46 results about "Multimodality image registration" patented technology

A method for multimodal image registration includes providing a pair of images acquired from differing imaging modalities, defining an intensity correction function that corrects the intensities of a first image in terms of the intensities of the second image, defining a registration transformation function that registers the second image with the first image, wherein said intensity correction function and transformation function are functions of a plurality of parameters, obtaining corrections to the plurality of parameters by minimizing an energy functional of a square difference of the intensity corrected first image and the registration transformed second image, and updating the intensity correction function and the registration transformation function based on the corrected plurality of parameters.

The invention provides an image registration method based on improved structural similarity. According to the invention, the improved structural similarity serves as the objective function of the image registration for the first time; four parameters of the two-dimensional image rigid body transformation are obtained through translation, rotation and consistent scaling along the X-axis and Y-axis; and the single-modal and multimodal images are analyzed in detail based on the registration algorithm and performance of the structural similarity and are compared with that based on a normalized mutual information registration algorithm. The result shows that when an absolute value is extracted during defining the structural similarity, the structural similarity has favorable features of a convex function; for either the single-modal image registration or the multimodal image registration, the structural similarity serving as the measure function can achieve the sub-pixel registration with registration precision and robustness better than that based on the classic normalized mutual information registration algorithm; and if K1 is less than or equal to 0.000001, and K2 is less than or equal to 0.000003, the two-value image can achieve the pixel registration.

The invention provides a head medical model quick forming method based on 3D printing. According to the method, CT/MR multimode medical images are used, a three-dimensional model is quickly established for head tissue/organs, and a 3D printing method is used for carrying out quick forming on the three-dimensional model. The method comprises the steps that (1) a multimode image registration technology is used, and the CT/MR images are registered into a unified space coordinate system; (2) according to medical information provided by the CT/MR images, different kinds of head tissue/organs are extracted; (3) the three-dimensional model is established for the extracted tissue/organs; and (4) the three-dimensional model is subjected to layering layer by layer, cross section data after layering are obtained, and 3D printing is carried out according to the cross section data. According to the CT/MR images, the head tissue/organs can be subjected to quick and accurate modeling, the manufacturing speed and the accuracy of a head medical model can be effectively improved, and a customized and personalized head medical model can be provided.

A multi-modality image registration method includes the following steps: sampling reference images and images to be registered according to a first sampling proportion to obtain the reference images and the images to be registered with first resolution ratio, adopting a B spline surface as a deformation model to perform matching on the reference images and the images to be registered, calculating matching measure and obtaining first matching parameters, and increasing sampling proportion to repeat the processes until the matching measure reaches a reset threshold value, wherein in matching, matching parameters of last time are used as initial matching parameters after each-time sampling proportion is increased. An operation navigation method based on multi-modality image registration is further provided.

The invention provides a multispectral-based fundus image registration method, comprising the steps: acquiring the multispectral fundus image through RHA, and selecting a template fundus image and a fundus image to be registered; constructing a regression network for obtaining a deformation field between the template fundus image and the fundus image to be registered, and adding a feature balancelayer and a pyramid structure into the regression network; constructing a loss function, wherein the loss function comprises similarity loss of the blood vessel label graph and flattening constraint of a deformation field; and training the network in a weak supervision mode, and taking the processed segmentation image of the fundus image as a label of the training network. According to the multispectral-based fundus image registration method, the registration model is trained by taking the segmentation image as guidance and adopting a weak supervision mode, and practice proves that the methodcan effectively solve the problem of multispectral fundus image registration, and a good effect can be achieved, and the weak supervision segmentation oriented registration method can be expanded to more multimodal image registration problems.

The invention discloses a prostate image registration system and method based on automatic segmentation and pelvis alignment, and mainly solves the problems of large modal difference and low registration precision when prostate multi-modal medical image registration is realized in the prior art. The implementation scheme includes: performing joint segmentation on the prostate CT image sequence andthe prostate MRI image sequence by using a trained prostate multi-modal image segmentation network U-net to obtain a prostate CT image segmentation result image set C and a prostate MRI image segmentation result image set M; and performing two-stage registration on the two image sets C and M to obtain a final registration image set L; and finally, fusing and displaying the final registration image set L and the prostate CT image segmentation result image set C. According to the invention, the prostate multi-modal image difference is reduced, the problem of small-deformation image registrationis solved, the multi-modal image registration efficiency and registration precision are improved, and the method can be used for observing the shape and position change of a prostate tumor anatomicalpart.

