Medical Image Data Research Articles

A Review of 3D Printing and Surgical Planning

Three-Dimensional Printing (3DP), also known as rapid prototyping or additive manufacturing, has revolutionized the field of medicine, particularly in preoperative planning. By transforming intricate Two-Dimensional (2D) medical imaging data into tangible, patient-specific Three-Dimensional (3D) models, 3DP enhances surgical precision and patient care. This technology facilitates detailed preoperative visualization, enabling surgeons to interact with accurate anatomical models, plan complex procedures, and anticipate potential challenges. The evolution of 3DP, driven by advances in Fused Deposition Modeling (FDM), Selective Laser Sintering (SLS), and Stereolithography (SLA), has democratized access to affordable and sophisticated 3D printers, spurring innovation in medical applications. In this study, a comprehensive review of the literature on the integration of 3DP technology into preoperative planning is conducted. Applications across various surgical disciplines, including cardiovascular, orthopedic, craniofacial, neurosurgical, oncological, transplant, and pediatric surgeries are examined. Benefits of 3DP in providing tactile feedback, creating patient-specific surgical guides and implants, and enhancing surgical simulation and training are evaluated. Additionally, challenges associated with the high costs, time-intensive processes, and material limitations of 3DP are assessed. Findings demonstrate that 3DP significantly reduces operative times, improves surgical outcomes, and lowers complication rates. The study highlights the transformative impact of 3DP on surgical practice and underscores its potential for future advancements. With ongoing research and innovation in materials science, bioprinting, artificial intelligence, and augmented reality, 3DP is poised to further revolutionize preoperative planning, offering more personalized, precise, and patient-centered healthcare solutions.

A journey from omics to clinicomics in solid cancers: Success stories and challenges.

The word 'cancer' encompasses a heterogenous group of distinct disease types characterized by a spectrum of pathological features, genetic alterations and response to therapies. According to the World Health Organization, cancer is the second leading cause of death worldwide, responsible for one in six deaths and hence imposes a significant burden on global healthcare systems. High-throughput omics technologies combined with advanced imaging tools, have revolutionized our ability to interrogate the molecular landscape of tumors and has provided unprecedented understanding of the disease. Yet, there is a gap between basic research discoveries and their translation into clinically meaningful therapies for improving patient care. To bridge this gap, there is a need to analyse the vast amounts of high dimensional datasets from multi-omics platforms. The integration of multi-omics data with clinical information like patient history, histological examination and imaging has led to the novel concept of clinicomics and may expedite the bench-to-bedside transition in cancer. The journey from omics to clinicomics has gained momentum with development of radiomics which involves extracting quantitative features from medical imaging data with the help of deep learning and artificial intelligence (AI) tools. These features capture detailed information about the tumor's shape, texture, intensity, and spatial distribution. Together, the related fields of multiomics, translational bioinformatics, radiomics and clinicomics may provide evidence-based recommendations tailored to the individual cancer patient's molecular profile and clinical characteristics. In this chapter, we summarize multiomics studies in solid cancers with a specific focus on breast cancer. We also review machine learning and AI based algorithms and their use in cancer diagnosis, subtyping, prognosis and predicting treatment resistance and relapse.

Mine Your Own Anatomy: Revisiting Medical Image Segmentation With Extremely Limited Labels.

Recent studies on contrastive learning have achieved remarkable performance solely by leveraging few labels in the context of medical image segmentation. Existing methods mainly focus on instance discrimination and invariant mapping (i.e., pulling positive samples closer and negative samples apart in the feature space). However, they face three common pitfalls: (1) tailness: medical image data usually follows an implicit long-tail class distribution. Blindly leveraging all pixels in training hence can lead to the data imbalance issues, and cause deteriorated performance; (2) consistency: it remains unclear whether a segmentation model has learned meaningful and yet consistent anatomical features due to the intra-class variations between different anatomical features; and (3) diversity: the intra-slice correlations within the entire dataset have received significantly less attention. This motivates us to seek a principled approach for strategically making use of the dataset itself to discover similar yet distinct samples from different anatomical views. In this paper, we introduce a novel semi-supervised 2D medical image segmentation framework termed Mine yOur owNAnatomy (MONA), and make three contributions. First, prior work argues that every pixel equally matters to the model training; we observe empirically that this alone is unlikely to define meaningful anatomical features, mainly due to lacking the supervision signal. We show two simple solutions towards learning invariances-through the use of stronger data augmentations and nearest neighbors. Second, we construct a set of objectives that encourage the model to be capable of decomposing medical images into a collection of anatomical features in an unsupervised manner. Lastly, we both empirically and theoretically, demonstrate the efficacy of our MONA on three benchmark datasets with different labeled settings, achieving new state-of-the-art under different labeled semi-supervised settings. MONA makes minimal assumptions on domain expertise, and hence constitutes a practical and versatile solution in medical image analysis. We provide the PyTorch-like pseudo-code in supplementary.

