Frontiers in Biomedical Technologies

Digital Twins in Nuclear Medicine: A Pathway to Personalized Theranostics

Hossein Arabi — 2026-06-28

Theranostics has revolutionized nuclear medicine by integrating diagnostic imaging and targeted radionuclide therapy, delivering precision oncology with proven survival benefits in cancers such as prostate cancer and neuroendocrine tumors. However, inter-patient variability in biodistribution, response, and toxicity remains a major challenge. This editorial explores the transformative potential of digital twins, dynamic virtual replicas of patients continuously updated with real-world data, as a natural synergy for theranostics. Theranostic digital twins enable predictive dosimetry, personalized treatment optimization, responder identification, and toxicity forecasting through hybrid AI-mechanistic models grounded in radiopharmacokinetics and radiobiology. Early development should prioritize clinically meaningful applications supported by comprehensive, harmonized multimodal datasets, robust hybrid modeling, and effective synchronization mechanisms. Large-scale collaboration and systematic evidence synthesis are essential to accelerate clinical translation. By bridging in-silico simulation with real-world theranostics, digital twins promise to evolve nuclear medicine toward truly proactive, equitable, and predictive personalized care.

Exploring Brain Functional Connectivity in Hand Motion and Motor Imagery through fNIRS Signals: A Graph Theory Approach

Mahsan Hajihosseini — 2026-06-28

Purpose: Functional Near-Infrared Spectroscopy (fNIRS) is a valuable and cost-effective neuroimaging technique, particularly in the context of sensorimotor tasks and its applications in brain-computer interface (BCI) research. While numerous studies have explored brain functional connectivity during sensorimotor tasks, they have often primarily focused on electrical brain activity. In this study, we present a signal processing algorithm utilizing fNIRS-HbO2 data to identify active brain regions involved in both actual motor execution and motor imagery within a motor imagery task.

Materials and Methods: Our algorithm incorporates several key steps: firstly, the application of wavelet transform to eliminate noise and preprocess the fNIRS signal. Subsequently, we employ correlation analysis to extract functional connectivity matrices for both motor execution and motor imagery. Finally, we compute global efficiency (GE) values, a significant graph theory parameter, to analyze network properties. Additionally, we investigate the small-world network characteristics within the connectivity matrices and classify motor execution and motor imagery using a t-test.

Results: To gather data, we recorded 20-channel fNIRS signals, measuring changes in HbO2 concentration in the motor cortex, from 12 healthy participants at a sampling frequency of 10 Hz. Our findings not only confirm the presence of small-world network properties in the correlation matrices but also reveal that meaningful classification between motor execution and motor imagery of both right and left hands occurs when we select the top 40% of the strongest connections between channels. Furthermore, the results indicate a tendency towards stronger connectivity between channels in the left hemisphere.

Conclusion: In summary, our study demonstrates that brain networks are organized as small-world networks during sensorimotor tasks and underscores the prominent role of the dominant hemisphere in executing these tasks.

Smart Prediction: Class Centric Focal XG- Boost for Accurate Diabetes Forecasting

Vandana C. Bavkar — 2026-06-28

Purpose: Diabetes, resulting from insufficient insulin production or utilization, causes extensive harm to the body. The conventional diagnostic methods are often invasive. The classification of diabetes is essential for effective management. The progression in research and technology has led to additional classification approaches. Machine Learning (ML) algorithms have been deployed for analyzing the huge dataset and classifying diabetes.

Materials and Methods: The classification and the regression of diabetic and non-diabetic are performed using the XGBoost mechanism. On the other hand, the proposed class-centric Focal XG-Boost is applied to elevate the model performance by measuring the similarity among the features. The prediction of the model is based on the classification and regression rates of diabetic and non-diabetic individuals, which are anticipated using applicable and effectual metrics to estimate their working performance.

The dataset used in the Class-Centric Focal XG Boost model is attained using the Arduino Uno Kit. The data collection is done under a sampling rate of 100 Hz. The data are gathered from Bharati Hospital Pathology Laboratories, located in Pune.

Results: The inclusive outcomes of the proposed model with their appropriate Exploratory Data Analysis (EDA) among classification and regression, with the suitable dataset used in the study are exemplified.

Conclusion: The proposed Class-Centric Focal XG Boost model has numerous advantages and is less delicate to the hyperparameters than the conventional XGBoost algorithm. As a part of the real-time application of the Class-Centric Focal XG Boost model, the model can be utilized in other communicable and communicable disease classification and detection.

UCGNet: Capsule-Guided GAN for Ultrasound Image Reconstruction from Single-Plane Wave RF Data

Maryam Asad Samani — 2026-06-28

Purpose: This study aims to improve ultrasound image reconstruction from single-plane wave RF data using Capsule Neural Networks, which can produce comparable image quality to Convolutional Neural Networks while requiring significantly fewer parameters. In addition to reducing model size, the proposed approach preserves clinically important image features and is better suited for real-time implementation in embedded systems with constrained computational resources.