The invention is suitable for the technical field of pattern recognition, and provides a multi-modal image registration method, which comprises the steps of obtaining a first image of a source modal and a second image of a target modal paired with the first image; correcting the intensity of the first image to obtain a first corrected image, so that the intensity of the first corrected image is matched with the intensity distribution of the second image; acquiring a deformation field of the first corrected image registered to a target mode according to the first corrected image and the secondimage; and according to the first image and the deformation field, obtaining a registration image of the first image registered to a target mode. Therefore, the influence of the image intensity characteristics is reduced, and the precision reduction caused by overlarge distance difference between the same object of the to-be-corrected first image and the ideal registration result can be avoided, so that the registration precision is improved.

The invention provides a multimode image registration method and system. The first point making the distance between every two straight lines across different surfaces in multiple straight lines across different surfaces shortest is determined by establishing a registration model provided with multiple straight lines across different surfaces and utilizing the uniqueness of the point shortest away from the straight lines across different surfaces, mapping data of the first point in a first image and a second image is respectively obtained, then rotation matrixes and translation vectors from the first image to the second image are obtained according to the mapping data, and the rotation matrixes and the translation vectors serve as registration parameters of the second image and the first image. Due to the uniqueness of the first point between every two straight lines across different surfaces, the accuracy of registration between the first image and the second image is ensured, and registration errors between the first image and the second image are decreased.

The invention discloses a multi-modal image registration method and device, electronic equipment and a storage medium. The method comprises the following steps: judging whether a to-be-registered areaexists in a to-be-registered image and a registration reference image or not, wherein the to-be-registered image and the registration reference image are medical images acquired by using imaging equipment with different modes; performing two-dimensional sliding window search in a first image direction and two-dimensional sliding window search in a second image direction perpendicular to the firstimage direction on the to-be-registered image and the registration reference image under the condition that the to-be-registered area exists in the to-be-registered image and the registration reference image; and determining registration information of the to-be-registered image registered to the registration reference image according to a sliding window search result. Coarse registration of themulti-modal image is realized, the calculation amount is small, and the efficiency is high.

The invention provides a cross-modal medical image registration method and a computer readable storage medium. The method comprises the following steps: constructing a cross-modal medical image registration model of a full-resolution residual registration network; connecting a floating image and a reference image by adopting a full-resolution stream, and sequentially adding residual errors returned by a multi-scale residual error stream parallel to the full-resolution stream to enrich full-resolution feature information of the full-resolution stream; gradually reducing the number of channels through a continuous residual learning module, and estimating a full-resolution deformation field through a 3D convolution module; warping the floating image through a spatial transformation network based on the full-resolution deformation field so as to evaluate the similarity between the warped floating image and the reference image; training a model; and inputting the floating image to be registered and the reference image for registration to obtain multi-modal image registration. The problems of potential information loss and poor local registration quality in full-resolution prediction ofa conventional single-stream encoder decoder structure are solved by adding full-resolution streams.

The invention relates to a multi-modal medical image rigid registration algorithm based on deep reinforcement learning, which is a new end-to-end multi-modal image registration method, is trained by an asynchronous dominant actor commentator (A3C), can simulate the gradual registration process of human experts, and enhances the interpretability of the registration result. In view of a severe challenge of a multi-modal registration task in the aspect of calculation complexity, reinforcement learning and an attention mechanism are combined, so that an intelligent agent can capture more abstract high-dimensional features, introduction of a deep neural network with huge parameter quantity is avoided, and the characteristics of light weight and high efficiency of the network are kept, so that a model is easy to train, and the accuracy of registration is improved. And the method has strong robustness and generalization ability, and can drive the model to distort the moving image along the correct direction.

