research direction are discussed.

Due to the difficulties of the copying, vein identification has developed rapidly in recent years. The light source selection directly affects the image quality. This paper acquired by experiment the reflectivities of vein and non-vein irradiation with different wavelengths of near infra-red. Comparing the strength of reflectivities of various wave lengths, we found that there were the strongest contrasts between vein and non-vein in the 810 nm, and 810 nm near infra-red was suitable to a vein imaging light source. Finally, clear hand vein images were obtained with the selected light source.

In order to detect the overlapped and clustered protein spots in protein gel images, we suggest an auto-separating algorithm based on the shape characteristics in this paper. Firstly, we used a marker-controlled watershed algorithm to detect the initial outline of the protein spots in application of the method. Secondly, we extracted shape markers adaptively by combining local minima depth and shape characteristics information, and calculated the shape distance image according to these shape markers. Then, we imposed these shape markers on the shape distance image as its local minima. Finally, we applied the watershed algorithm again on the modified shape distance image to separate the overlapped spots accurately. The experiments on the synthetic gel image and the different types of real gel images showed that compared with other improved watershed algorithm, the proposed one had a higher correct separation rate with an error less than 6%, and better separation results of overlapped spots.

Because of various effects of the imaging mechanism, noises are inevitably introduced in medical CT imaging process. Noises in the images will greatly degrade the quality of images and bring difficulties to clinical diagnosis. This paper presents a new method to improve singular value decomposition (SVD) filtering performance in CT image. Filter based on SVD can effectively analyze characteristics of the image in horizontal (and/or vertical) directions. According to the features of CT image, we can make use of discrete cosine transform (DCT) to extract the region of interest and to shield uninterested region so as to realize the extraction of structure characteristics of the image. Then we transformed SVD to the image after DCT, constructing weighting function for image reconstruction adaptively weighted. The algorithm for the novel denoising approach in this paper was applied in CT image denoising, and the experimental results showed that the new method could effectively improve the performance of SVD filtering.

Xinjiang local liver hydatid disease is an infectious parasitic disease in Xinjiang pastoral areas. Based on the image features, selecting the appropriate distance algorithms to retrieve the image quickly and accurately, different distance algorithms have been induced in this area, which can greatly assist the doctors to early detect, diagnose and cure the liver hydatid disease. This paper compared the performance of different distance algorithms to retrieve the image when using the liver hydatid disease medical image texture features. The results showed that: for the liver hydatid disease medical images retrieval based on gray level cocurrence matrix (GLCM) texture features, the Mahalanobis distance algorithm is superior to other distance algorithms.

This paper is aimed to fulfill a prototype system used to classify and retrieve retinal image automatically. With the content-based image retrieval (CBIR) technology, a method to represent the retinal characteristics mixing the fundus image color (gray) histogram with bright, dark region features and other local comprehensive information was proposed. The method uses kernel principal component analysis (KPCA) to further extract nonlinear features and dimensionality reduced. It also puts forward a measurement method using support vector machine (SVM) on KPCA weighted distance in similarity measure aspect. Testing 300 samples with this prototype system randomly, we obtained the total image number of wrong retrieved 32, and the retrieval rate 89.33%. It showed that the identification rate of the system for retinal image was high.

Breast mass is women's common disease, and it seriously affects the health of women. Accurate detection and location of the mass will greatly improve the precision of the diagnosis of the breast disease. Studies have shown that the organizational structure and surface roughness of the mass constitute the texture features of mass and it is an important basis for distinguishing the mass. In this paper, a multi-level fractal features extraction method is proposed, and, moreover, the fractal feature vector of the suspicious lesions has been established, thus realizing the feature extraction and analysis of the mammary gland. Finally, the result of classification with the support vector machine (SVM) was given. 110 mammograms are extracted and classified with an accuracy of 90%. The results showed that the multi-level fractal features extraction and classification methods proposed in this paper could improve the accuracy of mass detection and could be effective for the early diagnosis of the disease.

Magnetic induction tomography (MIT) image reconstruction is a typical ill-posed problem, and its numerical solution is unstable. A new image reconstruction algorithm based on the L(P)-norm, which solves the ill-posed inverse problem of MIT and improves the quality of reconstructed image, is presented in this paper. The new algorithm not only overcomes the problem of numerical instability of the MIT image reconstruction, but also improves the quality of the reconstructed image and enhances the spatial resolution of the reconstructed image. Simulation results showed that the quality of the reconstructed image obtained using the presented algorithm was better than that using Tikhonov regularization algorithm and that using the variation regularization algorithm, so it could be an effective method for the MIT.

In this paper, we present a new method which can reconstruct the three-dimensional model of nasolacrimal duct. We firstly resampled the volume data along nasolacrimal duct direction, then segmented the nasolacrimal duct into slices, and finally, completed the 3D reconstruction from the two-D contours. Using this method, we can not only reconstruct normal nasolacrimal duct, but also reconstruct catagmatic nasolacrimal duct. It overcomes the current shortcomings of some traditional methods. Consequently, the technology proposed is of great significance in computer aided technique and surgical planning related to ophthalmonogy.

