filter by: Publication Year
(Descending) Articles
Cancer/Radiotherapie (17696658) 29(4)
Patient outcomes are significantly impacted by the effectiveness and quality of radiation treatment planning. Deep learning, a branch of artificial intelligence, is a potent tool for enhancing and automating dose prediction processes. This article provides a comprehensive and critical analysis of deep learning-based dose prediction methods in radiotherapy, with a focus on convolutional neural networks. A comprehensive search throughout Elsevier Scopus®, Medline, and Web of Science™ literature databases was conducted to locate relevant papers published between 2018 and 2024. The use of deep learning methods for dose prediction is thoroughly examined in this paper. Analysis of these dose prediction approaches provides valuable insights into the potential of this technology to improve radiation treatment planning, particularly in the critical area of automating the dose prediction process. The findings aim to guide future research and facilitate the safe and effective integration of artificial intelligence in clinical workflows. © 2025
Rasti, R. ,
Kumbakumba, E. ,
Nanjebe, D. ,
Mlotshwa, P. ,
Nassejje, M. ,
Mzee, J. ,
Businge, S. ,
Akankwasa, G. ,
Nyehangane, D. ,
Gantelius, J. Bmc Infectious Diseases (14712334) 25(1)
Background: In low-resource settings, limited laboratory capacity adds to the burden of central nervous system (CNS) infections in children and spurs overuse of antibiotics. The commercially available BioFire® FilmArray® Meningitis/Encephalitis Panel (FA-ME) with its capability to simultaneously detect 14 pathogens in cerebrospinal fluid (CSF), could potentially narrow such a diagnostic gap. Methods: In Mbarara, Uganda, we compared clinical utility (clinical turnaround time [cTAT], microbial yield, and influence on patient outcome and antibiotic exposure) of FA-ME with bacterial culture, in children 0–12 years with suspected CNS infection. Results: Of 212 enrolled children, CSF was sampled from 194. All samples underwent bacterial culture, of which 193 also underwent FA-ME analyses. FA-ME analyses prospectively influenced care for 169 of the 193 patients, and they constituted an ‘Index group’. The remaining 43/212 patients constituted a ‘Reference group’. Of all 194 CSF-sampled patients, 87% (168) had received antibiotics before lumbar puncture. Median cTAT for FA-ME was 4.2 h, vs. two days for culture. Bacterial yield was 12% (24/193) and 1.5% (3/194) for FA-ME and culture, respectively. FA-ME viral yield was 12% (23/193). Fatality rate was 14% in the Index group vs. 19% in the Reference group (P = 0.20). From clinician receival of FA-ME results, median antibiotic exposure was 6 days for bacteria-negative vs. 13 days for bacteria-positive patients (P = 0.03). Median hospitalization duration was 7 vs. 12 days for FA-ME negative and positive patients, respectively (P < 0.01). Conclusions: In this setting, clinical FA-ME utility was found in a higher and faster microbial yield and shortened hospitalization and antibiotic exposure of patients without CSF pathology. More epidemiologically customized pathogen panels may increase FA-ME utility locally, although its use in similar settings would require major cost reductions. Trial registration: The trial was registered with clinicaltrials.gov (NCT03900091) in March 2019, and its protocol was published in November 2020. © The Author(s) 2025.
Hossein nikzad, M. ,
Heidari-rarani, M. ,
Rasti, R. Materials Letters (18734979) 379
Nikzad, M.H. ,
Heidari-rarani, M. ,
Rasti, R. ,
Sareh, P. Expert Systems with Applications (09574174) 264
Additive manufacturing (AM) has become a transformative technology in modern production, enabling complex geometric designs with minimal material waste. A significant aspect of AM, particularly in fused deposition modeling (FDM), is the need for precise prediction of mechanical properties, such as ultimate tensile strength (UTS), which is crucial for industrial applications. This study examines whether simple machine learning (ML) algorithms can accurately predict the UTS of 3D-printed polylactic acid (PLA) parts, and evaluates the effectiveness of ML techniques, especially ensemble methods, in enhancing prediction accuracy. To this end, the study compares simple ML algorithms to identify the most accurate model for predicting the UTS of 3D-printed PLA parts. Subsequently, an average ensemble technique combines four ML algorithms, namely categorical boosting (CatBoost), extreme gradient boosting (XGBoost), gradient boosting machine (GBM), and light gradient boosting machine (LGBM), to predict UTS. In this technique, the average predicted UTS values of CatBoost, XGBoost, GBM, and LGBM are taken as the final predicted UTS value. Additionally, 11 ensemble configurations of these algorithms are analyzed to determine the optimized ensemble configuration. The results show that the CatBoost algorithm, with an R2 of 94.46%, achieved the highest predictive accuracy among individual ML algorithms. Moreover, the CatBoost-XGBoost-GBM-LGBM ensemble was the most accurate configuration, achieving an R2 of 98.05% with less than 10% error in predicting 37 external data points not included in the training and testing sets. This study advances predictive modeling in AM by demonstrating that ML, particularly ensemble techniques, can reliably predict material properties, paving the way for more robust applications of AM in industry. © 2024 Elsevier Ltd
Aghapanah, H. ,
Rasti, R. ,
Tabesh, F. ,