Personalised in silico biomechanical modelling towards the optimisation of high dose-rate brachytherapy planning and treatment against prostate cancer.

High dose-rate brachytherapy presents a promising therapeutic avenue for prostate cancer management, involving the temporary implantation of catheters which deliver radioactive sources to the cancerous site. However, as catheters puncture and penetrate the prostate, tissue deformation is evident which may affect the accuracy and efficiency of the treatment. In this work, a data-driven in silico modelling procedure is proposed to simulate brachytherapy while accounting for prostate biomechanics. Comprehensive magnetic resonance and transrectal ultrasound images acquired prior, during and post brachytherapy are employed for model personalisation, while the therapeutic procedure is simulated via sequential insertion of multiple catheters in the prostate gland. The medical imaging data are also employed for model evaluation, thus, demonstrating the potential of the proposed in silico procedure to be utilised pre- and intra-operatively in the clinical setting.

An explainable ensemble approach for advanced brain tumor classification applying Dual-GAN mechanism and feature extraction techniques over highly imbalanced data.

Brain tumors are one of the leading diseases imposing a huge morbidity rate across the world every year. Classifying brain tumors accurately plays a crucial role in clinical diagnosis and improves the overall healthcare process. ML techniques have shown promise in accurately classifying brain tumors based on medical imaging data such as MRI scans. These techniques aid in detecting and planning treatment early, improving patient outcomes. However, medical image datasets are frequently affected by a significant class imbalance, especially when benign tumors outnumber malignant tumors in number. This study presents an explainable ensemble-based pipeline for brain tumor classification that integrates a Dual-GAN mechanism with feature extraction techniques, specifically designed for highly imbalanced data. This Dual-GAN mechanism facilitates the generation of synthetic minority class samples, addressing the class imbalance issue without compromising the original quality of the data. Additionally, the integration of different feature extraction methods facilitates capturing precise and informative features. This study proposes a novel deep ensemble feature extraction (DeepEFE) framework that surpasses other benchmark ML and deep learning models with an accuracy of 98.15%. This study focuses on achieving high classification accuracy while prioritizing stable performance. By incorporating Grad-CAM, it enhances the transparency and interpretability of the overall classification process. This research identifies the most relevant and contributing parts of the input images toward accurate outcomes enhancing the reliability of the proposed pipeline. The significantly improved Precision, Sensitivity and F1-Score demonstrate the effectiveness of the proposed mechanism in handling class imbalance and improving the overall accuracy. Furthermore, the integration of explainability enhances the transparency of the classification process to establish a reliable model for brain tumor classification, encouraging their adoption in clinical practice promoting trust in decision-making processes.

Heart Disease Prediction Using Machine Learning

In the medical field, diagnosing heart disease is one of the most challenging tasks. It relies on analyzing large sets of clinical and pathological data, making the process complex. As a result, there has been a growing interest among researchers and clinical professionals in developing efficient and accurate methods for heart disease prediction. Early diagnosis of heart disease is crucial due to its potential severity, as timely intervention can significantly impact patient outcomes. Heart disease is a leading cause of death globally, underscoring the importance of predicting it at an early stage. In recent years, machine learning has emerged as a reliable tool in the medical domain, offering valuable support in disease prediction when provided with appropriate training and testing data. A range of machine learning algorithms, encompassing decision trees, support vector machines, random forests, neural networks, and ensemble methods, have been employed to analyze varied datasets containing demographic information, clinical parameters, and medical imaging data. To improve prediction accuracy and interpretability, techniques for feature selection and model Optimizations have been utilized. The primary objective of this study is to explore various prediction models for heart disease and identify key features using the Random Forest algorithm. Random Forest is a supervised machine learning algorithm known for its high accuracy compared to other methods such as logistic regression. Utilizing the Random Forest algorithm, we aim to predict whether an individual has heart disease or not

Prevalence of COVID-19 neurological manifestations in patients referred to the Golestan Hospital in Ahvaz between March 2020 to the March 2023.

