Abstract
With the rapid development of artificial intelligence and medical technology, sensor technology, as an important tool to obtain medical data, provides a wealth of physiological parameter information, but the traditional medical image fusion methods are difficult to make full use of the information provided by sensors. Therefore, this paper aims to explore the application of artificial intelligence digital holographic technology based on medical sensor in medical image fusion, and improve the accuracy and effect of medical image diagnosis and treatment. Based on the physiological parameter data obtained by medical sensor, combined with deep learning and image processing algorithm, a medical image fusion method based on artificial intelligence digital holography technology was proposed. The method uses the physiological parameter information provided by the sensor and combines multiple medical image sources to realize holographic medical image presentation. The experimental results show that the artificial intelligence digital holographic technology based on medical sensor can effectively improve the resolution and information content of medical image fusion. Compared with traditional medical image fusion methods, this method has higher image quality and more accurate diagnosis results.
Keywords
Medical sensor,Artificial intelligence,Digital holographic technology,Medical image fusion,Development and application
1. Introduction
Since the 1970s, advances in electronic equipment technology, microcontroller parallel processing technology, and the rapid development of array storage systems have stimulated the continuous development of information resource management [1]. On this basis, sensor technology has also made major achievements. Sensors and machinery with excellent technology and various forms have been used in many industries such as national defense and civil information content [2]. With the different sensor principles and improved characteristics, the information content methods obtained in the system software are diverse and complex, and the amount of data information is huge [3]. We all know that in the same system software, multiple sensor groups must cooperate with each other to achieve the goal, and then promote the collection function, and its information content has redundant design and diversity. In order to better ensure the correct acquisition and fusion of important information in data processing methods, information fusion technology has emerged [4]. From the beginning, the B. Shalom expert and professor of the University of Connecticut clearly stated that the trial, application and promotion of the multi-information fusion model system software in the national defense industry has promoted the rapid development of national defense level, and information fusion technology has become the research of many experts and professors [5]. Image is a way for people to observe things, understand the world, and discover the source of information with human eyes. It is also a key way for everyone to obtain and convey information. After entering the 21st century, multilateral detection technology has received widespread attention, scientific research and continuous development, and its application scale has shifted from the defense industry to the civilian field, such as satellite remote sensing image processing, multi-focus image fusion, road traffic, service robots, medical diagnosis, etc [6]. Multi-modal medical image technology is the best way to solve the above problems. It is to combine different forms of medical images according to scientific and standardized combination optimization algorithms to obtain useful medical information content, and then organically and chemically recombine them into one sheet [7]. The complementary combined result images describe the anatomical structure, pathological features, morphological and functional changes of internal tissues and other comprehensive information, which is helpful for clinicians to accurately locate the location and scope of the disease and provide a strong and reliable basis for rational treatment of patients. Nowadays, medical image fusion has been successfully and widely used in the diagnosis of brain degeneration, the diagnosis plan of tumor radiotherapy, the identification and precise positioning of benign tumors, the diagnosis of Alzheimer’s disease, the formulation of treatment plans and many other aspects [8]. Therefore, according to the image features of digital holography technology, this article clearly proposes a medical image fusion method that constructs information content based on texture features and correlation. First, the source image is decomposed by NSCT using artificial intelligence technology to obtain the high and low frequency subband indicators [9]. Subsequently, considering the sensitivity of human visual effects to texture features, the box-counting dimension of some differential signals was clearly proposed, and the texture information content of the image was statistically analyzed; the NSCT high-frequency subband sibling index and the parent-child index were analyzed [10]. The strong correlation between the index and the structural similarity of the Laplacian kinetic energy sum of the connected domain is calculated separately, and the information content is constructed for the theoretical correlation index between the high-frequency subbands.