Analysis of neural stem cells' movements is one of the important parts in the fields of cellular and biological research. The main difficulty existing in cells' movement study is whether the cells tracking system can simultaneously track and analyze thousands of neural stem cells (NSCs) automatically. We present a novel cells' tracking algorithm which is based on segmentation and data association in this paper, aiming to improve the tracking accuracy further in high density NSCs' image. Firstly, we adopted different methods of segmentation base on the characteristics of the two cell image sequences in our experiment. Then we formed a data association and constituted a coefficient matrix by all cells between two adjacent frames according to topological constraints. Finally we applied The Hungarian algorithm to implement inter-cells matching optimally. Cells' tracking can be achieved according to this model from the second frame to the last one in a sequence. Experimental results showed that this approaching method has higher accuracy compared with that using the topological constraints tracking alone. The final tracking accuracies of average of sequence I and sequence II have been improved 10.17% and 4%, respectively.

Scoliosis, the abnormal lateral curvature of the spine, is an idiopathic disease often suffered by teenagers. Normally medical doctors use X-rays to measure the Cobb angle, and then assess the severity of scoliosis with it. In this paper, we point out the superiorities of image processing technique through analyzing the existing methods in the diagnosis of scoliosis. Two kinds of image processing technique are mainly introduced for scoliosis detection. The moiré stripe images show an asymmetric deformation pattern between the left-hand side and the right-hand side of human back, and these associate with Cobb angle to detect scoliosis. In order to check scoliosis through accurate three-dimensional surface features of human back, we in the second technique use different optical imaging methods to reconstruct the three-dimensional surface model of the patient's back, and extract the characteristics of the reconstructive human back to diagnose the disease. Both approaches tried to describe symmetry discrimination of human back and correlated it with the Cobb angles. Finally, we look forward to the future development in the application of image processing technique for scoliosis detection.

Noises are inevitably introduced to medical CT images because of various factors in the medical image processes. Noises in medical images will greatly degrade the quality of images and bring difficulties to clinical diagnosis. Therefore, it is necessary for medical CT images to be denoised. In the present paper, the sparse decomposition of images is used to denoise medical CT images. A new block-based sparse decomposition of images denoising method is proposed. Experimental results showed that block-based sparse decomposition of images denoising had a certain function. With the same number of atoms, the computing speed of the image constructed by the new algorithm was improved by about 15 times, compared with the whole image sparse decomposition denoising.

At present, ray-casting algorithm is the most widely used algorithm in the field of medical image visualization, and it can achieve the best image quality. Due to large amounts of computation like sampling, gradient, lighting and blending calculation, the cost of ray-casting algorithm is very large. The characteristic of Streaming single instruction multiple datum extensions (SSE) instruction--supporting vector computation--can satisfy the property of ray-casting algorithm well. Therefore, in this paper, we improved the implementation efficient significantly by vectorization of gradient, lighting and blending calculation, and still achieved a high quality image at the same time.

Medical image registration is a key step of image processing and analysis and the first problem of medical image fusion to be solved. In this paper, we focus on magnetic resonance images registration for preoperative and postoperative of deep brain stimulation for the Parkinson's disease. First of all, mutual distance is brought into the mutual information in this paper, realize the two groups of MR slices (for both preoperative and postoperative) mapping, and then 3D images are reconstructed from the corresponding MR slices. Finally, we use Powell algorithm to register 3D images. After the 3D images is registered, we can analyze the relative position between planted electrode (postoperative) and the subthalamic nucleus (preoperative), and measure the quality of deep brain stimulation scientifically.

Quantitative analysis of biological image data generally involves the detection of many pixel spots. In live mitochondria video image, for which fluorescent microscopy is often used, the signal-to-noise ratio (SNR) can be extremely low, making the detection and tracking of mitochondria particle difficult. It is especially not easy to get the movement curve when the movement of the mitochondria involves its self-move and the motion caused by the neuron. An tracking algorithm for live mitochondria is proposed in this paper. First the whole image sequence is frame-to-frame registered, in which the edge corners are chosen to be the feature points. Then the mitochondria particles are tracked by frame-to-frame displacement vector. The algorithm proposed has been applied to the dynamic image sequence including neuron and mitochondria, saving time without manually picking up the feature points. It provides an new method and reference for medical image processing and biotechnological research.

In this paper, a new method was presented which can generate dental panoramic radiographs from the 3D CT sectional data. The dental panoramic radiograph was generated by casting ray into the 3D sectional data from a curved surface close to the dental arch. With this method, the relationship between the 3D CT sectional data and the dental panoramic radiographs was built, which helped to overcome the defects in the real X-ray panoramic radiographs, such as structure overlap and unselectable content for displaying. The technology is of certain significance in computer aided technique and surgical planning related to dentistry.

Affine motion is common during PROPELLER magnetic resonance imaging (MRI) in abdomen or other soft tissues. The current algorithm, up till now, for affine motion compensation is based on frequency domain, which compensates the motion in k space and then reconstruct the final image based on gridding method. But aliasing and some tiny artifacts may exist. This paper proposed a new algorithm for affine motion compensation based on image domain. Firstly, exact affine motion information was obtained through the image registration, secondly k space coordinate was corrected for compensating the k space strips sampling density, then the images obtained from inverse FFT was compensated using motion information, finally the final results were composited after rotation. The experimental results showed that the proposed method could more effectively suppress the motion artifacts compared to the current algorithm.