The invention discloses a multi-modal image registration method and system based on depth global features. The method comprises the following steps: preparing a data set and making the data set; making a data set; preprocessing the image; designing a full-convolution twin network structure; extracting image features; obtaining a similarity score graph; based on the similarity score graph, using a comparison loss function to optimize the similarity score graph, using a peak value loss function to increase the convergence speed, and using a sorting loss function to enable similarity distribution of the positive samples to be close to real distribution; jointly optimizing the comparison loss, the sorting loss and the peak loss, loading the trained weight into the model, reading all test set data in sequence, and predicting translation parameters of a floating graph in a test set in a reference graph; and determining the position of the floating image on the reference image to realize multi-modal image matching. According to the method and system, a high-precision matching result can be obtained by deeply mining common characteristics of bottom layers of different-source images.

The invention discloses a brain MR medical image registration method. The method comprises the steps of 1, segmenting a reference image and a floating image by adopting a BCFCM method; 2, performing symmetry axis detection on the original reference image and the original floating image by adopting an MSR detection method, and extracting a symmetry axis equation; 3, performing threshold segmentation on the image processed in the step 1; 4, performing approximate symmetry constraint on the image processed in the step 3 according to the symmetry axis equation detected in the step 2, and marking the obtained image as a reference image and a floating image to be registered; 5, initializing rigid body transformation matrix registration parameters; and 6, registering the image through an SSD similarity measurement criterion to obtain an optimal transformation matrix registration parameter, segmenting the reference image and the floating image by using a BCFCM method; and performing binarization threshold processing on the segmented image to separate the background before separation, performing symmetry constraint on the image after binarization processing, and applying the image after binarization processing to multi-modal image registration under an SSD framework. The registration efficiency, the precision and the robustness are improved.

The invention discloses an unsupervised multi-modal image registration method based on integrated attention enhancement. The method comprises the following steps: 1, preprocessing a medical image; 2,designing a registration framework, constructing a convolutional neural network model, automatically learning network parameters by optimizing similarity measurement of image pairs, and directly estimating a deformation field of the image pairs; and 3, dividing the image data into a training set and a test set, training a network model by using the training set, and finally testing the test set byusing the trained network model. According to the method, the deformation parameters of the image pair are estimated by directly optimizing the target function through the deep learning technology, the method can adapt to different data for different data and has high generalization capacity, the cascade encoder and the EAM are designed to extract the characteristics useful for the registration task, and the registration performance is improved to a certain extent.

The invention belongs to the field of image processing, discloses a multi-modal image registration method based on deep learning, and is used for solving the problem that the error is large when a traditional image registration method is used for registering complex multi-modal images shot by cameras in different imaging modes. The image registration method comprises the following steps: firstly, putting original images in an image set A into a semantic segmentation network to obtain a segmented image set B; multiplying the corresponding images in the image set B and the image set A pixel by pixel to obtain an image set C; and then selecting one image from the image set C as a fixed image, calculating geometric transformation by using gray value information of the image to obtain a deformed image, calculating the similarity between the deformed image and the reference image, iterating the optimal transformation T with the maximum similarity of the two images through an optimization algorithm, and applying the transformation T to the image set A to complete registration of the images. According to the method, the complex multi-modal image can be registered, and the method has the characteristics of high registration precision and real-time performance.

The invention relates to a multi-modal image registration method, device and system and computer equipment, and the method comprises the steps: obtaining fixed first position information of a positioning part in a functional image imaging system, and storing the first position information in a data storage part; performing morphological image scanning on the positioning part to obtain a morphological positioning image of the positioning part; obtaining second position information of the positioning part in a morphological image imaging system based on the morphological positioning image; and registering the functional image and the morphological image based on the corresponding relationship between the first position information and the second position information. According to the multi-mode image registration method, device and system and the computer equipment, image registration can be carried out only by installing the positioning part in the functional image imaging system and carrying out morphological image scanning, operation is easy, meanwhile, a radioactive source does not need to be used, radiation of an operator is reduced, and the quality control cost is reduced.