Tamper Localisation Using Quantum Fourier Transform Signatures for Medical Image Authentication

Abstract—The Internet of Medical Things, smart healthcare systems are becoming ubiquitous in our daily lives. Patients, doctors, and other medical personnel rely on the safe and efficient storage, transmission, and analysis of electronic health records , medical images such as scanning reports X-rays for successful treatment and management of different ailments. A Multi watermarking method is proposed for medical images based on quantum random walk and optimization algorithm. A logo image is used to verifying medical image integrity is embedding in region of interest and text data are embedded in the region of non interest to conceal private hospital and patient informations. This method improves the authenticity and integrity of medical images. A number of experiments are conducted to validate the security ,robustness and capacity of the proposed multi watermarking scheme. Keywords—medical image, quantum random walk, optimization algorithm, Internet of Medical Things

In telemedicine it is a common practice to exchange medical images between hospitals located at distant places through unsecure networks like internet. During this transfer tampers may be introduced intentionally or unintentionally into medical images. In this circumstances authentication and integrity of medical images using watermarking has become a very popular area of research. In this paper we are proposing a Region of Interest (ROI) and Least Significant Bit (LSB) based fragile watermarking technique for tamper detection and recovery of medical images. At first, medical image is divided into ROI and RONI. Later, authentication information is inserted into ROI and recovery information into RONI. To increase embedding capacity in ROI, authentication information is compressed using Run Length Encoding (RLE) technique before inserting into ROI. Experimental results reveal that any tampering to ROI of medical image is identified and recovered without any loss.

Lupus Erythematosus, notably its systemic form (SLE), is one of the most difficult-to-diagnose autoimmune conditions owing to variable symptomatology, multi-organ involvement, and common delays in diagnosis. Given that traditional diagnostic methods are clinically grounded, these processes often do not have the speed, scalability, and precision offered by modern healthcare systems. The medical community has increasingly looked to data-driven and artificial intelligence (AI) methodologies in response to these gaps. In this paper, we provide an extensive survey on the existing methodologies for lupus detection and management, i.e., machine learning (ML), deep learning, hybrid model, and explainable AI (XAI) frameworks. The survey starts by exploring data collection methods, ranging from clinical databases to wearable devices and the integration of electronic health records (EHR), emphasizing the importance of carefully preprocessing data and selecting relevant features. Next, we provide a high-level summary of the methodological landscape— the collection of statistical models, support vector machines, neural networks, ensemble approaches, time-series methods, and others—focusing on the basic principles of these methods, and their relative utility to the lupus research paradigm. At the heart of this debate lies the question of the performance metrics (e.g., accuracy, F1-score, specificity, AUC) that gauge the clinical deployment potential of a model. The paper also notes real-world applications outside of algorithms including early diagnosis, flare prediction, personalized treatment planning, and patient self-monitoring through mobile health platforms. While these advances are encouraging, challenges still exist, including data heterogeneity, class imbalance, interpretability issues, and integration challenges in clinical workflows. In bringing these strands together, we also suggest future avenues for research, especially in the domains of federated learning, responsible AI deployment, and multi-modal data integration. This survey paper aims to reconcile the technical complexities of smart health with the practical needs in the healthcare domain by acting as a reference for clinicians, researchers and other developers hoping to leverage intelligence, bias-free human-in-the-loop system in managing lupus erythematosus efficiently and effectively towards the improvement of individuals suffering from such diseases.

Medical imaging and information management systems require transmission of medical images over the Internet. Many image watermarking techniques have been proposed in recent years to ensure the integrity and authenticity of medical images transferred over insecure networks. In this work, we propose a new medical image watermarking technique to detect tampered regions on medical images with finer accuracy by authenticating 4×4 blocks and without restricting region of interest (ROI) size. The proposed method can mark a 4×4 pixel block if it has even one tampered pixel, while similar methods (which have no ROI size restriction) mark 8×8, 16×16, and 40×40 pixel blocks. Modified difference expansion (MDE) and least significant bit (LSB) embedding techniques are used together first in the literature by the method to embed authentication bits into corresponding blocks. The method uses a small 4×4 window to mark the tampered region. Experimental results show that the proposed method detects tampered regions on medical images with high accuracy and can be used by all medical image modalities. The results also indicate that the method has finer accuracy and no ROI size restriction compared to similar works reported in the literature.