Pouraliakbar, H. ,
Sanei, H. ,
Kermani, S. Biomedical Signal Processing and Control (17468108) 100
Accurate segmentation of the left ventricle, right ventricle, and myocardium is essential for estimating key cardiac parameters in diagnostic procedures. However, automating Cardiovascular Magnetic Resonance Imaging (CMRI) segmentation faces challenges from diverse imaging vendors and protocols. This study introduces MECardNet framework as an innovative multiclass CMRI segmentation model, representing a prominent advancement in the field. MECardNet leverages a Multiscale Convolutional Mixture of Experts (MCME) ensemble technique with Adaptive Deep Supervision, seamlessly integrated into the U-Net architecture. The MCME framework improves representation learning in the U-Net workflow. It does this by adaptively adjusting the contribution of U-Net layers in the ensemble for better data modeling. Additionally, MECardNet incorporates a cross-additive attention mechanism, an EfficientNetV2L backbone, and a specialized compound loss function, leading to enhanced model performance. Through 10-fold Cross-Validation (CV) analysis on the ACDC dataset, MECardNet surpasses baseline models and state-of-the-art methods, showcasing promising performance levels with evaluation metrics such as Dice Similarity Coefficient (DSC) of 96.1 ± 0.4 %, Jaccard coefficient of 92.2 ± 0.4 %, Hausdorff distance of 1.7 ± 0.1 and mean absolute distance of 1.6 ± 0.1. Further validation on the M&Ms-2 dataset and a local dataset confirms promising performance of MECardNet, with DSC of 94.3 ± 0.7 % and 94.5 ± 0.6 %, respectively. The proposed MECardNet framework establishes a new benchmark in CMRI segmentation by outperforming existing models, offering efficient and reliable computer-aided technologies for cardiovascular disease diagnosis, with the potential for significant impact in the field. Researchers can access MECardNet repository and results on GitHub1 for comprehensive exploration and utilization. © 2024 Elsevier Ltd
Materials Today Communications (23524928) 41
Assessing the elastic modulus of 3D-printed polylactic acid (PLA) components is essential for understanding their stiffness and load capacity, which are crucial for predicting product performance and durability. In this study, the predictive accuracy of a Tabular Neural Network (TabNet) algorithm for determining the elastic modulus of 3D-printed PLA components via fused deposition modeling (FDM) was investigated. Utilizing a comprehensive dataset of 128 literature-sourced data points, divided into 80 % for training and 20 % for validation, the study proposed a new Taguchi-based method for efficient hyperparameter optimization of the TabNet algorithm. This optimization revealed that a configuration of 8 decision blocks, 16 attention blocks, and 5 decision steps, along with the “Adam” optimizer, a gamma of 1, learning rate of 0.1, and lambda-sparse of 0.01, yielded the highest prediction accuracy for the elastic modulus of PLA parts. The performance of the optimized TabNet model was evaluated using R-squared (R²), Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE) measures. The findings highlighted an R² of 96.855 %, an MAE of 0.158, an MSE of 0.037, and an RMSE of 0.193 in the validation dataset, demonstrating substantial predictive reliability. To further test the model's robustness, fourteen unseen data points were analyzed. The observed discrepancies between predicted and actual values were under 10 %, affirming the Taguchi-optimized TabNet algorithm's effectiveness in forecasting the elastic modulus of FDM 3D-printed PLA components. This investigation provides a significant advancement in additive manufacturing, introducing a precise and reliable method for predicting the mechanical properties of 3D-printed materials. © 2024 Elsevier Ltd
2025 29th International Computer Conference, Computer Society of Iran, CSICC 2025
The global significance of the steel industry as an economic cornerstone cannot be overstated, with its pivotal role in construction, automotive manufacturing, and pipe production. This paper investigates the transformative influence of deep learning, encompassing machine vision and artificial intelligence, on elevating performance standards within the steel industry. The industry's critical contribution to manufacturing building materials, automotive components, and high-value energy and fluid transmission pipes underscores the need for continuous technological evolution. Machine vision and artificial intelligence have emerged as pivotal catalysts in the pursuit of precision data analysis and enhanced industrial performance. This research explores the escalating importance of these technologies, elucidating their substantial impact on refining industrial processes within the steel sector. Recognized as powerful instruments for progression and optimization, machine vision and artificial intelligence contribute significantly to the industry's technological landscape. This study comprehensively reviews pertinent articles to delve into the myriad applications of machine vision and artificial intelligence in the steel industry. By scrutinizing the latest developments and applications, the paper aims to provide a thorough understanding of how these technologies are actively shaping the industry's landscape. The findings underscore the instrumental role of deep learning in augmenting efficiency, fostering innovation, and ultimately advancing the standards of the steel industry on a global scale. © 2024 IEEE.
