توجه: محتویات این صفحه به صورت خودکار پردازش شده و مقاله‌های نویسندگانی با تشابه اسمی، همگی در بخش یکسان نمایش داده می‌شوند.
۱Effect of Different Diffusion Maps on RegistrationResults
اطلاعات انتشار: هفتمین کنفرانس ماشین بینایی و پردازش تصویر، سال
تعداد صفحات: ۵
In this paper, we compare registration results obtainedusing different diffusion maps extracted from diffusion tensorimaging (DTI). Fractional Anisotropy (FA) and Ellipsoidal AreaRatio (EAR) are two diffusion maps (indices) that may be usedfor image registration. First, we use FA maps to find deformationmatrix and register diffusion weighted images. Then, we use EARmaps and finally we use both of FA and EAR maps to registerdiffusion weighted images. The difference between FA valuesbefore deformation and after registration using the FA alone orEAR alone has a median of 0.57 and using both of them has amedian of 0.29. Therefore, the results of registration using bothof the FA and EAR indices are superior to those obtained usingonly one of them alone.<\div>

۲MMRO: A Feature Selection Criterion for MRImages Based on Alpha Stable Filter Responses
نویسنده(ها): ،
اطلاعات انتشار: هفتمین کنفرانس ماشین بینایی و پردازش تصویر، سال
تعداد صفحات: ۵
In feature–based image registration, feature selectionand reduction methods play an important role in decreasingcomputational burden of these operations. In this paper, a newapproach is introduced to reduce the dimension of extractedfeature vectors of MR images. This approach is based on theselection of the maximum and minimum responses of the alphastable filter for the MR images over the extracted features withdifferent orientation in frequency domain. This algorithm selectsthe rotation invariant features which are suitable for imageregistration purposes. It has been shown that these features couldefficiently describe the image elements. The discriminating abilityof the features selected by the proposed method is compared withMRO method and an average improvement of 175 % wasobtained based on the defined discriminating value<\div>

۳Content Based Mammogram Image Retrieval Based On The Multiclass Visual Problem
نویسنده(ها): ،
اطلاعات انتشار: هفدهمین کنفرانس مهندسی پزشکی ایران، سال
تعداد صفحات: ۴
Since expertise elicited from past resolved cases plays an important role in medical application and images acquired from various cases have a great contribution to diagnosis of the abnormalities, Content based medical image retrieval has become an active research area for many scientists, In this article we proposed a new framework to retrieve visually similar images from a large database, in which visual relevanceis regarded as much as the semantic category similarity, we used optimized wavelet transform as the multi–resolution analysis of the images and extracted various statistical SGLDM features from different resolutions then after reducing feature space we used error correcting codes in order to untwist the existing multiclass visual problem introduced in preceding parts of the article, we implemented proposed algorithm on the 1000 mammograms provided by the DDSM database which consist of 2500 studies and their annotations provided by specialists.<\div>

۴Gradient Vector Flow Snake Segmentation of Breast Lesions in Dynamic Contrast–Enhanced MR Images
نویسنده(ها): ، ،
اطلاعات انتشار: هفدهمین کنفرانس مهندسی پزشکی ایران، سال
تعداد صفحات: ۴
The development of computer–aided diagnosis (CAD) for breast magnetic resonance (MR) images has encountered some big challenges. One of these challenges is related to breast lesion segmentation. Accurate segmentation of breast lesions has a vital role in other consequent applications such as feature extraction. Since malignant breast lesions typically appear with irregular borders and shapes in MR images whereas benign masses appear with more regular shapes, and smooth and lobulated borders, it seems that the accurate segmentation ofbreast lesion borders in MR images are important. To achieve this purpose, we have used the Gradient Vector Flow (GVF) snake segmentation method. This study included 52(33 malignant and 19 benign) histopathologically proven breast lesions and the stages of the proposed method are as follows: selecting the region of interest (ROI), segmentation using GVF, evaluation of GVF snake segmentation method. The results of GVF segmentation method in this study were satisfactory referred to the radiologist’s manual segmentation. The results showed the GVF snake segmentation method correctly segmented 97% of malignant lesion borders and 89.5% of benign lesion borders at the overlap threshold of 0.6. This indicates GVF snake segmentation method could provide us with a powerful method that can make an accurate segmentation in breast lesion borders.<\div>

۵Nonrigid Registration of Breast MR Images Using Residual Complexity Similarity Measure
اطلاعات انتشار: هشتمین کنفرانس ماشین بینایی و پردازش تصویر، سال
تعداد صفحات: ۴
Elimination of motion artifact in breast MR imagesis a significant issue in pre–processing step before utilizing imagesfor diagnostic applications. Breast MR Images are affected by slowvarying intensity distortions as a result of contrast agentenhancement. Thus a nonrigid registration algorithm consideringthis effect is needed. Traditional similarity measures such as sumof squared differences and cross correlation, ignore the mentioneddistortion. Therefore, efficient registration is not obtained.Residual complexity is a similarity measures that considersspatially varying intensity distortions by maximizing sparseness ofthe residual image. In this research, the results obtained byapplying nonrigid registration based on residual complexity, sumof squared differences and cross correlation similarity measuresare demonstrated which shows more robustness and accuracy ofRC comparing with other similarity measures for breast MRimages.<\div>