Medical images are transmitted to other remote places through the E-healthcare system. The protection of medical images is very crucial. Medical images need to be protected against any modification which the attacker may do through the unsecured channel. It is necessary for inspecting the integrity of the ROI (Region of Interest) of the received medical image prior to the adoption of any diagnostic decision to avoid a misdiagnosis. This paper aimed to confirm the integrity and authenticity of medical images by combining the hash signature with the watermarking technique based on the frequency domains. At first, the medical image is divided into ROI and RONI (Region of Non-Interest). The SLT and DCT transformations are employed together to extract the essential features set from ROI and then combined with EPR (Electronic Patient Registration) to produce a watermark, and then it is encoded by a chaotic map with a secret key to provide a signature. On the other hand, the RONI is mapping into randomly subblocks based on the linear system, and then the Signature is embedded in the chosen block using DWT. The hash code of the recovered image and EPR will be compared with the extracted watermark for integrity and authenticity. The experimental results demonstrate the watermark’s robustness against many of the more aggressive and geometric signal processing distortions without affecting the quality of the original medical image. In particular, compared with state-of-the-art technology, the proposed algorithm improves the average NC value larger than 0.90 under all types of attacks.

This article discusses an approach for both authentication of medical images and confidentiality for the related textual data in an online medical application paradigm. The image authentication is achieved in a soft manner through a feature-based digital signature while the confidentiality of the related patient information is achieved through reversible data hiding. The selected features are robust towards geometric transformations, while fragile towards texture alterations that are characteristic of medical images. The processing scheme is done in a block by block basis to permit the localization of tampered image’s regions. The effectiveness of the scheme, proven through experiments on a sample of medical images, enables us to argue that implementing mechanisms lying on this approach will help to maintain personal patient privacy and medical image integrity.

Telemedicine has gained popularity in recent years. Medical images can be transferred over the Internet to enable the telediagnosis between medical staffs and to make the patient's history accessible to medical staff from anywhere. Therefore, integrity protection of the medical image is a serious concern due to the broadcast nature of the Internet. Some watermarking techniques are proposed to control the integrity of medical images. However, they require embedding of extra information (watermark) into image before transmission. It decreases visual quality of the medical image and can cause false diagnosis. The proposed method uses passive image authentication mechanism to detect the tampered regions on medical images. Structural texture information is obtained from the medical image by using local binary pattern rotation invariant (LBPROT) to make the keypoint extraction techniques more successful. Keypoints on the texture image are obtained with scale invariant feature transform (SIFT). Tampered regions are detected by the method by matching the keypoints. The method improves the keypoint-based passive image authentication mechanism (they do not detect tampering when the smooth region is used for covering an object) by using LBPROT before keypoint extraction because smooth regions also have texture information. Experimental results show that the method detects tampered regions on the medical images even if the forged image has undergone some attacks (Gaussian blurring/additive white Gaussian noise) or the forged regions are scaled/rotated before pasting.

Enhanced security framework for medical data embedding based on octonionic steganographic transforms and FPGA-accelerated integrity verification

In telemedicine, the authenticity and integrity of medical images must be safeguarded. Copyright protection is provided through robust medical image watermarking (MIW) methods, and the original pictures may be retrieved at the receiver's end. But existing algorithms have limits in terms of balancing the tradeoff between robustness, imperceptibility, and embedded capacity. Aside from that, most MIW algorithms aren't built for color images. This article proposes a novel MIW technique based on the redundant discrete wavelet transform (RDWT) with singular value decomposition (SVD) to increase their performance in preserving medical color picture information. First and foremost, the RDWT -SVD is a reliable solution as compared to the conventional DWT. Second, modifying the wavelet domain coefficient ensures that integer values in the spatial domain change and that the watermarking process is reversible. Finally, the embedding approach makes full advantage of the original image's features and watermarking. The simulation results showed that the proposed method decreases the amount of original picture change and improves imperceptibility as compared to the conventional approaches.