Rasti, R. ,
Aghapanah, H. ,
Kermani, S. ,
Aghapanah, H. ,
Rasti, R. ,
Kermani, S. 2025 29th International Computer Conference, Computer Society of Iran, CSICC 2025 pp. 319-326
Accurate motion tracking and visualization of cardiac structures in MRI images are crucial for diagnosing and treating heart diseases. This paper introduces CardioTrackNet, a novel hybrid model that integrates an Active Mesh Model with a Pyramidal Warping and Cost Volume (PWC) Model for advanced cardiac MRI motion analysis and visualization. CardSegNet is first utilized for precise heart region segmentation. Subsequently, the Active Mesh and PWC models track the motion of each volume across frames. The Active Mesh Model provides a detailed, flexible representation of cardiac structures, while the PWC Model excels in optical flow estimation. These motion estimates are synthesized within CardioTrackNet, combining segmented and tracked data. The final output is refined using six bull's-eye data from CVI42 software, ensuring clinically relevant motion visualization. Experimental results demonstrate that CardioTrackNet significantly improves the accuracy of cardiac motion tracking, achieving a Dice Similarity Coefficient (DSC) of 93 ± 1%. Additionally, the system offers clear and informative visualizations, potentially enhancing clinical workflows and supporting better diagnostic and therapeutic decisions. © 2024 IEEE.
Aghapanah, H. ,
Rasti, R. ,
Kermani, S. ,
Tabesh, F. ,
Banaem, H.Y. ,
Aliakbar, H.P. ,
Sanei, H. ,
Segars, W.P. Computerized Medical Imaging and Graphics (18790771) 115
Cardiovascular MRI (CMRI) is a non-invasive imaging technique adopted for assessing the blood circulatory system's structure and function. Precise image segmentation is required to measure cardiac parameters and diagnose abnormalities through CMRI data. Because of anatomical heterogeneity and image variations, cardiac image segmentation is a challenging task. Quantification of cardiac parameters requires high-performance segmentation of the left ventricle (LV), right ventricle (RV), and left ventricle myocardium from the background. The first proposed solution here is to manually segment the regions, which is a time-consuming and error-prone procedure. In this context, many semi- or fully automatic solutions have been proposed recently, among which deep learning-based methods have revealed high performance in segmenting regions in CMRI data. In this study, a self-adaptive multi attention (SMA) module is introduced to adaptively leverage multiple attention mechanisms for better segmentation. The convolutional-based position and channel attention mechanisms with a patch tokenization-based vision transformer (ViT)-based attention mechanism in a hybrid and end-to-end manner are integrated into the SMA. The CNN- and ViT-based attentions mine the short- and long-range dependencies for more precise segmentation. The SMA module is applied in an encoder-decoder structure with a ResNet50 backbone named CardSegNet. Furthermore, a deep supervision method with multi-loss functions is introduced to the CardSegNet optimizer to reduce overfitting and enhance the model's performance. The proposed model is validated on the ACDC2017 (n=100), M&Ms (n=321), and a local dataset (n=22) using the 10-fold cross-validation method with promising segmentation results, demonstrating its outperformance versus its counterparts. © 2024 Elsevier Ltd
Nazem, F. ,
Rasti, R. ,
Fassihi, A. ,
Dehnavi, A.M. ,
Ghasemi, F. Journal of Computational Science (18777503) 81
One of the critical aspects of structure-based drug design is to choose important druggable binding sites in the protein's crystallography structures. As experimental processes are costly and time-consuming, computational drug design using machine learning algorithms is recommended. Over recent years, deep learning methods have been utilized in a wide variety of research applications such as binding site prediction. In this study, a new combination of attention blocks in the 3D U-Net model based on semantic segmentation methods is used to improve localization of pocket prediction. The attention blocks are tuned to find which point and channel of features should be emphasized along spatial and channel axes. Our model's performance is evaluated through extensive experiments on several datasets from different sources, and the results are compared to the most recent deep learning-based models. The results indicate the proposed attention model (Att-UNet) can predict binding sites accurately, i.e. the overlap of the predicted pocket using the proposed method with the true binding site shows statistically significant improvement when compared to other state-of-the-art models. The attention blocks may help the model focus on the target structure by suppressing features in irrelevant regions. © 2024
Nikzad, M.H. ,
Heidari-rarani, M. ,
Momenzadeh-kholenjani, A. ,
Rasti, R. Materials Today Communications (23524928) 39
Alloys are engineered materials aimed at enhancing mechanical properties. Extensive research has focused on identifying the optimal metal composition for alloys with superior tensile strength. This study validates the stiffness and strength values of an aluminum-copper alloy through a comparison with a molecular dynamics simulation. Subsequently, 100 data points were obtained from the simulation, and a deep neural network (DNN) with three hidden layers was employed. The DNN was trained, tested, and its structure optimized using the Taguchi design of experiment. The proposed DNN structures successfully predicted the maximum values of the stiffness and strength, which were further verified using molecular dynamics simulation. Notably, the results demonstrated the complete reliability of the Taguchi-designed DNN algorithm in this application. © 2024 Elsevier Ltd
Nazem, Fatemeh ,
Ghasemi, Fahimeh ,
Fassihi, Afshin ,
Rasti, Reza ,
Nazem, F. ,
Ghasemi, F. ,
Fassihi, A. ,
Rasti, R. ,
Dehnavi, A.M. JOURNAL OF MEDICAL SIGNALS & SENSORS (22287477) 13(1)pp. 1-10
Background: The first step in developing new drugs is to find binding sites for a protein structure that can be used as a starting point to design new antagonists and inhibitors. The methods relying on convolutional neural network for the prediction of binding sites have attracted much attention. This study focuses on the use of optimized neural network for three-dimensional (3D) non-Euclidean data. Methods: A graph, which is made from 3D protein structure, is fed to the proposed GU-Net model based on graph convolutional operation. The features of each atom are considered as attributes of each node. The results of the proposed GU-Net are compared with a classifier based on random forest (RF). A new data exhibition is used as the input of RF classifier. Results: The performance of our model is also examined through extensive experiments on various datasets from other sources. GU-Net could predict the more number of pockets with accurate shape than RF. Conclusions: This study will enable future works on a better modeling of protein structures that will enhance knowledge of proteomics and offer deeper insight into drug design process.
