185 results about "Lesion detection" patented technology

The invention provides an index construction method for a medical ultrasonic database, and relates to the technical field of ultrasonic data processing, and the method comprises the steps: reading a multi-frame image sequence from a DICOM ultrasonic database, and generating a diagnosis intention vector for each frame; analyzing the diagnosis report, and segmenting the report into a plurality of segments; lesion detection and segmentation are performed on each frame of image to obtain lesion information, and a cross-frame aggregation strategy is adopted to identify a unified lesion object; extracting a feature vector of each focus object to obtain a focus object feature vector; establishing an alignment mapping between the focus object and the report fragment by adopting triple constraints, and generating a focus report mapping table; and constructing a multi-layer index structure, and storing the focus report mapping table and the multi-layer index structure in parallel. According to the method, a multi-layer index structure comprising a metadata index, a text index, a focus object vector index and a cross-modal embedding index is formed, so that efficient organization and focus-level accurate retrieval of medical ultrasonic data are realized.

The invention discloses a diabetic foot early-stage detection method combining infrared and visible light imaging, and relates to the technical field of diabetic foot medical imaging diagnos.The diabetic foot early-stage detection method comprises the steps that multi-mode image collection and standardization processing are conducted, infrared images and visible light images are synchronously collected, and standardization processing such as size normalization is conducted; image registration and space alignment are carried out, marking points are set based on foot anatomical features, and feature extraction, mismatching point elimination and transformation matrix calculation are carried out; extracting and screening multi-dimensional features, extracting temperature and structural features, and screening by using a Relief-F algorithm; feature lesion recognition and classification are fused, and lesion probability is output through a double-branch convolutional neural network; carrying out detection result verification and feedback optimization, and comparing a clinical diagnosis optimization model; and generating a detection report and storing data, and generating a report containing the fused image. The early lesion detection precision is improved through multi-modal fusion, individual and environment differences are adjusted and adapted in a personalized mode, and reliable technical support is provided for clinic.

The invention discloses a cerebral apoplexy focus detection method and system, and belongs to the technical field of medical image detection. Extracting a fusion feature map of the brain image; for each region type, obtaining a representative feature which has the highest similarity with the feature at each pixel point position in the fused feature map in the class prototype set and carries a corresponding region type label, and further determining the region type to which each pixel point position in the fused feature map belongs so as to obtain a corresponding pseudo-label map; the class prototype set comprises representative features of different region types and is obtained in the training process of the system, and feature distribution of different region types in the memory bank is calculated through a Gaussian mixture model; for each region type, sampling is carried out based on feature distribution of the region type, and a plurality of representative features are obtained; the prototype-like set in the cerebral apoplexy detection method has real global context perception ability, can clearly distinguish the focus and various complex background structures, and can accurately realize cerebral apoplexy detection.

The invention discloses a tiny residual focus detection method and system based on ultra-high depth sequencing, and relates to the technical field of tiny residual focus intelligent detection.The tiny residual focus detection method comprises the following steps that on the basis of a sequencing library, splitting is conducted according to a sample index to obtain a to-be-detected sample, and a consensus sequence is obtained according to a molecular identifier of the to-be-detected sample; based on a consensus sequence, filtering out the consensus sequence of which the mass value is less than 25 or the family size is less than 3, and combining a variation type and a distance from a fragment edge as noise introduced into an original nucleic acid molecular chain; a context sequence (context) and a chain direction are used as noise for introducing the capture level of PCR amplification; on the basis of the noise level, the circulating tumor DNA level is estimated in combination with tumor priori knowledge, and the MRD state is determined by detecting the significance of molecular signal sources. According to the invention, the sensitivity and specificity of MRD detection are improved.

The invention discloses an intelligent bronchial endoscope focus detection method, computer equipment and a storage medium. The intelligent bronchial endoscope focus detection method comprises the steps that real, reliable and diversified bronchial endoscope examination images are collected; through data preprocessing, labeling and data division, a bronchial focus detection data set is constructed; an efficient bronchial focus detection network is constructed, and improvement is carried out through multiple attention mechanisms and a convolution module; designing a loss function and training the network to obtain an optimized bronchial focus detection model; and performing focus detection on the acquired image in the real-time bronchoscopy video, and displaying a result. Based on the deep learning technology, the deep neural network model is trained to carry out precise focus type detection on the bronchial endoscope image, real-time bronchial lesion prompt information is provided for endoscope operators, the lesion omission problem during bronchial lesion examination is effectively relieved, the bronchoscopy process is standardized, the endoscope examination quality is improved, and the bronchial lesion detection efficiency is improved. And the workload of operators is reduced.