Due to the abundance of ACE2 receptors in nervous system cells, the SARS-CoV-2 virus can cause damage to this system. This study aims to examine the prevalence of neurological symptoms in COVID-19 patients. In this cross-sectional observational study, 75 COVID-19 positive patients admitted to Golestan Hospital's neurology department in Ahvaz, Iran, from March 2020 to March 2023, were investigated. Neurological clinical symptoms were categorized into three groups: central nervous system, peripheral, and muscular symptoms. The relevant information was collected from patient files, including medical history, imaging data, and laboratory test results. Statistical analysis was performed using SPSS software, employing the rank-biserial correlation coefficient (r), Mann-Whitney U tests, Phi correlation, Cramer's V, and Kendall's Tau to evaluate the prevalence and significance of neurological symptoms. The most common clinical symptoms observed were hemiparesis, dysarthria, Central Facial Palsy (CFP), ataxia, and nausea, respectively. Among these symptoms, headaches (p = 0.001), seizures (p = 0.024), and nausea (p = 0.046) were found to be more prevalent in younger patients. Additionally, a significant relationship was identified between the level of serum Creatine phosphokinase (CPK) and seizures (p = 0.024), with lower levels observed in individuals with vomiting (p = 0.024), and higher levels observed in individuals with CFP (p = 0.040). This study highlights that patients with COVID-19 may experience serious neurological symptoms. The clinical spectrum and range of neurological symptoms associated with COVID-19 were found to be diverse and extensive, emphasizing the importance of considering this infection as a potential cause of neurological disorders.

LUneXt: Simple and Efficient U-shaped Network Design for Medical Image Segmentation with Nonlinear Activation

Medical image segmentation has always been a challenging task. This paper proposes a new LUneXt medical image segmentation model based on the characteristics analysis of medical image data sets and testing of different nonlinear activation units. Both normalized activations for the original negative input image activation have good optimization capabilities for the tokenization module parameters proposed in the original UneXt model. Using different activation coefficients for different foreground and background areas has achieved better results. The experimental results of this paper on the Breast Ultrasound Images (BUSI) data set reached an intersection over union (IoU) value of 62.64%, a Dice value of 76.12%, and a single inference speed of 807.57[Formula: see text]ms. The experimental IoU value of the International Skin Imaging Collaboration (ISIC 2018) data set reached 82.95%, and the Dice value reached 90.50%. The single inference speed reached 842.58[Formula: see text]ms. The LUneXt model is more robust than other models. While improving model performance, it does not introduce higher computational complexity and does not have a major impact on the processing speed of a single image.

Genetic Architectures of Medical Images Revealed by Registration of Multiple Modalities

The advent of biobanks with vast quantities of medical imaging and paired genetic measurements creates huge opportunities for a new generation of genotype–phenotype association studies. However, disentangling biological signals from the many sources of bias and artifacts remains difficult. Using diverse medical images and time-series (ie, magnetic resonance imagings [MRIs], electrocardiograms [ECGs], and dual-energy X-ray absorptiometries [DXAs]), we show how registration, both spatial and temporal, guided by domain knowledge or learned de novo, helps uncover biological information. A multimodal autoencoder comparison framework quantifies and characterizes how registration affects the representations that unsupervised and self-supervised encoders learn. In this study we (1) train autoencoders before and after registration with nine diverse types of medical image, (2) demonstrate how neural network-based methods (VoxelMorph, DeepCycle, and DropFuse) can effectively learn registrations allowing for more flexible and efficient processing than is possible with hand-crafted registration techniques, and (3) conduct exhaustive phenotypic screening, comprised of millions of statistical tests, to quantify how registration affects the generalizability of learned representations. Genome- and phenome-wide association studies (GWAS and PheWAS) uncover significantly more associations with registered modality representations than with equivalently trained and sized representations learned from native coordinate spaces. Specifically, registered PheWAS yielded 61 more disease associations for ECGs, 53 more disease associations for cardiac MRIs, and 10 more disease associations for brain MRIs. Registration also yields significant increases in the coefficient of determination when regressing continuous phenotypes (eg, 0.36 ± 0.01 with ECGs and 0.11 ± 0.02 for DXA scans). Our findings reveal the crucial role registration plays in enhancing the characterization of physiological states across a broad range of medical imaging data types. Importantly, this finding extends to more flexible types of registration, such as the cross-modal and the circular mapping methods presented here.