In recent times, technological advancements have remarkably improved picture archiving and communication system(PACS) capabilities beyond their conventional use in radiology departments. Researchers and instructors have started employing PACS functionalities to improve medical research processes, promote interdisciplinary collaborations, and facilitate learning. To illustrate this point further, the PACS enables researchers to handle and analyze huge amounts of imaging data with superior precision and speed, supporting innovative studies in areas like disease progression, treatment outcomes, and imaging modalities. Moreover, a PACSintegrated with artificial intelligence (AI) algorithms leads to significant improvements in image processing, diagnostic accuracy, and personalized treatment, thus marking a new approach to medical imaging. The PACSsupported by AI is mostly transformative since they allow for improved early disease detection capabilities as well as automated image processing and decision assistance, which increase diagnostic accuracy and clinical outcomes. Such systems can rapidly process large quantities of visual data with an accuracy rate surpassing earlier endeavors. In medical research, however, combining PACSwith AI allows challenging imaging datasets to be examined, thereby making findings that were not previously possible. The capacity to combine imaging outcomes with clinical information is valuable for medical students and professionals in the field of education. They can access extensive medical image collections and case studies using PACS. This link is critical for teaching and learning as it allows students to interact with concrete events and improve their diagnostic accuracy in a controlled environment. This review discusses how the PACS affects educational courses and clinical outcomes based on the available literature. Our aim was not only to outline recent research or developments but also to present a comprehensive overview regarding the growing role played by PACS in the modern healthcare system and academics. Similarly, we look at the challenges and opportunities associated with the wide adoption of PACS, highlighting possible future areas of study or teaching methodologies. Issues such as data security, interoperability, and the need for defined protocols are included to give an exhaustive understanding of what PACS can and cannot do. Through this study, we stress PACS's revolutionary potential in advancing research methodology and educational practices, eventually contributing to enhanced patient care and knowledge dissemination in healthcare areas. The continual growth of PACS technology and its applications is expected to reshape the landscape of medical research and education, making it a vital component in the quest for medical excellence. By knowing the present trends and future potential, stakeholders in healthcare and education may better employ PACS to reach their objectives and boost overall results.

The modern healthcare systems consider the secure transmission and storage of medical images as vital because of the sensitivity of patient data and the strict privacy regulations. This paper presents a secure and reversible RGB medical image scrambling and encryption scheme that aims to defend diagnostic images from unauthorized access, while, guaranteeing the integrity of the data. A 512- bit master key is generated through the memory-hard Argon2i algorithm which makes the system extremely resistant to brute-force and dictionary attacks. Domain-separated subkeys are generated using master key via AES in counter (CTR) mode. These subkeys are utilized at confusion and diffusion stages of the encryption and decryption process. The image confusion is performed with the use of a Fisher-Yates permutation which is controlled by cryptographically secure keystreams thereby disrupting spatial pixel relationships. The diffusion is done at both single colour channel and total colour channel levels utilizing a feedback-based mechanism that combines XOR operation, modular addition, and bit-level permutation. Both confusion and diffusion process ensures a strong avalanche effect across the entire image. Performance analysis shows that the proposed method provides an effective balance between security and computational efficiency, making it suitable for secure storage and transmission of medical images in telemedicine and healthcare information systems.