As an indispensable part of modern medical technology, medical sensor has the ability to obtain patients’ physiological parameters, and is widely used in the field of medical image acquisition and monitoring. The sensor can measure and record physiological parameters such as ECG signal, brain wave, respiratory rate, body temperature, etc., which provides a lot of clinical decision basis for doctors. However, the physiological data collected by sensors alone often cannot fully and accurately reflect the patient’s condition and disease characteristics. Traditional medical image fusion methods mainly rely on medical imaging techniques, such as CT, MRI, etc., to fuse different image sources into a comprehensive image. Under this method, it is difficult to obtain the physiological parameter information of patients only by relying on medical image data, which limits the comprehensive analysis and accurate judgment of the disease. Therefore, in order to make full use of the physiological parameter information provided by the sensor and improve the effect and accuracy of medical image fusion, artificial intelligence digital holographic technology based on medical sensor came into being. By fusing physiological parameters obtained by sensors with medical image data, the technology realizes comprehensive medical image presentation. Through digital holographic technology, doctors can simultaneously observe the physiological state and lesion location of patients, and further improve the accuracy and effect of medical diagnosis. The application of artificial intelligence digital holographic technology based on medical sensor in medical image fusion provides doctors with more comprehensive and accurate medical information, which is expected to provide strong support for early diagnosis and individualized treatment of diseases.
2. Related work
The literature introduces the causes of medical image fusion and its significance in clinical medicine. Then, it discusses the importance of China’s research on this subject and the status quo of research on this subject in countries around the world; once again analyzes the current multi-modal medical image fusion research is an urgent problem in scientific research [11]. The literature introduces the whole process and characteristics of CT, MRI, PET, and SPECT imaging; then introduces the hierarchical division of multimodal medical image fusion in detail, describes the basic framework of image fusion; then introduces the definition of multimodal medicine in detail Image fusion optimization algorithm; finally, the basic evaluation index system of the current combined image is introduced in detail [12]. The literature introduces the human visual effect characteristics of the responsive PCNN and the medical image fusion optimization algorithm: first, the NSCT conversion is introduced in detail, and then the research results of experts and scholars from all over the world are analyzed, and a method based on the NSCT field is clearly proposed [13]. The new medical image fusion optimization algorithm fully considers the characteristics of medical images and needs to meet the characteristics of human visual effects. For the low-frequency sub-bands obtained by NSCT dissolution, several closely-combined optimization algorithms are clearly proposed, which can be described more strongly the kinetic energy and contour characteristics in the sub-band index; for the high-frequency sub-band, global coupling is used [14]. PCNN is used as the physical model, the regional visual effect saturation is used as the external stimulus input of PCNN, and the visual effect comparison sensitivity is used as the connection compressive strength of PCNN. At the same time, it also integrates the sigmoid function, which promotes PCNN not only to reduce the iterative update time, but also to adjust the changes responsively [15]. Finally, the results of medical image fusion are tested and analyzed. The literature introduces the medical image fusion optimization algorithm of texture features and the theoretical relevance of the construction of information content: firstly, it combines the image features of multi-scale conversion, scientific research results and analysis of countries around the world, and proposes a new multi-modal medical image fusion algorithm [16]. Subsequently, we analyzed the strong correlation between the NSCT high-frequency subband sibling index and the father-son index, and calculated the structural similarity between the coefficients and the neighborhood Laplacian energy sum. It clearly proposes the adaptive fusion of the low-frequency Sigmoid function, and uses the generalized correlation structure information method for the high-frequency. Secondly, for the low-frequency subbands, the box-counting dimensions of some differential signals are clearly proposed, and the texture information content of the image is statistically analyzed. Finally, the effectiveness of the optimization algorithm is verified through experimental comparison. The literature introduces methods to improve the quality of reproduced images on frosted surfaces. The experiment verifies the methods of filtering out the affected images on the reproduced image field through the mean removal method, band filtering method and image surface filtering method; according to the test, the mean filtering and Lee are analyzed. The basic principle of filtering and wide filtering to remove the actual effect of speckle noise on frosted surface.
3. Artificial intelligence digital holography technology
3.1. Artificial intelligence
Medical sensors can sense and measure the body’s physiological indicators and environmental data, such as heart rate, blood pressure, temperature and so on. These sensors transmit data to AI systems that, through data analysis and pattern recognition, can help healthcare workers make accurate diagnosis and treatment decisions. The development and application of medical sensors provide more data support and real-time monitoring capabilities for artificial intelligence technology, further promoting the application of artificial intelligence in medical image fusion.
