A new algorithm is proposed for completing the missing parts caused by the removal of foreground or background elements from an image of natural scenery in a visually plausible way. The major contributions of the prop...A new algorithm is proposed for completing the missing parts caused by the removal of foreground or background elements from an image of natural scenery in a visually plausible way. The major contributions of the proposed algorithm are： （1） for most natural images, there is a strong orientation of texture or color distribution. So a method is introduced to compute the main direction of the texture and complete the image by limiting the search to one direction to carry out image completion quite fast; （2） there exists a synthesis ordering for image completion. The searching order of the patches is defined to ensure the regions with more known information and the structures should be completed before filling in other regions; （3） to improve the visual effect of texture synthesis, an adaptive scheme is presented to determine the size of the template window for capturing the features of various scales. A number of examples are given to demonstrate the effectiveness of the proposed algorithm.展开更多
There are two difficult in the existing image restoration methods.One is that the method is difficult to repair the image with a large damaged,the other is the result of image completion is not good and the speed is s...There are two difficult in the existing image restoration methods.One is that the method is difficult to repair the image with a large damaged,the other is the result of image completion is not good and the speed is slow.With the development and application of deep learning,the image repair algorithm based on generative adversarial networks can repair images by simulating the distribution of data.In the process of image completion,the first step is trained the generator to simulate data distribution and generate samples.Then a large number of falsified images are quickly generated using the generative adversarial network and search for the code of the closest damaged image.Finally,the generator generates missing content by using this code.On this basis,this paper combines the semantic loss function and the perceptual loss function.Experimental result show that the method successfully predicts the information of large areas missing in the image,and realizes the photorealism,producing clearer and more consistent results than previous methods.展开更多
Attention mechanism combined with convolutional neural network(CNN) achieves promising performance for magnetic resonance imaging(MRI) image segmentation,however these methods only learn attention weights from single ...Attention mechanism combined with convolutional neural network(CNN) achieves promising performance for magnetic resonance imaging(MRI) image segmentation,however these methods only learn attention weights from single scale,resulting in incomplete attention learning.A novel method named completed attention convolutional neural network(CACNN) is proposed for MRI image segmentation.Specifically,the channel-wise attention block(CWAB) and the pixel-wise attention block(PWAB) are designed to learn attention weights from the aspects of channel and pixel levels.As a result,completed attention weights are obtained,which is beneficial to discriminative feature learning.The method is verified on two widely used datasets(HVSMR and MRBrainS),and the experimental results demonstrate that the proposed method achieves better results than the state-of-theart methods.展开更多
Background:Completely endophytic renal tumors(CERT)pose significant challenges due to their anatomical complexity and loss of visual clues about tumor location.A facile scoring model based on three-dimensional(3D)reco...Background:Completely endophytic renal tumors(CERT)pose significant challenges due to their anatomical complexity and loss of visual clues about tumor location.A facile scoring model based on three-dimensional(3D)reconstructed images will assist in better assessing tumor location and vascular variations.Methods:In this retrospective study,80 patients diagnosed with CERT were included.Forty cases underwent preoperative assessment using 3D reconstructed imaging(3D-Cohort),while the remaining 40 cases were assessed using two-dimensional imaging(2D-Cohort).Vascular variations were evaluated by ascertaining the presence of renal arteries>1,prehilar branching arteries,and arteries anterior to veins.The proposed scoring system,termed RAL,encompassed three critical components:(R)adius(maximal tumor diameter in cm),(A)rtery(occurrence of arterial variations),and(L)ocation relative to the polar line.Comparison of the RAL scoring system was made with established nephrometry scoring systems.Results:A total of 48(60%)patients exhibited at least one vascular variation.In the 2D-Cohort,patients with vascular variations experienced significantly prolonged operation time,increased bleeding volume,and extended warm ischemia time compared with those without vascular variations.Conversely,the presence of vascular vari-ations did not significantly affect operative parameters in the 3D-Cohort.Furthermore,the 2D-Cohort demon-strated a notable decline in both short-and long-term estimated glomerular filtration rate(eGFR)changes com-pared with the 3D-Cohort,a trend consistent across patients with warm ischemia time≥25 min and those with vascular variations.Notably,the 2D-Cohort exhibited a larger margin of normal renal tissue compared with the 3D-Cohort.Elevated RAL scores correlated with larger tumor size,prolonged operation time,extended warm is-chemia time,and substantial postoperative eGFR decrease.The RAL scoring system displayed superior predictive capabilities in assessing postoperative eGFR changes compared with conventional nephrometry scoring systems.Conclusions:Our proposed 3D vascular variation-based nephrometry scoring system offers heightened proficiency in preoperative assessment,precise prediction of surgical complexity,and more accurate evaluation of postoper-ative renal function in CERT patients.展开更多
To resolve the completeness and independence of an invariant set derived by the traditional method, a systematic method for deriving a complete set of pseudo-Zernike moment similarity （translation, scale and rotation...To resolve the completeness and independence of an invariant set derived by the traditional method, a systematic method for deriving a complete set of pseudo-Zernike moment similarity （translation, scale and rotation） invariants is described. First, the relationship between pseudo-Zernike moments of the original image and those of the image having the same shape but distinct orientation and scale is established. Based on this relationship, a complete set of similarity invariants can be expressed as a linear combination of the original pseudo-Zernike moments of the same order and lower order. The problem of image reconstruction from a finite set of the pseudo-Zernike moment invariants （PZMIs） is also investigated. Experimental results show that the proposed PZMIs have better performance than complex moment invariants.展开更多
文摘A new algorithm is proposed for completing the missing parts caused by the removal of foreground or background elements from an image of natural scenery in a visually plausible way. The major contributions of the proposed algorithm are： （1） for most natural images, there is a strong orientation of texture or color distribution. So a method is introduced to compute the main direction of the texture and complete the image by limiting the search to one direction to carry out image completion quite fast; （2） there exists a synthesis ordering for image completion. The searching order of the patches is defined to ensure the regions with more known information and the structures should be completed before filling in other regions; （3） to improve the visual effect of texture synthesis, an adaptive scheme is presented to determine the size of the template window for capturing the features of various scales. A number of examples are given to demonstrate the effectiveness of the proposed algorithm.