Materials and Methods: We propose a novel ultrasound image reconstruction architecture, UCGNet (U-Caps-GAN Network), which combines Capsule Networks with a Generative Adversarial Network framework. UCGNet reconstructs high-quality B-mode ultrasound images directly from single-plane wave RF data and is evaluated on the Plane-wave Imaging Challenge in Medical Ultrasound (PICMUS) dataset. Capsule Networks play a key role in achieving parameter efficiency by encoding spatial hierarchies through vectorized feature representations. Their dynamic routing mechanism captures part–whole relationships and pose variations, enabling the network to preserve fine structural details essential for diagnostic imaging, without relying on deep, redundant convolutional layers. This makes the proposed architecture particularly well-suited for real-time applications in embedded systems with limited computational resources.

Results: The reconstructed images achieved a mean Signal-to-Noise Ratio (SNR) of 18.45 and a Peak Signal-to-Noise Ratio (PSNR) of 40.92, outperforming the baseline UNet model in terms of image quality. Additionally, UCGNet required about 23% of the training parameters compared to UNet, demonstrating its suitability for real-time applications on resource-constrained devices.

Conclusion: UCGNet provides an efficient and accurate solution for ultrasound image reconstruction from raw RF data. Its improved image fidelity and reduced computational complexity make it a strong candidate for practical use in portable and embedded medical imaging systems.

Deep Learning-Based Prediction of IVF Success: A Transformer Model Approach

Mahvash Zargar — 2026-06-28

Purpose: Predicting the success of assisted reproductive technology (ART) remains a significant challenge due to the complex interplay of clinical, embryological, and demographic factors. This study aimed to develop and evaluate machine learning models, particularly deep learning-based approaches, to identify key predictors of ART success and improve outcome prediction accuracy.

Materials and Methods: A retrospective study was conducted on 500 infertile couples undergoing ART treatment between 2019 and 2024. A comprehensive dataset, including 84 clinical, embryological, and demographic variables, was analyzed. The key predictors included endometrial thickness, endometrial pattern, embryo transfer day, and hormonal markers (PRL, LH). Four machine learning models were implemented: Decision Tree, Random Forest, XGBoost, and a Transformer-Based Model. Data preprocessing involved feature selection, missing data handling, normalization, and oversampling techniques to address class imbalance. The models were trained and validated using k-fold cross-validation, and performance was assessed using accuracy, precision, recall, and F1 score.

Results: The Transformer-Based Model achieved the highest accuracy (99.7%), outperforming traditional machine learning models. This performance was validated using k-fold cross-validation and oversampling to mitigate overfitting and ensure generalizability. Endometrial pattern (r = 0.69) and endometrial thickness (r = 0.82) were the strongest predictors of ART success, emphasizing the dominant role of uterine factors. While female age and infertility duration had a weak negative correlation, male infertility factors and lifestyle variables (smoking, alcohol consumption) showed minimal predictive significance. Model-based feature importance confirmed uterine and embryological factors as the primary determinants of ART success, indicating a potential shift in treatment strategies toward optimizing endometrial receptivity and embryo transfer timing.

Conclusion: This study highlights the superiority of deep learning models in ART success prediction, with uterine factors emerging as the strongest predictors. Integrating AI-driven predictive models into clinical practice can enable personalized ART treatment, improved patient counseling, and optimized embryo transfer strategies, ultimately enhancing fertility outcomes. However, the findings are based on data from a single medical center, and further multi-center validation is needed to confirm the model’s generalizability.

Sleep Stages Classification Using Music Made from EEG By LSTM Networks

Hamidreza Jalali — 2026-06-28

Purpose: Automatic classification of sleep stages is one of the fundamental factors in diagnosing sleep disorders to prevent and treat various diseases, and it can significantly aid in saving specialists' time and energy. In this study, a new method for converting Electroencephalogram (EEG) signals to music for sleep stages classification is proposed.

Materials and Methods: A total of 15.233, 30-second data segments from the Sleep-EDF database were used as the statistical population for this evaluation. Initially, the performance of Long Short-Term Memory (LSTM) networks for music sequence generation is evaluated with the music database and the best structure is selected. Subsequently, single-channel EEG data are mapped to music pieces using the selected network. Seven features are extracted from the generated music sequences and applied to classification structures.

Results: The selected LSTM structure was able to identify musical sequences with an accuracy of 93.3% of the musical pieces. The overall classification accuracy for the five sleep stages according to the AASM standard is 85.3% for the Sleep-EDF database. Accuracy of classifying W, N1, N2, N3, and REM stages are 86.1%, 77.3%, 95.4%, 96.3%, and 71.4%, respectively. Another objective of this study is to present a novel single-channel EEG sonification method, achieving classification accuracy that is either higher than or comparable to contemporary methods.

Conclusion: The results of this study show that audio signal mapping with LSTM networks contains effective information for sleep stage classification, and the classification accuracy increased by 1% compared to the method of a similar study and by 3% compared to most studies.

Effect of Maternal Diabetes on Amniotic Fluid Index during the Second and Third Trimesters of Pregnancy: A Sonographic Case-Control Study

Khadija Masood — 2026-06-28

Purpose: Patients with diabetes are more likely to develop polyhydramnios. The rate of polyhydramnios among diabetic patients is increasing compared to non-diabetic patients.

To compare the Amniotic Fluid Index (AFI) of diabetic and non-diabetic patients using sonography.