Rasti, R. ,
Biglari, A. ,
Rezapourian, M. ,
Yang, Z. ,
Farsiu, S. Ieee Transactions On Medical Imaging (02780062) 42(5)pp. 1413-1423
Optical coherence tomography (OCT) helps ophthalmologists assess macular edema, accumulation of fluids, and lesions at microscopic resolution. Quantification of retinal fluids is necessary for OCT-guided treatment management, which relies on a precise image segmentation step. As manual analysis of retinal fluids is a time-consuming, subjective, and error-prone task, there is increasing demand for fast and robust automatic solutions. In this study, a new convolutional neural architecture named RetiFluidNet is proposed for multi-class retinal fluid segmentation. The model benefits from hierarchical representation learning of textural, contextual, and edge features using a new self-adaptive dual-attention (SDA) module, multiple self-adaptive attention-based skip connections (SASC), and a novel multi-scale deep self-supervision learning (DSL) scheme. The attention mechanism in the proposed SDA module enables the model to automatically extract deformation-aware representations at different levels, and the introduced SASC paths further consider spatial-channel interdependencies for concatenation of counterpart encoder and decoder units, which improve representational capability. RetiFluidNet is also optimized using a joint loss function comprising a weighted version of dice overlap and edge-preserved connectivity-based losses, where several hierarchical stages of multi-scale local losses are integrated into the optimization process. The model is validated based on three publicly available datasets: RETOUCH, OPTIMA, and DUKE, with comparisons against several baselines. Experimental results on the datasets prove the effectiveness of the proposed model in retinal OCT fluid segmentation and reveal that the suggested method is more effective than existing state-of-the-art fluid segmentation algorithms in adapting to retinal OCT scans recorded by various image scanning instruments. © 1982-2012 IEEE.
Biomedical Signal Processing and Control (17468108) 78
Proposing a practical method for high-performance emotion recognition could facilitate human–computer interaction. Among existing methods, deep learning techniques have improved the performance of emotion recognition systems. In this work, a new multimodal neural design is presented wherein audio and visual data are combined as the input to a hybrid network comprised of a bidirectional long short term memory (BiLSTM) network and two convolutional neural networks (CNNs). The spatial and temporal features extracted from video frames are fused with Mel-Frequency Cepstral Coefficients (MFCCs) and energy features extracted from audio signals and BiLSTM network outputs. Finally, a Softmax classifier is used to classify inputs into the set of target categories. The proposed model is evaluated on Surrey Audio–Visual Expressed Emotion (SAVEE), Ryerson Audio–Visual Database of Emotional Speech and Song (RAVDESS), and Ryerson Multimedia research Lab (RML) databases. Experimental results on these datasets prove the effectiveness of the proposed model where it achieves the accuracy of 99.75%, 94.99%, and 99.23% for the SAVEE, RAVDESS, and RML databases, respectively. Our experimental study reveals that the suggested method is more effective than existing algorithms in adapting to emotion recognition in these datasets. © 2022 Elsevier Ltd
Accurate segmentation of the hippocampus head and body from MR images is routinely performed to examine the link between the hippocampus deformation and neurological diseases, such as Alzheimer's and epilepsy. State-of-the-art seminal segmentation methods (including hippocampus segmentation) are based on deep learning algorithms. Most studies focused on developing robust deep learning algorithms to achieve satisfactory performance in hippocampus body and head segmentation. A critical aspect that has been overlooked in these studies is the strategy adopted to train deep learning algorithms when there are two or more target structures. In this work, we examine which deep learning training strategies would be more effective. These strategies include simultaneous (parallel segmentation of the head and body), serial (first head and then body), independent, and attention-based training and segmentation of the target structures. To this end, the hippocampus dataset from the Decathlon challenge and a residual neural network (Resnet) were employed to compare the above-mentioned strategies for hippocampus head and body segmentation. The Dice similarity coefficient and Hausdorff distance were calculated for the outcome of each strategy versus the manually defined hippocampus head and body masks. The quantitative analysis of the outcomes of different training frameworks demonstrated the superior performance of the attention-based training framework with Dice index of 0.89±0.03 (body) and 0.88±0.04 (head) compared to simultaneous, serial, and independent training frameworks with Dice indices of 0.88±0.04 (body) and 0.87±0.04 (head), 0.88±0.04 (body) and 0.87±0.04 (head), and 0.88±0.04 (body) and 0.86±0.04 (head), respectively. The statistical analysis demonstrated the significantly superior performance of the attention-based training framework (p-value<0.0001). In conclusion, the attention-based training framework is recommended for multi-structure seminal segmentation. © 2021 IEEE.