In the debate about “machine intelligence” enlightenment education, the terms cybernetics and artificial intelligence are candidates side by side. In 1955, John McCarthy named the term “cybernetics” as “artificial intelligence technology”, which is different from Norbert Wiener (the originator of cybernetics). In the summer of 1956, after the Summer Dartmouth Artificial Intelligence Technology Seminar, artificial intelligence technology gradually became an independent research field.
McCarthy and Niels Nielsen later developed another expression of artificial intelligence technology, namely AI=Automation of Intelligence. From the perspective of engineering projects, this view is consistent with Wiener’s concept of cybernetics. The essence of artificial intelligence technology is the automation technology of professional knowledge.
In the book “Artificial Intelligence: A Modern Method”, Stuart Russell and Peter Noviget try to define artificial intelligence technology: artificial intelligence technology is a related “scientific research and design plan for intelligent agents” Knowledge, and “the main body of artificial intelligence technology refers to the system software that can observe the surrounding natural environment and take actions to achieve goals.”
Artificial intelligence technology is divided into three levels: weak artificial intelligence, strong artificial intelligence and super artificial intelligence. Also known as limited industry artificial intelligence technology/applied artificial intelligence technology, it refers to artificial intelligence technology that is dedicated to and only deals with special industry problems. For the time being, there is no need to allocate energy to worry about the future or to prepare for possible machine threats. An objective understanding of the nature of artificial intelligence technology and the level of development trends is conducive to grasping the current and predicting future development trends and preparing for the better development of artificial intelligence technology. The innovation factory predicts and analyzes the development trend of artificial intelligence technology in the diagnosis and treatment industry based on its technological maturity and future development trends. The development trend of artificial intelligence technology in the diagnosis and treatment industry will mature in the next 3–5 years. Therefore, combined with the current level of artificial intelligence, we should work to integrate artificial intelligence into medical treatment better and faster, and improve human health and living standards.
3.2. Digital holography technology
Data holographic projection is based on the basic development trend of electronic optical holographic projection. Choose CCD or CMOS to record the intelligent holographic projection, and simulate the light wavefront signal of the complex according to the Fresnel-Kirchhoff transmission integral formula, including the phase difference and intensity information content. According to the correction phase difference, the external economic appearance of the surface can be quantitatively analyzed. Among them, the theoretical basis of the holographic projection and reproduction optimization algorithm for fog data is the Fresnel transmission integral. Therefore, this chapter first introduces the Fresnel transmission integral in detail. On this basis, a mathematical analysis model for the recording and reproduction of rough surface data holographic projection is established. , And use Matlab to write Fresnel integral one-time Fourier transform optimization algorithm and convolution optimization algorithm to realize the analog generation and reconstruction of off-axis digital holograms.
The theoretical basis of the scientific research of light diffraction is Kirchhoff’s basic theory of transmission, Rayleigh-Sommerfeld’s basic theory and diffraction angle spectrum expansion basic theory. These three basic theories all use light as a scalar to solve this problem. Ignoring the nature of the magnetic field vector material, experiments conducted in the field of microwave heating spectroscopy show that if the scalar basic theory can meet two basic standards, one is that the transmission diameter must be much larger than the wavelength of the light wave; the other is not to be too close to the transmission diameter. Observing the transmission field, the result calculated by the basic theory of scalar transmission is appropriate. According to the basic theory of scalar transmission, Kirchhoff formula calculation, Rayleigh-Sommerfeld formula calculation and Fresnel transmission integral can usually be called classical transmission formula calculation.