The invention discloses a focus intelligent detection and diagnosis system based on multi-modal medical image fusion, and relates to the technical field of medical image processing. A molybdenum target image, an ultrasonic image and a magnetic resonance image of the mammary gland of the patient are acquired through the data acquisition module; a data registration module is used for carrying out partition affine and local refinement registration on the multi-modal image by taking the molybdenum target image as a geometric anchor point and combining parameters such as compression force; extracting registered fusion feature data through a data fusion module; generating a candidate focus set by a focus detection module; finally, the risk assessment module calculates a comprehensive risk score through weighting, correction and consistency verification processes based on a plurality of interpretable feature indexes. And the output module generates a visual diagnosis result. According to the invention, intelligent detection and risk assessment of rechecking of breast lesions are realized, and the diagnosis accuracy and clinical reliability are effectively improved.

The invention relates to the technical field of medical images, in particular to a medical image intelligent detection and auxiliary diagnosis system based on deep learning, and the system comprises a data access module which is used for obtaining medical image data and clinical text data of a patient; the data fusion module is used for generating a focus feature vector and a text feature vector, and performing cross-modal alignment and fusion to generate a fusion feature vector; the diagnosis analysis module is used for executing focus detection, focus segmentation and focus classification tasks, generating a diagnosis result and generating a diagnosis label based on the diagnosis result; the decision generation module is used for mapping the generated execution result to a preset medical knowledge base and performing deep reasoning to generate an auxiliary diagnosis decision; the decision auditing module is used for performing confidence scoring on the auxiliary diagnosis decisions and selecting the auxiliary diagnosis decision with the highest confidence score as the final auxiliary decision; and the data visualization module is used for carrying out visualization processing on the auxiliary decision and the diagnosis result.

The invention relates to the technical field of image processing, and discloses an improved StarNet-YOLOv13-based unmanned aerial vehicle field tobacco virus disease lightweight detection method, which comprises the following steps: acquiring a to-be-detected image of a field tobacco plant, and then inputting the to-be-detected image into a trained tobacco virus detection model to obtain a detection result of the tobacco plant; wherein the tobacco virus detection model is obtained by replacing a backbone network with a star network, replacing a C3K2 module in a neck network with a feature fusion module and replacing a detection head with a DetectMBConv detection head in a YOLOv13 model; the feature fusion module is constructed by introducing a partial convolution mechanism into the C3K2 module; thus, through the star network (StarNet), the feature fusion module (DSC3k2PConv module) and the DetectMBConv detection head, the small target scab detection capability and the complex scene robustness can be improved while the parameter and calculation overhead can be reduced, so that the tobacco virus detection model can meet the deployment requirement of the computing power limited equipment side, such as an unmanned aerial vehicle, and the detection precision is ensured at the same time.

The invention relates to the technical field of medical images, in particular to a multi-modal medical image registration and fusion analysis method, which comprises the following steps of: eliminating image noise and artifacts based on a modal adaptive filtering strategy; constructing a pyramid type feature extraction network to realize multi-scale feature extraction, calculating feature matching degrees among different modal images, and dynamically adjusting matching weights by combining feature differences among modals; a focus area attention mask is constructed, targeted enhancement of registration image features is realized, and a hierarchical fusion strategy is adopted to evaluate the quality of a fused image; and constructing a multi-task deep learning model to complete focus automatic detection, segmentation and benign and malignant preliminary judgment on the fused image. According to the multi-modal medical image registration and fusion analysis method, a three-layer feature pyramid is constructed, a multi-feature fusion matching cost function is introduced, and cross-modal feature matching is optimized through an adaptive weight iteration nearest point ICP algorithm, so that the information richness, marginal definition and focus discrimination of a fused image reach the standard.

This application relates to the field of medical technology and provides a medical application scenario matching method, electronic device, and computer program product. The medical application scenario matching method includes: acquiring a medical image sequence; determining image information of the medical images in the medical image sequence, the image information including key information, which includes one or more of the following: imaging object information, phase information, lesion detection information, and image quality information of the corresponding medical image; and outputting at least one target application scenario adapted to the medical image sequence based on the key information. Embodiments of this application can prevent doctors from using unsuitable medical image sequences in specific application scenarios, thus helping to improve doctors' work efficiency.