Song, Z. ,
Xu, L. ,
Wang, J. ,
Rasti, R. ,
Sastry, A. ,
Li, J.D. ,
Raynor, W. ,
Izatt, J.A. ,
Toth, C.A. ,
Vajzovic, L. American Journal of Ophthalmology (00029394) 221pp. 154-168
Purpose: Subretinal injections of therapeutics are commonly used to treat ocular diseases. Accurate dosing of therapeutics at target locations is crucial but difficult to achieve using subretinal injections due to leakage, and there is no method available to measure the volume of therapeutics successfully administered to the subretinal location during surgery. Here, we introduce the first automatic method for quantifying the volume of subretinal blebs, using porcine eyes injected with Ringer's lactate solution as samples. Design: Ex vivo animal study. Methods: Microscope-integrated optical coherence tomography was used to obtain 3D visualization of subretinal blebs in porcine eyes at Duke Eye Center. Two different injection phases were imaged and analyzed in 15 eyes (30 volumes), selected from a total of 37 eyes. The inclusion/exclusion criteria were set independently from the algorithm-development and testing team. A novel lightweight, deep learning–based algorithm was designed to segment subretinal bleb boundaries. A cross-validation method was used to avoid selection bias. An ensemble-classifier strategy was applied to generate final results for the test dataset. Results: The algorithm performs notably better than 4 other state-of-the-art deep learning–based segmentation methods, achieving an F1 score of 93.86 ± 1.17% and 96.90 ± 0.59% on the independent test data for entry and full blebs, respectively. Conclusion: The proposed algorithm accurately segmented the volumetric boundaries of Ringer's lactate solution delivered into the subretinal space of porcine eyes with robust performance and real-time speed. This is the first step for future applications in computer-guided delivery of therapeutics into the subretinal space in human subjects. © 2020 Elsevier Inc.
Rasti, R. ,
Brännström, J. ,
Mårtensson, A. ,
Zenk, I. ,
Gantelius, J. ,
Gaudenzi, G. ,
Alvesson, H.M. ,
Alfvén, T. BMJ Open (20446055) 11(11)
Objectives In many resource-limited health systems, point-of-care tests (POCTs) are the only means for clinical patient sample analyses. However, the speed and simplicity of POCTs also makes their use appealing to clinicians in high-income countries (HICs), despite greater laboratory accessibility. Although also part of the clinical routine in HICs, clinician perceptions of the utility of POCTs are relatively unknown in such settings as compared with others. In a Swedish paediatric emergency department (PED) where POCT use is routine, we aimed to characterise healthcare providers’ perspectives on the clinical utility of POCTs and explore their implementation in the local setting; to discuss and compare such perspectives, to those reported in other settings; and finally, to gather requests for ideal novel POCTs. Design Qualitative focus group discussions study. A data-driven content analysis approach was used for analysis. Setting The PED of a secondary paediatric hospital in Stockholm, Sweden. Participants Twenty-four healthcare providers clinically active at the PED were enrolled in six focus groups. Results A range of POCTs was routinely used. The emerging theme Utility of our POCT use is double-edged illustrated the perceived utility of POCTs. While POCT services were considered to have clinical and social value, the local routine for their use was named to distract clinicians from the care for patients. Requests were made for ideal POCTs and their implementation. Conclusion Despite their clinical integration, deficient implementation routines limit the benefits of POCT services to this well-resourced paediatric clinic. As such deficiencies are shared with other settings, it is suggested that some characteristics of POCTs and of their utility are less related to resource level and more to policy deficiency. To address this, we propose the appointment of skilled laboratory personnel as ambassadors to hospital clinics offering POCT services, to ensure higher utility of such services. © 2021 BMJ Publishing Group. All rights reserved.