---------------------------------------------------------------------***--------------------------------------------------------------------- Abstract– Cardiovascular diseases remain one of the leading causes of mortality worldwide, making early and reliable risk prediction a critical requirement in modern healthcare systems. Traditional diagnostic approaches often rely on either clinical parameters or medical imaging in isolation, which may limit the predictiveness of the disease. This project aims to build a multimodal heart disease prediction framework that integrates clinical tabular data and echocardiography-based image information to estimate the probability of heart disease in a robust way. The tabular dataset includes demographics and clinical risk factors such as age, sex, chest pain type, blood pressure, cholesterol, electrocardiographic findings, and exercise-related parameters. Multiple classical machine learning models were trained and evaluated, including Logistic Regression, Support Vector Machine, Random Forest, Decision Tree, and XGBoost. Among all these, XGBoost achieved the best performance, with a test accuracy of 89.13% and an F1-score of 0.9029, demonstrating strong predictive capability. For the imaging modality, an echocardiography video-based model was trained using the EchoNet Dynamic dataset, where cardiac abnormality was determined based on an ejection fraction threshold of 50%. The image model achieved a ROC-AUC of 0.915, indicating excellent discriminative performance. Threshold optimization was performed to improve clinical sensitivity, increasing recall from 66.77% to 72.70% and reducing false negative cases. A late fusion approach using weighted averaging of predicted probabilities was adopted to combine both modalities, leveraging their complementary strengths. The system outputs a probabilistic risk score rather than a definitive diagnosis, making it suitable as a clinical decision support tool. Future enhancements include the integration of a RAG-based assistant for user guidance and the adoption of explainable AI techniques to improve transparency and interpretability. Key Words: Heart Disease Prediction, Multimodal Learning, Echocardiography, Machine Learning, Deep Learning, Late Fusion, Clinical Decision Support, Prediction Threshold

Study question Deep-learning algorithms are known to be non-robust: can the variability and inconsistency of AI algorithms be reduced in embryo selection? Summary answer We reduced the variability of algorithms (measured on different tasks like rotations and brightness changes) by 86% while preserving their quality. What is known already Deep-learning methods are generally known to be non-robust, i.e., decisions change with even slight modification of input data. Current solutions for embryo scoring are not robust - for example rotating the input image results in a different score in most solutions on the market. Despite this fact and expressed concerns of embryologists, there are no other publications focusing on the problem of variance in AI solutions used in IVF. Most of the publications measure accuracy, sensitivity, specificity, and ROC AUC; there are no variance metrics. Study design, size, duration The data-set was collected within multiple clinics using various devices. It contains 34,821 embryos (4,510 were transferred with known pregnancy results), represented by time-lapse videos or images. This gives 3,290,481 frames of embryos at various maturity levels. From the data-set 925 randomly selected embryos were chosen as a test set. The frames were modified by methods that are not supposed to change the results of the algorithm. We measured the variability of the scores given by our algorithm. Participants/materials, setting, methods We have considered seven different modifications of images that should not influence embryo scoring: • Rotations (10 different angles); • Brightness and Contrast modifications; • Substitutions of Frames (from time-lapse monitoring taken from a 2 hours interval); • Blur (Generalised Normal filter); • Gaussian Noise; • Gaussian Blur; • Sharpening. We used several techniques to reduce variance of our deep neural network model (architecture commonly used for embryo selection): • Ensemble (of different models in cross validation); • Test time augmentation (TTA); • Robust training. Main results and the role of chance In order to measure the variance we have used the following method. First, the scores are stretched to the standard uniform distribution. In other words we look in which percentile the score lies. This way the range of the scores are normalised thus the variance can be compared. Second, we train the EMBROAID model on the augmented data that includes all the above modifications. Third, we compute the variance of the normalised scores on the test set. The mean variance dropped by 86% (0.0055 to 0.0008) across all measured input modifications. The individual drops in the variance on measured input modifications: Rotations: 77% (0.009 -> 0.002), Brightness and Contrast: 81% (0.0036 -> 0.0007), Substitution of Frames: 76% (0.0076 -> 0.0019), Blur 94% (0.012 -> 0.0008), Gaussian Noise: 96% (0.0049 -> 0.0002), Gaussian Blur: 95% (0.0052 -> 0.0003), Sharpening: 77% (0.0015 -> 0.0003). The significance was tested with Wilcoxon Rank Sum Test giving the p-value < 0.01 on all input modifications. Finally, we stress that these results were obtained without any loss in the ROC AUC metric. We have tested the algorithm both on the original test-set. Both models achieved an ROC AUC of 0.66 (CI 0.63-0.69) on both test-sets. Limitations, reasons for caution Further work needs to be done to extend the set of possible augmentations of data. Wider implications of the findings Increased reliability of AI scoring algorithms for embryo selection. It is possible to obtain consistent results over a wide range of data modifications. Trial registration number not applicable