基金supported by Scientific Research Starting Project of SWPU(No.0202002131604)Major Science and Technology Project of Sichuan Province(No.8ZDZX0143)+1 种基金Ministry of Education Collaborative Education Project of China(No.952)Fundamental Research Project(Nos.549,550).
文摘There are two difficult in the existing image restoration methods.One is that the method is difficult to repair the image with a large damaged,the other is the result of image completion is not good and the speed is slow.With the development and application of deep learning,the image repair algorithm based on generative adversarial networks can repair images by simulating the distribution of data.In the process of image completion,the first step is trained the generator to simulate data distribution and generate samples.Then a large number of falsified images are quickly generated using the generative adversarial network and search for the code of the closest damaged image.Finally,the generator generates missing content by using this code.On this basis,this paper combines the semantic loss function and the perceptual loss function.Experimental result show that the method successfully predicts the information of large areas missing in the image,and realizes the photorealism,producing clearer and more consistent results than previous methods.
基金Supported National Natural Science Foundation of China (No.62171321)Tianjin Municipal Natural Science Foundation (No.20JCZDJC00180,19 JCZDJC31500)the Open Projects Program of National Laboratory of Pattern Recognition (No.202000002)。
文摘Attention mechanism combined with convolutional neural network(CNN) achieves promising performance for magnetic resonance imaging(MRI) image segmentation,however these methods only learn attention weights from single scale,resulting in incomplete attention learning.A novel method named completed attention convolutional neural network(CACNN) is proposed for MRI image segmentation.Specifically,the channel-wise attention block(CWAB) and the pixel-wise attention block(PWAB) are designed to learn attention weights from the aspects of channel and pixel levels.As a result,completed attention weights are obtained,which is beneficial to discriminative feature learning.The method is verified on two widely used datasets(HVSMR and MRBrainS),and the experimental results demonstrate that the proposed method achieves better results than the state-of-theart methods.
基金We thank researchers for patients enrolled from the FUSCC cohort.This work was supported by grants from the National Natural Science Foundation of China(grant numbers:81802525 and no.82172817)the Natural Science Foundation of Shanghai(grant number:20ZR1413100)+3 种基金Beijing Xisike Clinical Oncology Research Foundation(grant number:Y-HR2020MS-0948)the Shanghai“Science and Technology Innova-tion Action Plan”medical innovation research Project(grant num-ber:22Y11905100)the Shanghai Anti-Cancer Association Eyas Project(grant number:SACA-CY21A06 and no.SACA-CY21B01)Fudan University Fuqing scholars Project(grant number:FQXZ202304A).
文摘Background:Completely endophytic renal tumors(CERT)pose significant challenges due to their anatomical complexity and loss of visual clues about tumor location.A facile scoring model based on three-dimensional(3D)reconstructed images will assist in better assessing tumor location and vascular variations.Methods:In this retrospective study,80 patients diagnosed with CERT were included.Forty cases underwent preoperative assessment using 3D reconstructed imaging(3D-Cohort),while the remaining 40 cases were assessed using two-dimensional imaging(2D-Cohort).Vascular variations were evaluated by ascertaining the presence of renal arteries>1,prehilar branching arteries,and arteries anterior to veins.The proposed scoring system,termed RAL,encompassed three critical components:(R)adius(maximal tumor diameter in cm),(A)rtery(occurrence of arterial variations),and(L)ocation relative to the polar line.Comparison of the RAL scoring system was made with established nephrometry scoring systems.Results:A total of 48(60%)patients exhibited at least one vascular variation.In the 2D-Cohort,patients with vascular variations experienced significantly prolonged operation time,increased bleeding volume,and extended warm ischemia time compared with those without vascular variations.Conversely,the presence of vascular vari-ations did not significantly affect operative parameters in the 3D-Cohort.Furthermore,the 2D-Cohort demon-strated a notable decline in both short-and long-term estimated glomerular filtration rate(eGFR)changes com-pared with the 3D-Cohort,a trend consistent across patients with warm ischemia time≥25 min and those with vascular variations.Notably,the 2D-Cohort exhibited a larger margin of normal renal tissue compared with the 3D-Cohort.Elevated RAL scores correlated with larger tumor size,prolonged operation time,extended warm is-chemia time,and substantial postoperative eGFR decrease.The RAL scoring system displayed superior predictive capabilities in assessing postoperative eGFR changes compared with conventional nephrometry scoring systems.Conclusions:Our proposed 3D vascular variation-based nephrometry scoring system offers heightened proficiency in preoperative assessment,precise prediction of surgical complexity,and more accurate evaluation of postoper-ative renal function in CERT patients.
基金The National Natural Science Foundation of China(No.61071192,61073138)
文摘To resolve the completeness and independence of an invariant set derived by the traditional method, a systematic method for deriving a complete set of pseudo-Zernike moment similarity （translation, scale and rotation） invariants is described. First, the relationship between pseudo-Zernike moments of the original image and those of the image having the same shape but distinct orientation and scale is established. Based on this relationship, a complete set of similarity invariants can be expressed as a linear combination of the original pseudo-Zernike moments of the same order and lower order. The problem of image reconstruction from a finite set of the pseudo-Zernike moment invariants （PZMIs） is also investigated. Experimental results show that the proposed PZMIs have better performance than complex moment invariants.