Rasti, R. ,
Allingham, M.J. ,
Mettu, P.S. ,
Kavusi, S. ,
Govind, K. ,
Cousins, S.W. ,
Farsiu, S. Biomedical Optics Express (21567085) 11(2)pp. 1139-1152
Anti-vascular endothelial growth factor (VEGF) agents are widely regarded as the first line of therapy for diabetic macular edema (DME) but are not universally effective. An automatic method that can predict whether a patient is likely to respond to anti-VEGF therapy can avoid unnecessary trial and error treatment strategies and promote the selection of more effective first-line therapies. The objective of this study is to automatically predict the efficacy of anti-VEGF treatment of DME in individual patients based on optical coherence tomography (OCT) images. We performed a retrospective study of 127 subjects treated for DME with three consecutive injections of anti-VEGF agents. Patients’ retinas were imaged using spectral-domain OCT (SD-OCT) before and after anti-VEGF therapy, and the total retinal thicknesses before and after treatment were extracted from OCT B-scans. A novel deep convolutional neural network was designed and evaluated using pre-treatment OCT scans as input and differential retinal thickness as output, with 5-fold cross-validation. The group of patients responsive to anti-VEGF treatment was defined as those with at least a 10% reduction in retinal thickness following treatment. The predictive performance of the system was evaluated by calculating the precision, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC). The algorithm achieved an average AUC of 0.866 in discriminating responsive from non-responsive patients, with an average precision, sensitivity, and specificity of 85.5%, 80.1%, and 85.0%, respectively. Classification precision was significantly higher when differentiating between very responsive and very unresponsive patients. The proposed automatic algorithm accurately predicts the response to anti-VEGF treatment in DME patients based on OCT images. This pilot study is a critical step toward using non-invasive imaging and automated analysis to select the most effective therapy for a patient’s specific disease condition. © 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Gaudenzi, G. ,
Kumbakumba, E. ,
Rasti, R. ,
Nanjebe, D. ,
Réu, P. ,
Nyehangane, D. ,
Mårtensson, A. ,
Nassejje, M. ,
Karlsson, J. ,
Mzee, J. JMIR Research Protocols (19290748) 9(11)
Background: A timely differential diagnostic is essential to identify the etiology of central nervous system (CNS) infections in children, in order to facilitate targeted treatment, manage patients, and improve clinical outcome. Objective: The Pediatric Infection-Point-of-Care (PI-POC) trial is investigating novel methods to improve and strengthen the differential diagnostics of suspected childhood CNS infections in low-income health systems such as those in Southwestern Uganda. This will be achieved by evaluating (1) a novel DNA-based diagnostic assay for CNS infections, (2) a commercially available multiplex PCR-based meningitis/encephalitis (ME) panel for clinical use in a facility-limited laboratory setting, (3) proteomics profiling of blood from children with severe CNS infection as compared to outpatient controls with fever yet not severely ill, and (4) Myxovirus resistance protein A (MxA) as a biomarker in blood for viral CNS infection. Further changes in the etiology of childhood CNS infections after the introduction of the pneumococcal conjugate vaccine against Streptococcus pneumoniae will be investigated. In addition, the carriage and invasive rate of Neisseria meningitidis will be recorded and serotyped, and the expression of its major virulence factor (polysaccharide capsule) will be investigated. Methods: The PI-POC trial is a prospective observational study of children including newborns up to 12 years of age with clinical features of CNS infection, and age-/sex-matched outpatient controls with fever yet not severely ill. Participants are recruited at 2 Pediatric clinics in Mbarara, Uganda. Cerebrospinal fluid (for cases only), blood, and nasopharyngeal (NP) swabs (for both cases and controls) sampled at both clinics are analyzed at the Epicentre Research Laboratory through gold-standard methods for CNS infection diagnosis (microscopy, biochemistry, and culture) and a commercially available ME panel for multiplex PCR analyses of the cerebrospinal fluid. An additional blood sample from cases is collected on day 3 after admission. After initial clinical analyses in Mbarara, samples will be transported to Stockholm, Sweden for (1) validation analyses of a novel nucleic acid-based POC test, (2) biomarker research, and (3) serotyping and molecular characterization of S. pneumoniae and N. meningitidis. Results: A pilot study was performed from January to April 2019. The PI-POC trial enrollment of patients begun in April 2019 and will continue until September 2020, to include up to 300 cases and controls. Preliminary results from the PI-POC study are expected by the end of 2020. Conclusions: The findings from the PI-POC study can potentially facilitate rapid etiological diagnosis of CNS infections in low-resource settings and allow for novel methods for determination of the severity of CNS infection in such environment. © 2020 JMIR Publications. All rights reserved.