A novel blockchain and fractional DCT based image storage and retrieval framework

Securing medical images are a challenging task for researchers as they are prone to cyber- attacks. A potential solution to this problem can be blockchain, which provide secure and unchangeable storage. Fractional Discrete Cosine Transform (frDCT) has the capability to minimizes the amount of data necessary for expressing an image in a secure way. This paper presents a novel framework for compressing as well as securely storing and retrieving medical images by extraction of feature maps from medical images by frDCT, followed by encoding and storing the feature map on decentralized cloud and linking them on blockchain using fractional angle α. The proposed framework provides numerous benefits such as improved data sharing and collaboration, enhanced security, compression as well as efficient retrieval and processing of medical image data which in turn decreases the computational energy requirement providing sustainability. The performance of proposed framework has been evaluated by employing image quality metric such as MSE, PSNR, SSIM and multi-SSIM by comparing it with correct and incorrect α values on six different values of frDCT parameter α.

Mathematical analysis of wavelet-based multi-image compression in medical diagnostics

Accurate diagnosis and well-informed treatment choices are only possible with the help of sophisticated medical imaging technology. The introduction of digital imaging methods has led to a dramatic rise in the volume of data associated with medical images, which in turn has led to difficulties in their storage, transmission, and administration. It is crucial to utilize effective image compression techniques to address these challenges without compromising the diagnostic integrity of the pictures. The wavelet transform has matured into a potent method for striking a good compromise between picture quality and file size reduction while compressing. The safe and efficient transfer of medical image data is a major concern in today’s healthcare settings. In this academic investigation, we investigate how wavelet transform-based techniques may be used to enhance medical picture compression. The proposed techniques optimize compression ratios while maintaining diagnostic picture quality by making use of the wavelet transform’s multi-resolution and frequency localization features. To address the unique challenges given by medical image collections, various iterations of the wavelet transform and compression techniques are investigated. Through a series of detailed tests involving several medical picture modalities, the effectiveness of these technologies is thoroughly evaluated, demonstrating their effectiveness in achieving significant data reduction without sacrificing clinical information.

Score-Based Counterfactual Generation for Interpretable Medical Image Classification and Lesion Localization.

Deep neural networks (DNNs) have immense potential for precise clinical decision-making in the field of biomedical imaging. However, accessing high-quality data is crucial for ensuring the high-performance of DNNs. Obtaining medical imaging data is often challenging in terms of both quantity and quality. To address these issues, we propose a score-based counterfactual generation (SCG) framework to create counterfactual images from latent space, to compensate for scarcity and imbalance of data. In addition, some uncertainties in external physical factors may introduce unnatural features and further affect the estimation of the true data distribution. Therefore, we integrated a learnable FuzzyBlock into the classifier of the proposed framework to manage these uncertainties. The proposed SCG framework can be applied to both classification and lesion localization tasks. The experimental results revealed a remarkable performance boost in classification tasks, achieving an average performance enhancement of 3-5% compared to previous state-of-the-art (SOTA) methods in interpretable lesion localization.

Combined use of 3D printing and mixed reality technology for neurosurgical training: getting ready for brain surgery.

Learning surgical skills is an essential part of neurosurgical training. Ideally, these skills are acquired to a sufficient extent in an ex vivo setting. The authors previously described an in vitro brain tumor model, consisting of a cadaveric animal brain injected with fluorescent agar-agar, for acquiring a wide range of basic neuro-oncological skills. This model focused on haptic skills such as safe tissue ablation technique and the training of fluorescence-based resection. As important didactical technologies such as mixed reality and 3D printing become more readily available, the authors developed a readily available training model that integrates the haptic aspects into a mixed reality setup. The anatomical structures of a brain tumor patient were segmented from medical imaging data to create a digital twin of the case. Bony structures were 3D printed and combined with the in vitro brain tumor model. The segmented structures were visualized in mixed reality headsets, and the congruence of the printed and the virtual objects allowed them to be spatially superimposed. In this way, users of the system were able to train on the entire treatment process from surgery planning to instrument preparation and execution of the surgery. Mixed reality visualization in the joint model facilitated model (patient) positioning as well as craniotomy and the extent of resection planning respecting case-dependent specifications. The advanced physical model allowed brain tumor surgery training including skin incision; craniotomy; dural opening; fluorescence-guided tumor resection; and dura, bone, and skin closure. Combining mixed reality visualization with the corresponding 3D printed physical hands-on model allowed advanced training of sequential brain tumor resection skills. Three-dimensional printing technology facilitates the production of a precise, reproducible, and worldwide accessible brain tumor surgery model. The described model for brain tumor resection advanced regarding important aspects of skills training for neurosurgical residents (e.g., locating the lesion, head position planning, skull trepanation, dura opening, tissue ablation techniques, fluorescence-guided resection, and closure). Mixed reality enriches the model with important structures that are difficult to model (e.g., vessels and fiber tracts) and advanced interaction concepts (e.g., craniotomy simulations). Finally, this concept demonstrates a bridging technology toward intraoperative application of mixed reality.