Multimodal fake news detection requires strong models with the ability to perform well under real-world data imperfections and class imbalance. This paper introduces a CLIP-based approach validated on a stratified fraction of the Fakeddit dataset (50,000 samples, 60.7% real vs. 39.3% fake news) with state-of-the-art performance using a fault-tolerant pipeline with BART text summarization, URL-based image loading with fallback strategies, and class-imbalance reduction. Our model achieves 91.24% test accuracy and 0.9715 ROC AUC, with excellent per-class performance: 93% F1-score for fake news (94% precision, 91% recall) and 89% F1-score for real news (87% precision, 91% recall), showing balanced effectiveness in spite of the 1.54:1 class imbalance. The confusion matrix shows strong minority-class detection, correctly identifying 2,851/3,136 fake samples (90.9% recall) while keeping 4,430/4,844 real news detections (91.5% precision). Major innovations are dynamic truncation of text (23% of posts abridged through BART-large-CNN), adaptive attention gates for modality reliability weighting, and focal loss optimization, reducing training epochs by 43% compared to baseline CLIP fine-tuning. Batch-wise evaluation results in uniform performance (87.9-94.9% per-batch accuracy) with a final test loss of 0.2105, achieving better performance than current multimodal architectures by 6.8% F1-score. The system's computational efficiency-40% less GPU memory usage than typical CLIP implementations-combined with the robustness of the system to 6.2% inherent image URL failures and synthetic corruption tests (only 2.1% performance loss at 20% failure rates), makes it a feasible solution for real-time social media applications. This piece ensures that strategic architectural treat-ments can surpass data scarcity and imbalance issues, gaining high reliability without full-scale dataset requirements.

This article discusses the practical aspects of applying logistic regression for binary data classification. Logistic regression determines the probability of an object belonging to one of two classes. This probability is calculated with the help of a sigmoid function, the argument of which is a linear convolution of the feature vector of the object with the weighting coefficients obtained during the minimization of the logarithmic loss function. Predicted class labels are determined by comparing the calculated probability with a given threshold value. The logistic regression study was performed using the computer simulation method. For this, a software complex was developed, the work of which reproduces the main stages of logistic regression: preparation of input data, training, testing with determination of quality metrics of binary classification, application of the logistic regression method for data classification in practice. The paper examines the effect of overlapping and imbalance of classes in the input data set on the efficiency of binary classification. The overlapping of classes is modeled by the formation of input data based on two shifted relative to each other density functions of the normal distribution of random variables. Class imbalance is simulated by the probability of switching between these features. It is shown that when the distance between the mathematical expectations of the density functions of the normal distribution decreases or when the dispersion of random variables increases, the overlapping of relevant classes increases, which leads to an increase in the number of objects that the classifier can assign to one or another class. Approaching the probability of switching between the distribution functions of random variables to the extreme values of the unit interval leads to an increase in class imbalance, which is manifested in an increase in the number of elements of the input data set labeled with the label of the same class. It has been experimentally confirmed that the AUC ROC metric, popular in binary classification problems, is dependent on the degree of class overlap and relatively resistant to class imbalance.