Rasti, R. ,
Dehnavi, A.M. ,
Rabbani, H. ,
Hajizadeh, F. Journal Of Medical Signals And Sensors (22287477) 9(1)pp. 1-14
Background: Macular disorders, such as diabetic macular edema (DME) and age-related macular degeneration (AMD) are among the major ocular diseases. Having one of these diseases can lead to vision impairments or even permanent blindness in a not-so-long time span. So, the early diagnosis of these diseases are the main goals for researchers in the field. Methods: This study is designed in order to present a comparative analysis on the recent convolutional mixture of experts (CMoE) models for distinguishing normal macular OCT from DME and AMD. For this purpose, we considered three recent CMoE models called Mixture ensemble of convolutional neural networks (ME-CNN), Multiscale Convolutional Mixture of Experts (MCME), and Wavelet-based Convolutional Mixture of Experts (WCME) models. For this research study, the models were evaluated on a database of three differen macular OCT sets. Two first OCT sets were acquired by Heidelberg imaging systems consisting of 148 and 45 subjects respectively and set3 was constituted of 384 Bioptigen OCT acquisitions. To provide better performance insight into the CMoE ensembles, we extensively analyzed the models based on the 5-fold cross-validation method and various classification measures such as precision and average area under the ROC curve (AUC). Results: Experimental evaluations showed that the MCME and WCME outperformed the ME-CNN model and presented overall precisions of 98.14% and 96.06% for aligned OCTs respectively. For non-aligned retinal OCTs, these values were 93.95% and 95.56%. Conclusion: Based on the comparative analysis, although the MCME model outperformed the other CMoE models in the analysis of aligned retinal OCTs, the WCME offers a robust model for diagnosis of non-aligned retinal OCTs. This allows having a fast and robust computer-aided system in macular OCT imaging which does not rely on the routine computerized processes such as denoising, segmentation of retina layers, and also retinal layers alignment. © 2019 Journal of Medical Signals & Sensors.
Rhedin, S.A. ,
Eklundh, A. ,
Ryd-rinder, M. ,
Naucler, P. ,
Mårtensson, A. ,
Gantelius, J. ,
Zenk, I. ,
Andersson-svahn, H. ,
Nybond, S. ,
Rasti, R. JMIR Research Protocols (19290748) 8(4)
Background: There is a need to better distinguish viral infections from antibiotic-requiring bacterial infections in children presenting with clinical community-acquired pneumonia (CAP) to assist health care workers in decision making and to improve the rational use of antibiotics. Objective: The overall aim of the Trial of Respiratory infections in children for ENhanced Diagnostics (TREND) study is to improve the differential diagnosis of bacterial and viral etiologies in children aged below 5 years with clinical CAP, by evaluating myxovirus resistance protein A (MxA) as a biomarker for viral CAP and by evaluating an existing (multianalyte point-of-care antigen detection test system [mariPOC respi] ArcDia International Oy Ltd.) and a potential future point-of-care test for respiratory pathogens. Methods: Children aged 1 to 59 months with clinical CAP as well as healthy, hospital-based, asymptomatic controls will be included at a pediatric emergency hospital in Stockholm, Sweden. Blood (analyzed for MxA and C-reactive protein) and nasopharyngeal samples (analyzed with real-time polymerase chain reaction as the gold standard and antigen-based mariPOC respi test as well as saved for future analyses of a novel recombinase polymerase amplification-based point-of-care test for respiratory pathogens) will be collected. A newly developed algorithm for the classification of CAP etiology will be used as the reference standard. Results: A pilot study was performed from June to August 2017. The enrollment of study subjects started in November 2017. Results are expected by the end of 2019.Conclusions: The findings from the TREND study can be an important step to improve the management of children with clinical. © 2019 Journal of Medical Internet Research. All rights reserved.
Rivas, L. ,
Reuterswärd, P. ,
Rasti, R. ,
Herrmann, B. ,
Mårtensson, A. ,
Alfvén, T. ,
Gantelius, J. ,
Andersson-svahn, H. Talanta (00399140) 183pp. 192-200
Paper-based biosensors offer a promising technology to be used at the point of care, enabled by good performance, convenience and low-cost. In this article, we describe a colorimetric vertical-flow DNA microarray (DNA-VFM) that takes advantage of the screening capability of DNA microarrays in a paper format together with isothermal amplification by means of Recombinase Polymerase Amplification (RPA). Different assay parameters such as hybridization buffer, flow rate, printing buffer and capture probe concentration were optimized. A limit of detection (LOD) of 4.4 nM was achieved as determined by tabletop scanning. The DNA-VFM was applied as a proof of concept for detection of Neisseria meningitidis, a primary cause of bacterial meningitis. The LOD was determined to be between 38 and 2.1 × 106 copies/VFMassay, depending on the choice of DNA capture probes. The presented approach provides multiplex capabilities of DNA microarrays in a paper-based format for future point-of-care applications. © 2018 The Author(s)
Rasti, R. ,
Dehnavi, A.M. ,
Rabbani, H. ,
Hajizadeh, F. Journal of Biomedical Optics (15602281) 23(3)
The present research intends to propose a fully automatic algorithm for the classification of three-dimensional (3-D) optical coherence tomography (OCT) scans of patients suffering from abnormal macula from normal candidates. The method proposed does not require any denoising, segmentation, retinal alignment processes to assess the intraretinal layers, as well as abnormalities or lesion structures. To classify abnormal cases from the control group, a two-stage scheme was utilized, which consists of automatic subsystems for adaptive feature learning and diagnostic scoring. In the first stage, a wavelet-based convolutional neural network (CNN) model was introduced and exploited to generate B-scan representative CNN codes in the spatial-frequency domain, and the cumulative features of 3-D volumes were extracted. In the second stage, the presence of abnormalities in 3-D OCTs was scored over the extracted features. Two different retinal SD-OCT datasets are used for evaluation of the algorithm based on the unbiased fivefold cross-validation (CV) approach. The first set constitutes 3-D OCT images of 30 normal subjects and 30 diabetic macular edema (DME) patients captured from the Topcon device. The second publicly available set consists of 45 subjects with a distribution of 15 patients in age-related macular degeneration, DME, and normal classes from the Heidelberg device. With the application of the algorithm on overall OCT volumes and 10 repetitions of the fivefold CV, the proposed scheme obtained an average precision of 99.33% on dataset1 as a two-class classification problem and 98.67% on dataset2 as a three-class classification task. © 2018 Society of Photo-Optical Instrumentation Engineers (SPIE).