MedTransCluster: Transfer learning for deep medical image clustering

This work introduces the “MedTransCluster” framework, a novel approach to medical image clustering in chest radiography through the application of transfer learning, leveraging the capabilities of pre-trained deep learning models. Our evaluation encompassed a variety of neural networks, considering their adaptability to the nuances of medical imaging data. The study incorporated four renowned clustering algorithms and an expanded set of evaluation metrics, offering a comprehensive comparison and a refined analysis of these models’ ability to cluster complex diagnostic features. Notably, EfficientNetB0 coupled with DBSCAN clustering algorithm achieved a silhouette score of 0.924131, and ResNet152 with KMeans displayed a Calinski Harabasz score of 9655.213964, indicating their superior proficiency in capturing the intricacies of medical features. These results emphasize the critical importance of model refinement within the healthcare imaging sphere and underscore the potential of methodologies like MedTransCluster in enhancing diagnostic accuracy and patient outcomes.

Enhancing bra design for post-mastectomy patients: Incorporating MRI data and innovative textiles for optimal support and comfort

This study introduces an innovative approach to designing a customized bra to the specific needs of mastectomy patients, aiming to reduce the discrepancies in breast displacement and consequently alleviate associated pain. Through mastectomy patient interviews, recognizing the limitations of current mastectomy bras, which often fail to accommodate the unique body shapes and needs of breast cancer survivors, this research leverages Magnetic Resonance Imaging (MRI) and Three-Dimensional Surface Imaging (3D-SI) to develop a custom-fit bra prototype. The methodology involves the analysis of body shape changes post-mastectomy and the development of a prototype bra that effectively supports prostheses while minimizing common issues such as displacement, discomfort, and skin irritation. The results showcased a pronounced reduction in breast movement in all directions with the prototype bra, further enhanced using a novel cooling textile, improving the wearer’s overall comfort. Moreover, our investigation delved into the potential of integrating MRI and 3D-SI anthropometric data in bra design. By comparing two Prototype bras with differing dimensions but identical design features, we discerned marked differences in their performance. Crucially, the Prototype bra, shaped predominantly using MRI data, reduced breast displacement and evaluated better fit compared to the bra using 3D-SI data. Results from the study indicate that the prototype significantly improves fit and wearer comfort, suggesting a promising direction for future development of mastectomy bras. The findings highlight the potential of utilizing medical imaging data in garment design to create more inclusive and considerate solutions for post-surgical care.

A transfer learning approach for the classification of liver cancer

Abstract Problem The frequency of liver cancer is rising worldwide, and it is a common, deadly condition. For successful treatment and patient survival, early and precise diagnosis is essential. The automated classification of liver cancer using medical imaging data has shown potential outcome when employing machine and deep learning (DL) approaches. To train deep neural networks, it is still quite difficult to obtain a large and diverse dataset, especially in the medical field. Aim This article classifies liver tumors and identifies whether they are malignant, benign tumor, or normal liver. Methods This study mainly focuses on computed tomography scans from the Radiology Institute in Baghdad Medical City, Iraq, and provides a novel transfer learning (TL) approach for the categorization of liver cancer using medical images. Our findings show that the TL-based model performs better at classifying data, as in our method, high-level characteristics from liver images are extracted using pre-trained convolutional neural networks compared to conventional techniques and DL models that do not use TL. Results The proposed method using models of TL technology (VGG-16, ResNet-50, and MobileNetV2) successfully achieves high accuracy, sensitivity, and specificity in identifying liver cancer, making it an important tool for radiologists and other healthcare professionals. The experiment results show that the diagnostic accuracy in the VGG-16 model is up to 99%, ResNet-50 model 100%, and 99% total classification accuracy was attained with the MobileNetV2 model. Conclusion This proves the improvement of models when working on a small dataset. The use of new layers also showed an improvement in the performance of the classifiers, which accelerated the process.