Materials and Methods: A case-control study was conducted with 200 participants, comprising 100 diabetic patients and 100 non-diabetic patients. The study utilized a Toshiba XARIO XG ultrasound machine with a convex probe of 3.5-7.5 MHz frequency at the university ultrasound clinic in Green Town. All diabetic and gestational diabetic patients aged 18-45 years in their 2nd and 3rd trimesters were included. Patients with underlying pathologies such as hypertension or multiple gestations were excluded. Data analysis was performed using SPSS version 25.0.

Results: The mean amniotic fluid index in diabetic and non-diabetic groups was 21.19 and 13.20, respectively. The difference in the AFI between diabetics and non-diabetics was statistically significant (p=0.000). Chi-square analysis revealed a significant association between the AFI category and diabetes status. The diabetic group exhibited a higher proportion of cases in the polyhydramnios AFI category and a lower proportion in the normal AFI category compared to the non-diabetic group. The mean estimated fetal weight for diabetics and non-diabetics was 1341.64 and 1372.53 grams, respectively. There was no significant difference in estimated fetal weight between diabetic and non-diabetic patients (p=0.088).

Conclusion: The study concluded that diabetes during pregnancy is significantly associated with increased amniotic fluid levels, leading to a higher likelihood of polyhydramnios.

Injectable Nano-Chitosan /CaCO3 (NCsC) Composite as Novel Pulpotomy Paste in Partially Pulpotomized Rabbit Incisors

Atheer Abdulhussain Ali — 2026-06-28

Purpose: Maintaining pulp vitality and functions are of paramount importance in different branches of modern dentistry like endodontic, conservative, pedodontics, and dental traumatology, the applying of tissue-engineering principles to create simple and effective scaffold material suitable for maintaining pulp vitality, represent a promising therapeutic option for treating immature and mature tooth with compromised pulp vitality.

To evaluate the pathophysiological response of the pulp to (nano-chitosan/CaCO₃) scaffold as a potential novel capping material on traumatically exposed pulp of rabbit incisors.

Materials and Methods: Twenty-four upper incisors of twelve completely dentulous healthy New Zealand male rabbits were used, subdivided into two groups of 12 teeth. In each group, six teeth were used as a control group, where the pulp was traumatically exposed and partially amputated, left free of capping material, and the other six teeth, used as an experimental group, where the amputated pulp was capped with an injectable (nano-chitosan/CaCO₃) composite. The cavities of both groups sealed with resin modified glass ionomer cement as the final restoration. Animals were sacrificed, and the teeth were collected for histological examination according to the sacrifice time (1 and 4 weeks).

Results: At 1 week, both groups showed a non-significant difference in inflammatory extent. Highly significant difference in calcific bridge formation (P=0.002), and dentin morphology (P=0.002). At a 4-week period, there is a non-significant difference in inflammatory response. High significant difference in dentine bridge formation (P=0.002). Non-significant difference in the dentin morphology (P=0.06). The nano-chitosan/CaCO₃ group showed faster dentin bridge formation in both periods compared to the control group.

Conclusion: (nano-chitosan/CaCO₃) composite is a promising novel pulpotomy material.

White Matter Microstructural Changes in Primary Progressive Aphasia: Insights from Diffusion Tensor Imaging

Leila Golchin — 2026-06-28

Purpose: Primary Progressive Aphasia (PPA) is a neurodegenerative syndrome characterized by progressive language impairment. The present study investigated White Matter (WM) microstructural changes in PPA patients and their relationship with language and neuropsychological functions.

Materials and Methods: Diffusion Tensor Imaging (DTI) was used to examine 29 PPA patients and 13 healthy controls, focusing on 18 white matter tracts in both hemispheres.

Results: Significant differences in diffusivity values were observed between PPA patients and controls in multiple tracts, including the Cingulum, Arcuate Fasciculus (AF), Superior Longitudinal Fasciculus (SLF), Inferior Fronto-Occipital Fasciculus (IFOF), Inferior Longitudinal Fasciculus (ILF) bilaterally, as well as the left Uncinate Fasciculus (UF). Correlations between WM integrity and language functions were found in both hemispheres, with the left Cingulum showing positive correlations with various language measures. Notably, right hemisphere tracts (IFOF, ILF, SLF) positively correlated with several language domains, suggesting a potential compensatory role. White matter microstructural changes also correlated with neuropsychological functions (left Cingulum, Left ILF, right IFOF), highlighting PPA's interconnections of language and cognitive domains.

Conclusion: To our knowledge, the present study is the first to identify specific correlations between right hemisphere tracts, language domains, and cognitive functions in PPA patients. Our findings contribute to understanding the neural basis of language impairment in PPA, emphasizing the bilateral nature of language processing in neurodegenerative disorders. The results have implications for diagnosis, prognosis, and treatment planning in PPA, suggesting the need for therapeutic approaches that consider both hemispheres and the interplay between language and broader cognitive functions.

Stacking Ensemble Learning Approach for Non-Alcoholic Fatty Liver Disease Identification: Leveraging Explainable Machine Learning for Enhanced Prediction Models

Dolley Srivastava — 2026-06-28

Purpose: The prevalence of Non-Alcoholic Fatty Liver Disease (NAFLD) has significantly increased over the past two decades, becoming a leading cause of liver disease in industrialized nations, particularly among individuals who do not consume alcohol. This study aims to develop an efficient detection method for NAFLD using a stacked ensemble learning approach, which integrates multiple machine learning models to enhance predictive accuracy.