4. Research on medical image fusion system based on artificial intelligence digital holography technology
4.1. Medical image fusion classification
4.1.1. Hierarchical division of multimodal medical image fusion
Different visualization machines and devices can visualize the same location of the body in the same time frame, and medical sensors are able to pick up a variety of physiological signals, such as heart rate, blood pressure, and temperature, and convert them into digital signals. But simple accumulation alone is not enough to eliminate image overlap. To overcome this problem, scientists have developed a reasonable combination method of multimodal medical images based on scientific standard combinatorial optimization algorithms. This method forms valuable images by complementing different information contents, so that the resulting images are more visual, more comprehensive and more accurate. This provides a strong scientific research guarantee for clinical medicine. The whole process of multimodal medical image fusion includes image preprocessing, support vector machine algorithm, feature classification, management decision and analysis, and comprehensive utilization of results. In the process of multi-modal medical image fusion, medical sensor plays an important role. Abstract results of information content at pixel level, feature level and management decision level can be obtained from physiological signals collected by medical sensors. These levels of information content abstraction results can help doctors better understand a patient’s condition and make accurate diagnosis and treatment decisions.
- (1)Pixel-level medical image fusion
The more commonly used is the pixel-level image fusion optimization algorithm, which is to immediately combine all source images that have undergone strict registration. The bottom layer of the combined layer provides powerful analysis, acquisition, and resolution for the next layer. This method performs real-time operations on the original record and can integrate different forms of the same overall target. It not only contains the residual information content and supplementary information content of the source image, but also enables the combined result image to affect the overall target tissue and internal organs. The description of organs or diseases is more vivid, comprehensive and accurate. It can capture the key point information content, edge features and texture changes of the source image, which is more conducive to doctors’ understanding and analysis of diseases, as well as diagnosis and providing reliable basis for treatment plans. With the scientific research and independent innovation of Chinese experts and professors, the definition-level image fusion optimization algorithm is becoming more and more perfect, but the pixel-level fusion has a large amount of data, complex calculations, and a relatively long time, which requires a system that can quickly process information.
- (2)Feature-level medical image fusion
Feature-level image fusion uses the eigenvalues of the low-level matrix to represent most of the information content of the image. This method is to obtain capillaries, human organs, diseases and other features from multi-modal medical images with demand and use value, and performs classification analysis and information content selection for these features, and completes the reasonable reduction of the source image information content. Reduce calculation time and improve combination efficiency. Feature-level medical image fusion is the acquisition or segmentation of the key features of the entire target of interest. It does not know the features, physiological information content and indoor spatial location of the overall target, which increases the complexity of the collection or segmentation optimization algorithm. For the difficulty coefficient, a more reasonable compound optimization algorithm must be used for feature testing to improve the accuracy of acquisition. In the whole process of feature-level medical image fusion, the feature information content with use value is selected, and the original record in the source image is rarely displayed, but the loss of information content cannot be prevented.
- (3)Decision-level medical image fusion
The medical image fusion of the management decision-making layer does not require strict registration of the source image, and is the topmost layer of the combination layer. This combined method makes basic management decisions for the same overall target institution, internal organs or diseases, and solves them according to the combined criteria and index weight values, and combines this basic ruling to get the best overall final diagnosis management decision. This is a highly coordinated management decision, the whole process of collaborative management decision-making ability, high efficiency, strong anti-interference ability, and strong adaptability. However, the basic diagnosis preparation processing cost of image fusion at the management decision level is high, which leads to the loss of part of the information content in the source image.
In general, the three levels of multimodal medical image fusion have similar connections, but they can also be combined with actual operations, which mean that in the entire process of specific applications, they can be flexibly combined with actual needs according to different necessary conditions. , To achieve a satisfactory combination of actual effects. Due to the strict management of registered pixel-level image fusion and the advantages of capturing information content at key points such as the edge and texture of the source image, it has been scientifically researched and discussed by many experts and scholars.
4.1.2. Pixel-level multi-modal medical image fusion
The origin of the multi-modal image, the location of the imaging equipment, the combination of sensing technology and the main purpose, etc., determine the complexity and diversity of the multi-modal image fusion optimization algorithm.
It is not difficult to see that compared with other multi-modal image fusion, the standardization of medical image fusion is relatively high. It must not only clearly preserve the information content of the source image, such as edge features, texture features, and overall target targets, but also colorfully reflect the key medical information contained in the multi-modal medical image, and must conform to the characteristics of the human eye. Therefore, the multi-modal medical image fusion can reduce image loss and false information while ensuring the quality of the combination, and prevent the clinical medical community from misjudging and misdiagnosing the patient’s condition.