An ultrasound diagnostic apparatus includes: an image acquisition unit (36) that performs a scan using an ultrasound probe (1) to acquire ultrasound images of a breast; a mammary gland region extraction unit (25) that extracts a mammary gland region from the ultrasound images; a thickness calculation unit (26) that calculates a thickness of the mammary gland region in a depth direction; a lesion detection unit (27) that detects a suspected lesion region for the ultrasound images; a frame selection unit (29) that selects, as an evaluation target frame group, at least frames which exclude a frame in which the suspected lesion region is detected, and in which the thickness of the mammary gland region is equal to or greater than a thickness threshold value, among a plurality of frames; and an evaluation unit (32) that performs a glandular tissue component evaluation on the ultrasound image of the evaluation target frame group.

The invention relates to the technical field of medical artificial intelligence and public health data management crossing, in particular to an intelligent tuberculosis screening and data analysis management system which comprises a data acquisition module, an intelligent screening module, a data analysis management module, a clinical interaction module and a system management module. The multi-source data fusion and the dynamic threshold reasoning enable the early-stage tiny focus detection rate to be larger than or equal to 85%, the screening time is shortened to 5 minutes per case (80% higher than that of manual film reading), cross-mechanism and cross-dimension data integrated acquisition and sharing are achieved, 'hospital-disease control-grassroots' data intercommunication is achieved, and the screening efficiency is improved. Complete data support is provided for epidemiological analysis, longitudinal individual tracking and transverse group analysis are combined, the early warning period of regional outbreak is larger than or equal to 2 weeks, and a public health prevention and control gateway is assisted to move forwards.

The invention belongs to the technical field of medical instruments, and relates to a telescopic intracranial aspiration and flushing system and an operation method. The problems that an existing endoscope is not high in function integration level, large in occupied operation space, long in operation time and lack of reference in the operation process are solved. The system comprises an endoscope module and a visual guiding module, the endoscope module and the visual guiding module cooperate to form an intracranial focus precise treatment integrated system under visual guiding, and full-process collaborative operation from focus detection, deep management and control to layered suction and synchronous flushing is achieved; the operation method comprises the steps of instrument assembly, focus detection and depth marking, operation chamber establishment, suction parameter presetting, negative pressure pipeline connection, layered suction and synchronous flushing, operation ending and the like. The endoscope tube, the suction tube, the flushing tube, the control button, the unlocking button, the adjusting hole and the like are integrated on the handle, one-hand operation is achieved, the operation steps are simplified, the operation space is saved, and the cooperation efficiency is improved.

The invention relates to the technical field of medical information processing, in particular to a cervical cancer cell early screening and intelligent evaluation system based on multi-modal data, which comprises a data acquisition module, a form detection module and a screening early warning module, the data acquisition module is used for acquiring multi-modal data of a current detection object in a plurality of inspection processes; the morphology detection module is used for calculating a cellular morphology score of a current detection object in any target inspection process; the screening early warning module is used for determining similar detection objects of the current detection object, and correcting differences between the current detection object and the similar detection objects in the aspects of cellular morphology scores and lesion detection indexes according to the approximation degree between different types of lesion detection indexes of the current detection object; and obtaining the screening early warning coefficient of the current detection object by using the relative change condition of the correction result. According to the invention, the early screening and evaluation effects on cervical cancer cells are improved.

The application discloses a kind of hysterosalpingogram automatic lesion detection method, system, comprising the following steps: 1: CMOS image sensor gathers analog signal in uterine cavity, and illumination is provided using LED;2: convert analog image signal into digital image signal by analog-digital conversion chip;3: using ISP processor, image signal is carried out ISP image processing;4: image quality is evaluated by image quality control module, and LED brightness is dynamically adjusted, 5: using NPU chip, image AI uterine cavity lesion detection is carried out to the image after processing;6: real-time output image and superimposed detection result display.The application system integrates advanced image acquisition, processing and display technology, proposes a kind of hysterosalpingogram automatic lesion detection method, through automation, intelligent way, the speed and accuracy of hysterosalpingography lesion detection are significantly improved, and strong support is provided for clinical diagnosis and treatment.