Materials and Methods: The dataset utilized in this research was sourced from an open platform and includes critical attributes such as age, gender, Body Mass Index (BMI), height, time to death or last follow-up, and survival status. We implemented a variety of machine learning algorithms, including XGBoost, CatBoost, Decision Trees, and AdaBoost, within a stacking framework to optimize performance. The proposed methodology involved several steps: data preparation, feature engineering, model training, and evaluation. Additionally, we employed explainable AI techniques to identify the most influential features contributing to NAFLD prediction, thereby enhancing the model's interpretability.

Results: The stacked ensemble model achieved an impressive classification accuracy of 95.9%, outperforming individual models and demonstrating the robustness of the ensemble approach. Confusion Metrics, ROC curves, and Calibration Curves are used to evaluate the proposed approach with state-of-the-art approaches. The suggested stacking methodology demonstrates superior performance in all contexts. When Explainable Machine Learning is applied to the proposed approach, it reveals that NAFLD is more common in middle-aged and elderly individuals, but is also present in younger age groups to some extent. Also, the prevalence of NAFLD is higher in males.

Conclusion: The results underscore the potential of a stacked ensemble approach for clinical applications in NAFLD screening and diagnosis, highlighting its importance in healthcare decision-making. By combining various machine learning techniques, we have developed a reliable and resilient model that improves detection accuracy and offers transparency in its predictions. Combining XGBoost, CatBoost, Decision Tree, and AdaBoost in a stacked ensemble model yielded the best results. The findings also indicate that age and gender are significant predictors of NAFLD, with the model providing valuable insights into the underlying patterns associated with the disease. This research contributes to the growing body of knowledge on machine learning applications in gastroenterology and emphasizes the need for explainable models in clinical settings.

Feasibility of Patient Quality Assurance Method Based on Log File and Onboard Detector in Helical Tomotherapy Technique

Ghazal Etemadi — 2026-06-29

Purpose: The phantom-less Patient-Specific Quality Assurance (PSQA) for Intensity‐Modulated Radiotherapy (IMRT) plan verification has been exploited recently. This study aimed to evaluate the feasibility of PSQA based on log files and onboard detectors for prostate patients in helical tomotherapy.

Materials and Methods: For 15 prostate patients, the Quality Assurance (QA) of the helical tomotherapy plan was performed using the Delta4 phantom and Cheese phantom to evaluate the spatial dose distribution and point dose, respectively. These parameters were also reconstructed by Delivery Analysis (DA) software using measured Leaf Open Times (LOTs). Gamma analysis and relative dose difference were used to compare the measured and reconstructed doses with the calculated values. Then, using the relative discrepancy, the log file and onboard detector data were compared to the expected data to assess machine performance.

Results: The mean relative dose difference was within 1.3% among the measurement, reconstruction, and calculation. Statistical analysis and p-value results indicated that there was no statistically significant difference in the dose difference between the DA-based and conventional QA methods. The gamma values for the DA-based QA method were similar to the measurement QA method for the criteria 3%/3mm, 3%/2mm, 2%/3mm, 2%/2mm, 2%/1mm, and 1%/1mm. However, the gamma values for the criteria 3%/1mm, 1%/3mm, and 1%/2mm were comparable. The mean percentage difference in LOTs was 0.07%, with most discrepancies occurring in very low and some high LOTs. The relative difference between the log file and expected data was lower than 2.30% for the couch speed, couch movement, monitor unit, and gantry rotation per minute.

Conclusion: The DA software is an efficient alternative to measurement-based PSQA methods. However, the accuracy of the DA software requires further investigation for gamma analysis with strict criteria. The very low and high LOTs may lead to the dose discrepancy. The tomotherapy machine can accurately implement the planned parameters.

An Effective Method to Repair Poor Signal of Magnetoencephalography Channel Data

Hanie Arabian — 2026-06-29

Purpose: Magnetoencephalography is the recording of magnetic fields resulting from the activities of brain neurons and provides the possibility of direct measurement of their activity in a non-invasive manner. Despite its high spatial and temporal resolution, magnetoencephalography has a weak amplitude signal, drastically reducing the signal-to-noise ratio in case of environmental noise. Therefore, signal reconstruction methods can be effective in recovering noisy and lost information.

Materials and Methods: The magnetoencephalography signal of 11 healthy young subjects was recorded in a resting state. Each signal contains the data of 148 channels which were fixed on a helmet. The performance of three different reconstruction methods has been investigated by using the data of adjacent channels from the selected track to interpolate its information. These three methods are the surface reconstruction methods, partial differential equations algorithms, and finite element-based methods. Afterward to evaluate the performance of each method, R-square, root mean square error, and signal-to-noise ratio between the reconstructed signal and the original signal were calculated. The relation between these criteria was checked through proper statistical tests with a significance level of 0.05.

Results: The mean method with the root mean square error of 0.016 ± 0.009 (mean ± SD) at the minimum time (3.5 microseconds) could reconstruct an epoch. Also, the median method with a similar error but in 5.9 microseconds with a probability of 99.33% could reconstruct an epoch with an R-square greater than 0.7.