Comparison of the Packet Wavelet Transform Method for Medical Image Compression

Medical images are often used for educational, analytical, and medical diagnostic purposes. Medical image data requires large amounts of storage on computers. Three types of codecs, namely Haar, Daubechies, and Biorthogonal, were used in this study. This study aims to find the best wavelet method of the three tested wavelet methods (Haar, Daubechies, and Biorthogonal). This study uses medical images representing USG and CT-scan images as testing data. The first test is carried out by comparing the threshold ratio. Three threshold values are used, namely 30, 40, and 50. The second test looks for PSNR values with different thresholds. The third test looks for a comparison of the rate (image size) to the PSSR value. The final test is to find each medical image's compression and decompression times. The first compression ratio test results on both medical images showed that CT scan images on Haar and Biorthogonal wavelets were the best, with an average compression ratio of 40.76% and a PSNR of 33.77. The PSNR obtained is also getting more significant for testing with a larger image size. The average compression time is 0.52 seconds, and the decompression time is 2.27 seconds. Based on the test results, this study recommends that the Daubechies wavelet method is very good for compression, which is 0.51 seconds, and the Biorthogonal wavelet method is very good for medical image decompression, which is 1.69 seconds.

The The Classification and Segmentation of Pneumonia using Deep Learning Algorithms: A Comparative Study

The paper uses convolutional neural networks (CNNs) to analyze radiography pictures and discriminate between areas afflicted by pneumonia and normal lung tissue. A sizable dataset of annotated chest X-rays is used to train the deep learning model, which enables it to pick up on complex patterns and characteristics linked to pneumonia. Pneumonia is a significant respiratory disease that affects a large number of individuals worldwide. Timely and accurate diagnosis of pneumonia plays a crucial role in effective treatment and management of the disease. We evaluate the performance of several up-to-date convolution neural network (CNN) architectures, namely ResNet-50, VGG-16, and DenseNet-121, then compare their results with traditional machine learning classifiers. Recent advancements in deep learning methods have shown accurate results in the investigation and diagnosis of medical image data, including the detection of pneumonia. This paper examines different deep-learning methods for categorizing pneumonia from lung X-ray imagery. Our results show that deep learning techniques performed better than conventional machine learning techniques in classifying pneumonia, with an estimated accuracy of 95% across all of the examined CNN models. These results demonstrate the potential of deep learning algorithms to significantly improve the accuracy and effectiveness of pneumonia diagnosis, supporting physicians in making knowledgeable decisions about patient care.

SE-NDEND: A novel symmetric watermarking framework with neural network-based chaotic encryption for Internet of Medical Things

The development of the Internet of Medical Things heavily relies on big data, and data security based on medical images has become a growing concern in society. Digital watermarking serves as a crucial technique for protecting and tracing medical image data copyright, as well as enabling forensic analysis. However, existing deep watermarking methods often neglect the protection of watermarks after extraction, leading to potential copyright disputes. To address this issue, this paper proposes SE-NDEND, a novel symmetric watermarking framework with neural network-based chaotic encryption for the Internet of Medical Things that significantly enhances the effectiveness and security of watermarking while maintaining robustness. Specifically, the SE-NDEND leverages neural networks to simulate chaotic systems and generate chaotic sequences, mitigating the complexity and high cost of implementing chaotic systems using hardware circuits. Moreover, we introduce a new noise layer with Moiré distortion that interacts with the decoder, forming a symmetric network structure that bolsters the robustness of watermarking. Parameters are jointly trained and shared during the training process to counteract potential interference from the noise layer. Experimental results validate the effectiveness of SE-NDEND in enhancing copyright protection, traceability, and forensic capabilities, surpassing existing deep learning methods in terms of visual quality (with PSNR of 45.8492 dB and SSIM of 0.9874), security, and robustness. The proposed framework can find application in protecting medical image data in the Internet of Medical Things.

Deep learning-powered biomedical photoacoustic imaging

Photoacoustic Imaging (PAI) is an emerging hybrid imaging modality that combines optical imaging and ultrasound imaging, offering advantages such as high resolution, strong contrast, and safety. Despite demonstrating superior imaging capabilities, PAI still has certain limitations in its clinical application, such as the trade-off between imaging depth and spatial resolution, and the need for further improvement in imaging speed. Deep Learning, as a novel machine learning technique, has gained significant attention for its ability to improve medical image data and has been widely applied in PAI in recent years to overcome these limitations. In this review, we first introduce the principles of photoacoustic imaging, followed by the development and applications of popular deep neural network structures such as U-Net and GAN networks. Furthermore, we comprehensively discuss the recent advancements in the application of deep learning in photoacoustic imaging. Finally, a summary and discussion are provided.