The invention discloses an endoscopic gastrointestinal tract image detection and recognition system and method based on artificial intelligence, and belongs to the technical field of medical image detection. Comprising an image acquisition module, an image preprocessing module, a feature extraction module, a lesion detection module, a lesion recognition module, a model training module, a result output module, a data storage module, a man-machine interaction module and a quality evaluation module, wherein the image acquisition module is used for acquiring image data in the gastrointestinal tract and transmitting the image data to the image preprocessing module; the image preprocessing module is used for preprocessing the image data. Through cooperative work of multiple modules, a whole-process intelligent detection and recognition system from image collection to result output is constructed, all the modules are clear in division of labor, data transmission is efficient, accurate processing and analysis of gastrointestinal tract images under an endoscope can be achieved, and the defects of traditional manual diagnosis and an existing AI diagnosis technology are effectively overcome.

This invention discloses an automatic segmentation and scoring method and system for FDG PET-CT lesions in lymphoma, belonging to the field of medical image analysis technology. It aims to improve the segmentation accuracy of lymphoma lesions and the objectivity of the Deauville score. First, standardized uptake values ​​are calculated for FDG PET images, and lymph node morphological features are extracted from CT images based on multi-scale Hessian enhancement filtering to construct a dual-modality PET-CT image pair. Then, a dual-channel depth network is used to extract anatomical structural features and metabolic distribution features respectively. Cross-modal gating fusion is used to suppress physiological uptake interference, outputting preliminary lesion segmentation results. Next, three-dimensional connected component analysis is performed on the segmentation mask, and metabolic heterogeneity index is extracted by combining kurtosis and Haar wavelet multi-scale energy. A graph attention network is used to identify key lesions. Finally, the ratio of key lesions to standardized liver uptake values ​​is combined with the metabolic heterogeneity index to correct the Deauville score, achieving automation from lesion detection to treatment efficacy evaluation.

The application provides a pathological section image analysis method and device based on an image coding large model, and belongs to the technical field of medical images. The method comprises the following steps: cutting an original pathological section image to obtain a pathological section sub-image set; sequentially inputting the pathological section sub-images in the set into a target region segmentation large model to obtain corresponding local target region distribution maps, and then obtaining an overall target region distribution map and generating a set of positive local target region distribution maps; based on the set of positive local target region distribution maps, using an abnormal lesion detection network to generate an abnormal lesion distribution map; and based on the target region distribution map and the abnormal lesion distribution map, finally realizing statistical analysis on the original pathological section image. The application can analyze various pathological section images, automatically identify and classify abnormal lesion regions in the images, and automatically statistically analyze to obtain key clinical indicators such as lesion rates, thereby improving the accuracy and efficiency of diagnosis.

The application discloses a micro blood flow imaging method based on adaptive density filtering and a related device, and relates to the technical field of ultrasonic imaging. The method comprises the following steps: extracting a static frame in original ultrasonic data, screening a high-quality static frame through correlation analysis to eliminate motion interference; performing SVD clutter filtering, and cutting off high spatiotemporal coherence singular values to separate tissue clutter and microbubble signals; performing adaptive density filtering and subtracting a preliminary filtering result to realize noise suppression and overlapping microbubble decoupling; and enhancing microbubble motion signals through time-lag autocorrelation processing, and finally completing power Doppler imaging and ultrasonic localization microscope reconstruction. The scheme effectively solves the problem that the prior art cannot simultaneously consider PDI noise suppression and ULM microbubble decoupling, significantly improves image signal-to-noise ratio and super-resolution, avoids problems such as signal distortion, artifacts and loss of low-speed blood flow, and provides more reliable imaging support for clinical diagnosis such as blood vessel lesion detection and tumor blood flow evaluation.

An adaptable deep learning method is provided that delivers sound hepatic lesion identification in NETs, while significantly reducing human effort for data annotation and improving model generalizability for PET image quantification. A region-guided GAN (RGGAN) model conducts image-to-image translation between list-mode simulated PET images and real-world clinical data, while preserving semantic content of interest, e.g., lesions. The RG-GAN model is integrated with a lesion detection model into an end-to-end, unified framework for joint-task learning, such that the two models can benefit from each other. The RG-GAN translates the list-mode simulated data into real world-style images, which appear to be drawn from the real clinical PET image dataset, and feeds the translated images into the lesion detection model for training. In order to deal with the limited diversity of list mode-simulated PET image data, a specific data augmentation module is incorporated into the unified framework to improve model training.