Conclusion: The mean and median methods can reconstruct the noisy or lost signal in magnetoencephalography with a suitable percentage of similarity to the reference by using the signal of adjacent channels from the damaged sensor.

A Deep Learning Approach: Effective Multi-Class Classification of Alzheimer's Disease using Unified Integration in the Tri-Branch Network with Efficient Net

Kumar Prasun — 2026-06-29

Purpose: One of the increasing neurological disorders is Alzheimer's, which progressively weakens brain cells and leads to critical cerebral impairments like memory loss. The present diagnostic techniques comprise PET scans, MRI scans, CSF biomarkers, and others that frequently need manual power and time-consuming process which might not offer appropriate results. This emphasizes the requirement for more precise and potential diagnostic solutions.

Materials and Methods: The proposed model utilizes AI-based Deep Learning (DL) techniques for effective multi-class classification of AD such as Early Mild Cognitive Impairment (EMCI), Late Mild Cognitive Impairment (LMCI), Mild Cognitive Impairment (MCI), Cognitive Normal (CN) and Alzheimer’s Disease (AD) using Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset. The proposed study utilizes Tri Branch Attention Network (TBAN) with Unified Component Incorporation (UCI) by capturing both spatial and channel attention information, by replacing the Squeeze and Excitation (SE) component in the conventional EfficientNet model and helps in addressing the concerns associated to imbalanced spatial feature distribution in images. Further, the incorporation of the proposed TBAN module in the Conv Layer helps, not only in terms of capturing the long-term dependence between the different channels of the network but also helps in retaining the specific location information to enhance the performance of the model. Similarly, the proposed UCI which is used in the MBConv layer deals with regularization, as the accuracy of the model can be dropped due to unbalanced regularization, hence the incorporation of UCI advocates strong regularization for combatting the concerns associated with overfitting and aids in providing better accuracy.

Results: Eventually, the proposed framework is evaluated with different metrics and the accuracy value obtained by the proposed model is 0.95. Likewise, precision, recall, and F1 scores gained by the proposed work are 0.95, 0.95, and 0.95.

Conclusion: The proposed research resolves significant gaps in the present diagnostic practices by implementing emerged AI techniques to improve the efficacy and accuracy of Alzheimer's diagnosis by medical imaging. Through enhancing the abilities of early detection, this proposed model holds the prospective to majorly affect treatment tactics for people affected with Alzheimer's. Finally, it led to better patient consequences and life quality.

Assessing the Difference between Equilibrium Dose and CTDI in Effective Dose Estimation

Soheila Sharifian Jazi — 2026-06-29

Purpose: The dose of Computed Tomography (CT) scan exams consists of a large proportion of all medical imaging modalities' dose burdens. There are different methods to measure and describe radiation in CT. A standardized way is to measure the Computed Tomography Dose Index (CTDI). However, due to the increase in the detector system size along the z-axis in new CT scanner generations, new measurement methods are described in the American Association of Physicists in Medicine-Task Group No.111(AAPM-TG111). This study aims to estimate the equilibrium dose and compare it with the amount displayed in the volume Computed Tomography Dose Index (CTDI_vol) at the end of each exam. Eventually, the effective dose was calculated for both methods.

Materials and Methods: Using a pencil ionization chamber and standard polymethylmethacrylate (PMMA phantom), the following values were calculated: CTDI₁₀₀, CTDI_vol, cumulative dose, equilibrium dose, and effective dose.

Results: Six protocols performed in two centers, and the results indicated that the measurements with a standard CT dosimetry phantom, was varied between average equilibrium dose and CTDI_vol, and the discrepancies ranged between 27% to 33%.

Conclusion: The CTDI_Vol is not suitable for evaluating the radiation dose at the end of each scan, and the use of an equilibrium dose for dosimetry of new systems is recommended.

Comparative Analysis of Diffusion Tensor Imaging Estimation Methods

Somaye Jabari — 2026-06-29

Purpose: Diffusion Tensor Imaging (DTI) is a noise-sensitive method, where a low Signal-to-Noise Ratio (SNR) results in significant errors in the estimated tensor field. This topic focuses on a comprehensive evaluation of various DTI estimation methods, such as Linear Least Squares (LLS), Weighted Linear Least Squares (WLLS), iterative re-weighted Linear Least Squares (IRLLS), and Non-linear Least Squares (NLS). The article will explore how each method performs in terms of accuracy, efficiency in estimating the diffusion tensor and robustness against noise.

Materials and Methods: The study compares the methods using simulated diffusion-weighted Magnetic Resonance Imaging (MRI) data. Time complexity and performance of the LLS, WLLS, IRLLS, and NLS methods were evaluated across key metrics such as TRMSE, RMSE, MSD, and ΔSNR.

Results: The results of the study demonstrate that LLS and IRLLS consistently outperform other methods in terms of TRMSE, MSD, and SNR, particularly in high-noise scenarios. NLS performs best in reducing RMSE but high noise causes it to fit to noise, so it is not robust. WLLS showed the weakest performance across all metrics.

Conclusion: The paper suggests that LLS, despite its simplicity, remains a competitive option in terms of capturing the true underlying diffusion properties. IRLLS further refines this by iteratively reducing the effect of outliers in tensor estimation.