۶Comparison of Classification and Dimensionality Reduction Methods Used in fMRI Decoding
نویسنده(ها): ،
اطلاعات انتشار: هشتمین کنفرانس ماشین بینایی و پردازش تصویر، سال
تعداد صفحات: ۵
In the last few years there has been growinginterest in the use of functional Magnetic Resonance Imaging(fMRI) for brain mapping. To decode brain patterns in fMRIdata, we need reliable and accurate classifiers. Towards this goal,we compared performance of eleven popular pattern recognitionmethods. Before performing pattern recognition, applying thedimensionality reduction methods can improve the classificationperformance; therefore, seven methods in region of interest(ROI) have been compared to answer the following question:which dimensionality reduction procedure performs best? Inboth tasks, in addition to measuring prediction accuracy, weestimated standard deviation of accuracies to realize morereliable methods. According to all results, we suggest usingsupport vector machines with linear kernel (C–SVM and ν–SVM),or random forest classifier on low dimensional subsets, which isprepared by Active or maxDis feature selection method toclassify brain activity patterns more efficiently.<\div>

۷Fuzzy Local Binary Patterns: A Comparison between Min–Max and Dot–Sum Operators in the Application of Facial Expression Recognition
نویسنده(ها): ،
اطلاعات انتشار: هشتمین کنفرانس ماشین بینایی و پردازش تصویر، سال
تعداد صفحات: ۵
The Local Binary Patterns (LBP) featureextraction method is a theoretically and computationally simpleand efficient methodology for texture analysis. The LBP operatoris used in many applications such as facial expression recognitionand face recognition. The original LBP is based on hardthresholding the neighborhood of each pixel, which makestexture representation sensitive to noise. In addition, LBP cannotdistinguish between a strong and a weak pattern. In order toenhance the LBP approach, Fuzzy Local Binary Patterns (FLBP)is proposed. In FLBP, any neighborhood does not representedonly by one code, but, it is represented by all existing codes withdifferent degrees. In FLBP, any fuzzy Intersection and Unionoperators may be used. In this study, the following operators areapplied and their results are compared together: Dot–Sum, Min–Max and normalized Min–Max. Based on the extensiveexperiments, the fuzzy Min–Max operators are more useful andcan improve the accuracy in the application of Facial ExpressionRecognition (FER) about 4% (i.e., form 82.98% to 86.88%).<\div>

۸High Angular Resolution Diffusion Image Registration
اطلاعات انتشار: هشتمین کنفرانس ماشین بینایی و پردازش تصویر، سال
تعداد صفحات: ۵
Diffusion Tensor Imaging (DTI) is a commonmethod for the investigation of brain white matter. In thismethod, it is assumed that diffusion of water molecules isGaussian and so, it fails in fiber crossings where this assumptiondoes not hold. High Angular Resolution Diffusion Imaging(HARDI) allows more accurate investigation of microstructuresof the brain white matter; it can present fiber crossing in eachvoxel. HARDI contains complex orientation information of thefibers. Therefore, registration of these images is morecomplicated than the scalar images. In this paper, we propose aHARDI registration algorithm based on the feature vectors thatare extracted from the Orientation Distribution Functions(ODFs) in each voxel. Hammer similarity measure is used tomatch the feature vectors and thin–plate spline (TPS) basedregistration is used for spatial registration of the skeleton and itsneighbors. A re–orientation strategy is utilized to re–orient theODFs after spatial registration. Finally, we evaluate our methodbased on the differences in principal diffusion direction and wewill show that utilizing the skeleton as landmark in theregistration results in accurate alignment of HARDI data.<\div>

۹MRI IMAGE RECONSTRUCTION VIA NEW K–SPACE SAMPLING SCHEME BASED ON SEPARABLE TRANSFORM
اطلاعات انتشار: هشتمین کنفرانس ماشین بینایی و پردازش تصویر، سال
تعداد صفحات: ۴
Reducing the time required for MRI, has taken a lot ofattention since its inventions. Compressed sensing (CS) isa relatively new method used a lot to reduce the requiredtime. Usage of ordinary compressed sensing in MRI imagingneeds conversion of 2D MRI signal (image) to 1D signalby some techniques. This conversion of the signal from 2Dto 1D results in heavy computational burden. In this paper,based on separable transforms, a method is proposed whichenables the usage of CS in MRI directly in 2D case. Bymeans of this method, imaging can be done faster and withless computational burden.<\div>

۱۰Cell Segmentation Unlimited Algorithm Based on Cytoplasm by Using BM3D Filter and Gradient Vector Flow
اطلاعات انتشار: کنفرانس بین المللی پژوهش های کاربردی در فناوری اطلاعات، کامپیوتر و مخابرات، سال
تعداد صفحات: ۱۱
Cell dynamic review is a very important issue in life sciences. For reviewing this dynamic, we need a complete and unlimited algorithm based on type and number of the cells. In this paper, we provide a Cell Segmentation Comprehensive Algorithm (CSUA) or a novel method which segment the cell without any limitation and fully automatic. In this method, first the noise of image is omitted by BM3D filter. Then it is calculated the intensity difference of each image. If the intensity difference is less than threshold, the image will become histogram equalization, otherwise we use a new method that the image will be passed through two parallel Gaussian filters (TPGF), and then two filtered images are subtracted to get an image with background trace omission. After some morphological operation, the image is segmented by using a gradient vector flow (GVF), then two new formulas about each structural data of image are provided based on cells area and we remove big and small redundant objects. Finally, we have a segmented and noiseless image of the cells. The results of the trials showed segmentation accuracy rate is 97.28% for all cell types.<\div>
نمایش نتایج ۱ تا ۱۰ از میان ۱۰ نتیجه