The invention relates to the field of medical image processing, and provides an intelligent screening and benign and malignant auxiliary identification system for a tiny tumor in an ultrasonic image, which comprises a dynamic ultrasonic sequence preprocessing module, a space-time double-flow feature extraction module, a multi-scale focus detection module, a space-time feature fusion module and an interpretable classification decision module, the system extracts the elasticity and mobility characteristics of a focus in an ultrasonic dynamic image through an optical flow method, performs multi-scale focus detection and benign and malignant classification in combination with static morphological characteristics, outputs an interpretable thermodynamic diagram to provide decision support for a doctor, realizes parameter adaptive optimization through a closed-loop feedback mechanism, and improves the detection accuracy. The method is particularly suitable for screening small tumors with the diameter smaller than 1 cm, and the classification accuracy and the detection sensitivity are remarkably improved.

The invention relates to the technical field of medical diagnosis assistance, and discloses an ear-nose-throat endoscope intelligent diagnosis assistance system based on AI image recognition, and the system comprises an image collection module which is used for receiving and preprocessing video stream data from an electronic nasopharyngolaryngoscope or an otoscope in real time; the process quality monitoring module is used for analyzing the video stream data in real time so as to judge the motion state and the image definition of the endoscope; the focus recognition and marking module is internally provided with a trained AI image recognition model, image quality is guaranteed through preprocessing of the AI image acquisition module, and inspection operation is standardized through the process quality monitoring module; the AI model accurately identifies and classifies multiple types of lesions, warns and assists doctors in focusing key problems in real time, improves the detection rate of the lesions and reduces the risk of missed diagnosis; a risk sorting report is automatically generated and hospital uploading is supported, so that the burden of doctors is relieved and the efficiency is improved; and in combination with a feedback optimization model, long-term clinical adaptation is realized, and stable and intelligent assistance is provided for ear-nose-throat endoscope diagnosis.

The invention discloses an X-ray machine intelligent examination scheme creation method based on pet identity recognition, and relates to the technical field of veterinary diagnostic images.The method has the advantages that variety-specific artifacts are accurately inhibited, modeling is performed through breathing movement driven by skeleton points, high-frequency vibration noise is separated, dynamic blurring of a trachea area of a short-nose dog is eliminated, and the accuracy of the examination scheme is improved. The tiny lesion detection capability is improved; the limitation of a single angle threshold value is avoided by adopting a composite geometric criterion, so that pets with different body types can be clearly imaged under abnormal body positions; full-process identity data binding, biological characteristic penetration scanning protocol formulation, radiation protection and data storage links are provided, real-time association of images and individual medical records is ensured, and manual checking errors are avoided; a block chain evidence storage mechanism is introduced to meet the electronic medical record compliance requirement of a diagnosis and treatment institution, and cross-institution data security sharing is realized; historical parameters are automatically loaded, repeated debugging is reduced, new pets are rapidly initialized through a variety parameter library, and the equipment preparation time is remarkably shortened.

The application discloses an intelligent fish body surface lesion detection system for an underwater aquaculture scene, and relates to the technical field of intelligent aquaculture monitoring.The system comprises an underwater image acquisition module, which is used for acquiring underwater images of fish in real time in an aquaculture water body, ensuring illumination through a light supplement lamp, and transmitting image data to an edge computing processing module; the edge computing processing module is based on an edge inference chip and is used for quality enhancement, target positioning, feature extraction and enhancement, target detection and multi-task decoding of the underwater images, and real-time detection and analysis of fish body lesions; an aquaculture management cloud platform module is used for receiving and storing detection results, statistically analyzing lesion data, and providing disease early warning, and for pushing alarm information to aquaculture personnel when a lesion is found, and for realizing account and permission management.The system is used to realize high-precision, real-time and robust detection and classification of various lesions on the surface of fish, and to provide remote monitoring and disease early warning services.

The invention provides a face acne skin lesion detection matching method, system and device, and relates to the field of image processing, and the method comprises the steps: calling a face key point detection interface, and obtaining face key point coordinates on face images at different angles; obtaining a first mapping transformation reference point of each face key point corresponding to the side face and the front face; according to the first mapping transformation reference point, generating a perspective transformation matrix of mapping the side face to the front face; according to the perspective transformation matrix and the human face key point coordinates, mapping the skin lesion detection frame on the side face to a position corresponding to the front face; matching the skin lesion detection frame on the front face with the skin lesion detection frame mapped to the front face by the side face; and unifying the skin lesion type labels of the skin lesion with the front face and the side face successfully matched. According to the invention, the accuracy of acne evaluation results can be improved.