Impact of Image Reconstruction on Quantitative Analysis of 18F-FDG PET in Epilepsy Evaluation: A Preliminary Study

Naghmeh Firouzi — 2026-06-29

Purpose: In epilepsy pre-surgical evaluations, semi-automated quantitative analysis of Fluorine-18-fluorodeoxyglucose (¹⁸F-FDG) brain PET complements visual assessment for localizing the seizure onset zone. This study evaluates how adjusting reconstruction parameters enhances quantitative accuracy, aiming to identify optimal configurations for reliable clinical decision-making.

Materials and Methods: 234 reconstruction methods were applied to ¹⁸F-FDG brain PET images of a 47-year-old male with focal epilepsy. The parameters encompassed the 3D-Ordered-Subset Expectation Maximization image reconstruction method, both with Resolution Recovery (RR) and without (non-RR), various numbers of iterations×subset (#it×sub), pixel sizes, and Gaussian filters. The accuracy errors were determined by the Relative Difference Percentage (RDP) in measured maximum standardized uptake value SUV_max and absolute Z-scores from all 234 reconstructed images, compared to reference values from the normal database reconstruction set as the benchmark

Results: The study revealed that reconstructed images with 5 mm or 8 mm full width at half maximum (FWHM) Gaussian filters yielded RDP values above 5% for SUV_max and Z-scores, indicating potential inaccuracy with higher values of post-smoothing filters. The recommended reconstruction sets with RDP values below 5% for both RR and non-RR images were those with a 3 mm FWHM Gaussian filter and higher (#it×sub), specifically (5×21, 8×21), (5×21, 6×21), and (7×21, 8×21) for pixel sizes of 1.01 mm, 1.35 mm, and 2.03 mm, respectively.

Conclusion: The findings underscore the significant impact of altering the image reconstruction sets on the SUVmax and Z-scores. Furthermore, the inconsistent fluctuations of Z-scores emphasize the importance of using standardized image reconstruction sets to ensure accurate and reliable quantitative outcomes in epilepsy pre-surgical evaluations.

Evaluating the Role of Channel Selection in EEG Anxiety Recognition Rates Utilizing a Chaotic Map

Faezeh Daneshmand-Bahman — 2026-06-29

Purpose: Today, the human lifestyle has led to an increase in anxiety. Its diagnosis is usually made with questionnaires and by specialist physicians. Recently, objective techniques such as brain-behavior analysis have captivated the attention of scientists for the early detection of this disorder. This study aimed to provide a method for diagnosing anxiety based on electroencephalogram (EEG) signals. Also, presents a new methodology by examining different approaches to brain channel selection and feature extraction based on chaotic maps.

Materials and Methods: The DASPS database was used, containing a 14-channel EEG of 23 people (10 men and 13 women, average age: 30 years). The self-assessment manikin was applied to divide anxiety into 2 and 4 levels. Firstly, four methods were assessed to select the optimal channel; two methods were based on the minimum coefficient of variation, and two methods were based on the maximum relative power. Then, Chebyshev’s chaotic map was reconstructed, and two features, including 1) the maximum density and 2) its corresponding sample, were extracted. Finally, the k-Nearest Neighbor (K-NN) and Support Vector Machine (SVM) classifiers were applied.

Results: The results indicated a maximum accuracy of 100% for both two/four-level anxiety detection. In addition, the K-NN outperformed the SVM classifier.

Conclusion: It highlighted the role of some brain channels, as well as the classifier structure, in distinguishing anxiety levels. The outstanding result of the proposed algorithm nominated it as a suitable approach for anxiety detection.

Identifying the Arm Joint Dynamics Using Muscle Synergy Patterns and SVMD-BiGRU Hybrid Mechanism

Seyyed Ali Zendehbad — 2026-06-29

Purpose: In this study, we propose a novel generalizable hybrid underlying mechanism for mapping Human Pose Estimation (HPE) data to muscle synergy patterns, which can be highly efficient in improving visual biofeedback.

Materials and Methods: In the first step, Electromyography (EMG) data from the upper limb muscles of twelve healthy participants are collected and pre-processed, and muscle synergy patterns are extracted from it. Concurrently, kinematic data are detected using the OpenPose model. Through synchronization and normalization, the Successive Variational Mode Decomposition (SVMD) algorithm decomposes synergy control patterns into smaller components. To establish mappings, a custom Bidirectional Gated Recurrent Unit (BiGRU) model is employed. Comparative analysis against popular models validates the efficacy of our approach, revealing the generated trajectory as potentially ideal for visual biofeedback. Remarkably, the combined SVMD-BiGRU model outperforms the alternatives.

Results: The results show that the trajectory generated by the model is potentially suitable for visual biofeedback systems. Remarkably, the combined SVMD-BiGRU model outperforms the alternatives. Furthermore, empirical assessments have demonstrated the adept ability of healthy participants to closely adhere to the trajectory generated by the model output during the test phase.

Conclusion: Ultimately, incorporating this innovative mechanism at the heart of visual biofeedback systems has been revealed to significantly elevate both the quantity and quality of movement.