Pixel-level image fusion has many characteristics, flexibility and diversity. Since the combinatorial optimization algorithm of multi-scale geometric analysis was clearly proposed, it has caused an uproar in academia. Its wide application range and retention of the characteristics of the spectrometer have been recognized by the joint industry. Multi-scale geometric figure analysis combination optimization algorithm is a definition-level combination optimization algorithm belonging to the conversion domain.
- (1)Medical image fusion algorithm in spatial domain
The medical image fusion optimization algorithm in the indoor space field is to immediately perform actual calculations on the sharpness of the source image that has been strictly registered, and apply a special combination optimization algorithm for fusion. Common combination methods include: weighted average optimization algorithm, principal component analysis method, color space fusion algorithm, intelligent domain fusion method, etc.
- a.Weighted average algorithm
The weighted average optimization algorithm is a very simple and most basic resolution-level indoor spatial domain combination optimization algorithm. It is a weighted average optimization algorithm that adjusts the weight value of two or more medical images. The optimization algorithm is fast and convenient in calculation, short in operation time, and strong in practicability. However, it reduces the frequency stability of the combined result image, averages the characteristic information content of the source image, such as edges, lines, and contours, and reduces the saturation of the combined result image, indicating that the features are insignificant. This greatly reduces the overall quality of the resulting image, jeopardizing the resolution and progress of the next step.
- b.Principal component analysis (PCA)
In the entire process of specific application of the weighted average optimization algorithm, due to the single weight index, the ideal combination of actual effects cannot be obtained. In order to better solve this problem, the principal component analysis method is clearly proposed.
- c.Color space fusion algorithm
Color medical images include PET, SPECT, etc., which play a key role in the display of human body system metabolism, but this color effect has low spatial resolution and cannot accurately reflect the actual characteristics of internal organs or diseases. The above must be combined with high-resolution anatomical images on the screen. Color space conversion is to project the multi-modal medical image into another color safe channel suitable for human eye characteristics, and then use the weight feature in this hue safe channel to combine with the corresponding medical anatomical image to achieve special countermeasures, and finally select the matching The color is inversely transformed, and the digital image of the final fusion result is obtained. IHS (Intensity-Hue-Saturation) entity model and RGB (Red-Green-Blue) entity model are typical color channel models. Among them, the IHS entity model is a safe channel for image fusion, consisting of three components: chroma (Intensity), hue (Hue) and contrast (Saturation).
- d.Smart domain method
With the continuous development trend of the artificial intelligence technology industry, multi-modal medical image fusion has also applied intelligent domain combination optimization algorithms, and has achieved excellent combined practical effects, including: neural network algorithms, Bayesian inference, DS direct evidence theory Methods and so on.
- (2)Medical image fusion algorithm in transform domain
With the continuous development of scientific and technological progress, researchers have been independently exploring and innovating optimization algorithms for multi-modal image fusion. Analysis of several common indoor spatial domain optimization algorithms have their own shortcomings: the weighted average optimization algorithm trivializes the characteristics of the image, reduces the saturation of the combined result image, and does not highlight the relevant key information content; the PCA optimization algorithm for the combined image The relevance requirements of the medium feature information and content are relatively high, and black and white can be distinguished. If the two multimodal medical images are very different, it is difficult to obtain a satisfactory combination of realistic effects. The color space conversion can easily cause the performance of the spectrometer to decrease and change the characteristics of the spectrometer. Therefore, once the multi-scale geometric transformation was proposed, it attracted great attention from the image fusion community. The conversion domain image fusion can not only preserve the characteristics of the spectrometer well, but also obtain excellent image quality in terms of information entropy (IE), edge information content evaluation factor (QABF), standard deviation (SD) and spatial frequency (SF) And actual results.