This invention discloses a method and system for targeted detection of internal lesions in mangoes, belonging to the field of agricultural product testing technology. It includes: acquiring X-ray images of the mangoes to be tested; a processing unit analyzing the X-ray images to determine the presence of internal lesions, outputting the determination result and confidence level; determining whether the confidence level meets a preset threshold; if it does, generating a rejection control signal; otherwise, generating an angle adjustment signal; an angle adjustment actuator adjusting the mango placement angle according to the angle adjustment signal; re-acquiring X-ray images based on the adjusted mangoes; and a processing unit fusing the two X-ray images to re-determine the presence of internal lesions. The system includes a material handling area, an X-ray detection component, a sorting area, an angle adjustment component, a conveyor line, and trays. The trays include an inner tray and an outer tray that are nested together and connected by a rotating shaft, allowing adjustment of the mango placement angle. This invention improves the accuracy of detecting internal lesions in mangoes through multi-angle X-ray imaging and image fusion analysis.

This invention relates to the field of biomedical technology, and more particularly to a real-time lesion AR annotation method for endoscopic systems based on augmented reality. The method first acquires and preprocesses a real-time video stream from the endoscope; then, a two-stage cascaded model is used to detect and segment lesions in the preprocessed video stream, identifying suspicious lesion areas, generating a segmentation mask, and removing interfering areas in the endoscopic scene; real-time pose data of the endoscope is acquired using an electromagnetic-visual fusion positioning method; based on the segmentation mask and real-time pose data, two-dimensional lesion information is mapped to three-dimensional space to construct a three-dimensional lesion model, which is then completed by combining anatomical structural features and adjusted according to tissue deformation; finally, the adjusted three-dimensional lesion model is back-projected onto the endoscopic field of view to generate AR annotation information, which is then fused with the real-time endoscopic image and output. This invention enables real-time and accurate annotation of endoscopic lesions, providing auxiliary support for endoscopic diagnosis and treatment.

The invention relates to the technical field of enhanced imaging, in particular to a fluorescence metabolism enhanced imaging method and system based on three-mode fusion, and the method comprises the steps: confirming three-mode colposcope imaging equipment, carrying out the illumination uniformity calibration operation on the three-mode colposcope imaging equipment, obtaining calibration imaging equipment, obtaining a three-mode registration image set, and carrying out the illumination uniformity calibration operation on the three-mode colposcope imaging equipment; performing dynamic monitoring based on a plurality of preset acetic acid action time points to obtain a lesion evolution dynamic video, performing lesion area identification analysis operation on the registered fluorescence metabolism image to obtain a metabolism characteristic parameter set, performing time sequence dynamic characteristic analysis operation on the lesion evolution dynamic video to obtain a time sequence dynamic characteristic parameter set, and performing dynamic analysis on the time sequence dynamic characteristic parameter set; and obtaining a lesion risk grading result based on the metabolism characteristic parameter set and the time sequence dynamic characteristic parameter set, and completing fluorescence metabolism enhancement imaging based on three-mode fusion based on the lesion risk grading result. According to the invention, the imaging effect and the lesion detection sensitivity can be improved.

The invention discloses a generative artificial intelligence cerebral artery lesion detection method based on plain-scan brain CT. The method comprises the steps that 1, a plain-scan brain CT image to be processed is acquired; 2, converting the plain scanning brain CT image into a CTA image through an adaptive noise elimination network; and 3, carrying out multi-modal lesion detection on the basis of the CTA image generated in the step 2. According to the method, a CTA image is generated through an adaptive noise elimination network (ANE-NET), and a real-time lesion feature analysis engine (RTAL-FE) is embedded in the generation process, so that real-time classification prediction of lesion types is realized. And furthermore, a detection result is output through a dynamic weight decision model (DWD-M), so that the generation quality and the detection efficiency are remarkably improved. The problems that a traditional method is low in generation quality and lags behind detection are solved, and the method is particularly suitable for low-dose and non-invasive cerebrovascular disease screening scenes and has important clinical application value.

Embodiments of the present application provide a medical image synthesis device and method. The method comprises: obtaining a first medical image and a second medical image; registering the first medical image and the second medical image; determining first parameter values of each pixel position on the registered first medical image and second parameter values of each pixel position on the second medical image; multiplying the first parameter values and the second parameter values on the same pixel positions of the registered first medical image and the second medical image, and generating a synthesis image data according to the multiplication result. Therefore, the multiplication of the first parameter values and the second parameter values is equivalent to amplifying the difference of pixel values of the positions of normal tissue and lesion tissue, so that the contrast and resolution of the synthesis image are higher, thereby making the lesion area clearer and improving the lesion detection efficiency.