Facilitating Timely Decision-Making in Healthcare: An Object Detection Approach for Automated Coronary Artery Stenosis Detection

Hadis Keshavarz — 2026-06-29

Purpose: Coronary Artery Disease (CAD), characterized by coronary artery stenosis—the narrowing of arteries supplying blood to the heart—is a leading global cause of morbidity and mortality. Timely detection and management of stenosis are crucial to preventing severe outcomes such as myocardial infarction and heart failure. Despite advancements in medical imaging, current diagnostic methods rely heavily on the manual interpretation of coronary angiograms, which is time-consuming, subjective, and prone to variability. To address these limitations, this study proposes an automated object detection-based framework for identifying coronary artery stenosis in medical imaging.

Materials and Methods: The study employs two state-of-the-art deep learning models, RetinaNet and EfficientDet D3, to detect stenotic regions in X-ray angiography images. A dataset of 8,325 annotated images from 100 patients with single-vessel CAD, sourced from the Research Institute for Complex Issues of Cardiovascular Diseases in Kemerovo, Russia, was used for training and evaluation. To enhance model performance, a comprehensive preprocessing pipeline was applied, including image resizing, data augmentation, and intensity normalization. These steps ensured robustness and generalizability across diverse imaging conditions.

Results: Both models demonstrated high accuracy in stenosis detection. RetinaNet achieved a mean Average Precision (mAP) of 93.2%, while EfficientDet D3 outperformed with an mAP of 96.6%. These results highlight the models' ability to accurately identify stenosis, even in noisy and variable angiographic images. The superior performance of EfficientDet D3 underscores its potential for clinical integration, offering precise and reliable stenosis localization.

Conclusion: This study presents a robust and efficient deep learning framework for the automated detection of coronary artery stenosis. By reducing reliance on manual interpretation and enhancing diagnostic accuracy, the proposed approach supports timely and informed clinical decision-making. This innovation has the potential to streamline diagnostic workflows, improve patient outcomes, and advance the application of artificial intelligence in cardiovascular healthcare.

Brain Structural Changes are Associated with Motor Function: A Study of Healthy Young Adults from the Human Connectome Project

Yunus Soleymani — 2026-06-29

Purpose: There is a known decline in brain volume with age, impacting cognitive health and increasing the risk of diseases such as dementia and Alzheimer's. Physical activity has been shown to have positive effects on brain structure and cognitive function with aging. Still, the association between motor function and brain volume in young adults remains unclear.

Materials and Methods: This study utilized high-resolution T1-weighted MRI images and motor function test results from 1082 healthy young adults aged 22-37, sourced from the Human Connectome Project Young Adult (HCP-YA). Motor functions were assessed using four tests: Endurance, Gait Speed, Dexterity, and Strength. Correlation analysis and multiple linear regression models were used to evaluate the association between motor functions and brain volumes, adjusting for demographic variables and Body Mass Index (BMI).

Results: Significant positive correlations were found between Endurance and Strength tests with multiple brain volumes. In contrast, the Dexterity test showed negative correlations reflecting intricate patterns of neural connectivity and plasticity, which may not directly correlate to brain volumes. No significant correlations were observed for the Gait Speed test, indicating that it may not be a sensitive indicator of brain health in younger adults. Multiple linear regression analyses revealed that total brain (β = 0.045, SE = 0.020), total gray matter (GM) (β = 0.035, SE = 0.016), left white matter (WM) (β = 0.058, SE = 0.025), right WM (β = 0.056, SE = 0.025), total WM (β = 0.057, SE = 0.025), and left accumbens (β = -0.072, SE = 0.031) volumes were significantly associated with motor function scores (p < 0.05).

Conclusion: Physical fitness, as measured by motor function tests, is significantly associated with brain structural integrity in young adults. These findings highlight the potential importance of physical activity in maintaining brain health, which could inform strategies to promote active lifestyles and prevent neurodegenerative diseases.

Long-Term EEG-Based Modeling and Classification of Migraine Phases Using Hidden Markov Models

Safoura Ashoorisefat — 2026-06-29

Purpose: Migraine is a complex neurological disorder characterized by dynamic alterations in brain activity during multiple phases: interictal (baseline), preictal, ictal, and postictal. This study aims to model and differentiate these migraine phases using Electroencephalogram (EEG) and a Hidden Markov Model (HMM).

Materials and Methods: EEG signals were collected from each subject over several months through frequent, short sessions often multiple times per day. The recordings were temporally aligned with self-reported symptom diaries, allowing for precise labeling of migraine phases. A comprehensive set of features was extracted from the EEG signals, including spectral, temporal, and nonlinear measures such as Dynamic Mode Decomposition (DMD) and Katz Fractal Dimension (KFD) across various frequency bands. Despite the limited number of participants, the dense long-term recordings captured multiple migraine episodes, enabling reliable phase modeling.

Results: The HMM identified migraine-specific neural patterns, achieving an average classification accuracy of approximately 87% for all 15 patients, with individual patient performance ranging from 70% to 95%, depending on signal length and normalization. Only three patients are shown in detail in the results section as illustrative examples.

Conclusion: The HMM identified distinguishable neural patterns corresponding to migraine states, suggesting the feasibility of temporal EEG modeling for clinical applications in personalized migraine management.

Targeted Radiosensitization in Cancer Radiotherapy Using Functionalized Nanocarriers: A Systematic Review

Fatemeh Zare — 2026-06-29

Purpose: This study aims to provide a comprehensive review of recent advances in the application of nanocarriers for targeted drug delivery and radiosensitization in cancer Radiotherapy (RT), as well as to examine the challenges, solutions, and prospects of this technology.