4.2. Algorithm model
The medical image fusion optimization algorithm in the field of indoor space is to immediately perform actual calculations on the pixels of the source image that have been strictly registered, and apply a special combination optimization algorithm for fusion. Common combination methods include: weighted average optimization algorithm, principal component analysis (PCA), color space combination optimization algorithm, intelligent domain combination method, etc. In these methods, data from medical sensors are often considered to have a higher weight because they can provide physiological information that is closely related to disease or health status. Through comprehensive analysis and processing of physiological signal data and image data, more accurate and reliable medical image fusion results can be obtained. Therefore, the medical image fusion of medical sensors in the field of indoor space provides rich physiological signal data, provides more information for the optimization algorithm, and helps the algorithm make more scientific and accurate decisions in the fusion process. The application of medical sensor makes the application of medical image fusion in the field of indoor space more efficient and reliable, and provides strong support for the diagnosis and analysis of medical images.
4.3. Evaluation system
Different drug combination optimization algorithms will obtain different combination image quality and results in the entire process of specific application. The image quality evaluation standard is one of the key index values for evaluating combination optimization algorithms and improved combination optimization algorithms. The current image quality review specifications are aimed at some special industries, and it is impossible to quantitatively analyze and describe image quality from a human subjective point of view. However, in the medical image fusion review specification, there is no clear description of a detailed evaluation method describing the human visual recognition system and its mental activity description, which lacks objectivity. Generally, medical image fusion reviews include: subjective reviews and objective reviews.
People’s evaluation of image quality is usually based on subjective evaluation, supplemented by objective evaluation. Subjective comments are based on human observation and combined with the professional knowledge of imaging medicine to immediately determine the comments, analysis and consideration of the combined result image. Observers must stand from a different perspective, make judgments based on their own work experience and their own evaluation methods, and usually must also combine other people’s views and suggestions on the image. The subjective evaluation method has its own advantages:
- (1)The observation is simple and fast, and the evaluation results can be directly obtained;
- (2)The doctor can purposefully observe the organs or lesions that need to be understood, and ignore other irrelevant image information;
- (3)The observer can distinguish whether the combined information content is missing, whether the edges of the overall target in the image are clear, and whether there is a mosaic effect.
In clinical medicine, subjective evaluation is also affected by many factors:
- (1)Physicians must use their own subjective work experience data to analyze the overall target of the disease and the information content of nearby capillary structures in multimodal medical images. Under normal circumstances, it is impossible to obtain accurate evaluation results without long-term accumulation;
- (2)To a certain extent, rely on the doctor’s comprehensive experience, speculative judgment and imagination;
- (3)Observation requires professional knowledge and image observation ability;
- (4)Observing requires good psychological quality.
Because subjective reviews are harmed by many factors to distinguish results, objective reviews rely on statistical analysis based on image data information to quantitatively analyze and smoothly analyze related or comparative features in the results. It is universal and will not damage subjectivity and psychological state due to human factors, so as to eliminate subjective variability. Perform quantitative data analysis on the combined result image, and annotate the results with visual values.
5. Conclusion
Based on the technology of medical sensor and artificial intelligence, this paper proposes a kind of digital holographic technology in the field of medical image fusion. By combining medical image information with digital holographic technology, 3D visualization and depth perception of medical images can be realized, so as to provide more comprehensive and accurate medical diagnosis and treatment programs. Artificial intelligence has made significant progress in medical image analysis, being able to automatically identify and extract important information from medical images. The sensor technology can collect the physiological parameters and physical signs of patients in real time. The combination of these two technologies can achieve comprehensive monitoring and analysis of patients’ conditions. Digital holographic technology can transform medical images into three-dimensional holograms, allowing doctors to more intuitively observe and analyze patients’ conditions. Compared with the traditional two-dimensional image, digital hologram can provide more information, including depth information and texture information. This allows doctors to make more accurate diagnoses and surgical planning. Artificial intelligence digital holographic technology based on medical sensor has a broad application prospect in the field of medical image fusion. By combining sensor technology, artificial intelligence technology and digital holographic technology, the comprehensive analysis and diagnosis of medical images can be realized, and the medical level and treatment effect of patients can be improved.
References
Mixed reality holograms for heart surgery planning: first user experience in congenital heart disease
Epicardial echocardiography in pediatric and congenital heart surgery
Three-dimensional computer-assisted surgical planning and manufacturing in complex mandibular reconstruction