Materials and Methods: This systematic review was conducted in accordance with PRISMA guidelines and protocol registered in PROSPERO (CRD420251154905). A comprehensive literature search was conducted in PubMed, Scopus, and Web of Science, identifying 373 records. Following PRISMA guidelines, 40 studies met the inclusion criteria focusing on functionalized nanocarriers in cancer RT. Data extraction covered nanoparticle types, functionalization, therapeutic payloads, cancer models, radiation modalities, and outcomes.

Results: Forty studies were analyzed, categorized into iron oxide-based (10), silver (10), bismuth-based (7), graphene-based (4), gadolinium-based (4), and titanium-based (2) nanoparticles (NPs). Bismuth-based NPs demonstrated superior radiosensitization with sensitizer enhancement ratios (SERs) of 1.25–1.48 and up to 450% increase in reactive oxygen species (ROS) in vivo, achieving ~70% tumor volume reduction without systemic toxicity. Silver NPs demonstrated dose enhancement factors (DEF) rising from 1.4 to 1.9 and synergistic effects with docetaxel plus 2 Gy radiation. Iron oxide NPs functionalized with HER2 and RGD ligands reduced cell viability by 1.95-fold and achieved DEF of 89.1 in targeted systems. Gadolinium NPs reached SERs up to 2.44 at 65 keV, while graphene-based systems enhanced ROS production by 75.2%. Titanium-based NPs increased ROS levels 2.5-fold. Combination therapies integrating chemotherapeutics, including cisplatin and curcumin with nanocarriers, yielded SERs up to 4.29. The radiation modalities included megavoltage X-rays (4–10 MV, n=24), synchrotron keV X-rays (n=2), gamma rays (0.38–1.25 MeV, n=3), and electron beams (6–12 MeV, n=3).

Conclusion: Bismuth-based NPs represent the most promising radiosensitizers due to their high efficacy, safety, and clinical relevance, supporting their advancement toward clinical translation.

Radiopharmaceuticals: A Brief Overview of Basic Pharmacological Parameters

Mahshid Kiani — 2026-06-29

Radiopharmaceuticals are combinations of two main components including a pharmaceutical ingredient that targets specific moieties, and radionuclide, which acts through spontaneous degradation to create diagnostic or therapeutic effects, as well as both effects simultaneously known as theranostics. By combining diagnostic and therapeutic methods, radiotheranostics play an important role in reducing patient radiation dosages, increasing treatment effectiveness, controlling side effects, improving patient outcomes, and reducing overall treatment costs. Despite the diagnostic and therapeutic roles, radiopharmaceuticals are beneficial for assessing prognosis, disease progression, and the possibility of recurrences, treatment planning strategies, and assessing response to treatment. The most incredible role of radiopharmacy is establishing new radiopharmaceuticals with the aim of better targeting functions and enhanced tolerability for imaging and treatment purposes in a clinic. These approaches are supported by nuclear medicine non-invasive procedures. It is crucial for radiopharmaceuticals that drug delivery occurs in a highly selective and sensitive manner to minimize the potential radiation risk to non-targeted organs of patients. This report will provide an overview of basic pharmacological patterns related to clinical radiopharmaceuticals for diagnosis and therapy, including the latest radiotheranostic tracers, key concerns within the field, and future trends and prospects. Additionally, the available and useful radiopharmaceuticals are categorized into separate tables based on their specific characteristics. Presenting information in table format enhances organization and makes the data more understandable and accessible for users. This structured approach allows users to quickly locate relevant information, compare different radiopharmaceuticals, and grasp essential details at a glance. By utilizing tables, we ensure that critical information is not only easy to read but also effectively highlights the unique attributes of each radiopharmaceutical, ultimately improving the decision-making process for healthcare professionals.

Towards Routine AI-Based PET/CT and SPECT/CT Lesion Segmentation and Tracking in PSMA Theranostics

Fereshteh Yousefirizi — 2026-06-29

Quantitative molecular imaging is central to treatment response assessment in oncology, yet clinical practice remains largely dominated by patient-level or limited target-lesion criteria that ignore inter-lesion heterogeneity. This limitation is particularly important in prostate cancer, where PSMA PET/CT can reveal extensive skeletal and nodal metastatic disease that often evolves heterogeneously under therapy. Accurate and scalable lesion segmentation and tracking across serial PSMA PET/CT and post-therapy SPECT/CT scans is therefore essential for implementing emerging PSMA-specific response frameworks, such as RECIP 1.0, and for enabling lesion-level dosimetry in ¹⁷⁷Lu-PSMA Radiopharmaceutical Therapies (RPTs).

This article examines clinical motivations, technical foundations, and future pathways for automated lesion tracking in prostate cancer imaging. We focus on the unique requirements introduced by PSMA PET/CT compared with FDG PET/CT and highlight the critical role of quantitative SPECT/CT in linking imaging-derived disease characterization with delivered therapeutic dose. Recent advances in AI-based segmentation and automated lesion matching now make scalable longitudinal lesion correspondence feasible, providing comprehensive infrastructure for standardized response assessment and personalized PSMA-based theranostics.