Rasti, R. ,
Rabbani, H. ,
Dehnavi, A.M. ,
Hajizadeh, F. Ieee Transactions On Medical Imaging (02780062) 37(4)pp. 1024-1034
Computer-aided diagnosis (CAD) of retinal pathologies is a current active area in medical image analysis. Due to the increasing use of retinal optical coherence tomography (OCT) imaging technique, a CAD system in retinal OCT is essential to assist ophthalmologist in the early detection of ocular diseases and treatment monitoring. This paper presents a novel CAD system based on a multi-scale convolutional mixture of expert (MCME) ensemble model to identify normal retina, and two common types of macular pathologies, namely, dry age-related macular degeneration, and diabetic macular edema. The proposed MCME modular model is a data-driven neural structure, which employs a new cost function for discriminative and fast learning of image features by applying convolutional neural networks on multiple-scale sub-images. MCME maximizes the likelihood function of the training data set and ground truth by considering a mixture model, which tries also to model the joint interaction between individual experts by using a correlated multivariate component for each expert module instead of only modeling the marginal distributions by independent Gaussian components. Two different macular OCT data sets from Heidelberg devices were considered for the evaluation of the method, i.e., a local data set of OCT images of 148 subjects and a public data set of 45 OCT acquisitions. For comparison purpose, we performed a wide range of classification measures to compare the results with the best configurations of the MCME method. With the MCME model of four scale-dependent experts, the precision rate of 98.86%, and the area under the receiver operating characteristic curve (AUC) of 0.9985 were obtained on average. © 1982-2012 IEEE.
Pattern Recognition (00313203) 72pp. 381-390
This work addresses a novel computer-aided diagnosis (CAD) system in breast dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). The CAD system is designed based on a mixture ensemble of convolutional neural networks (ME-CNN) to discriminate between benign and malignant breast tumors. The ME-CNN is a modular and image-based ensemble, which can stochastically partition the high-dimensional image space through simultaneous and competitive learning of its modules. The proposed system was assessed on our database of 112 DCE-MRI studies including solid breast masses, using a wide range of classification measures. The ME-CNN model composed of three CNN experts and one convolutional gating network achieves an accuracy of 96.39%, a sensitivity of 97.73% and a specificity of 94.87%. The experimental results also show that it has competitive classification performances compared to three existing single-classifier methods and two convolutional ensemble methods. The proposed ME-CNN model could provide an effective tool for radiologists to analyse breast DCE-MRI images. © 2017
Rasti, R. ,
Nanjebe, D. ,
Karlström, J. ,
Muchunguzi, C. ,
Mwanga-amumpaire, J. ,
Gantelius, J. ,
Mårtensson, A. ,
Rivas, L. ,
Galban, F. ,
Reuterswärd, P. PLoS ONE (19326203) 12(7)
Background: Point-of-care (POC) tests have become increasingly available and more widely used in recent years. They have been of particular importance to low-income settings, enabling them with clinical capacities that had previously been limited. POC testing programs hold a great potential for significant improvement in low-income health systems. However, as most POC tests are developed in high-income countries, disengagement between developers and end-users inhibit their full potential. This study explores perceptions of POC test end-users in a low-income setting, aiming to support the development of novel POC tests for low-income countries. Methods: A qualitative study was conducted in Mbarara District, Southwestern Uganda, in October 2014. Fifty health care workers were included in seven focus groups, comprising midwives, laboratory technicians, clinical and medical officers, junior and senior nurses, and medical doctors. Discussions were audio-recorded and transcribed verbatim. Transcripts were coded through a data-driven approach for qualitative content analysis. Results: Nineteen different POC tests were identified as currently being in use. While participants displayed being widely accustomed to and appreciative of the use of POC tests, they also assessed the use and characteristics of current tests as imperfect. An ideal POC test was characterized as being adapted to local conditions, thoughtfully implemented in the specific health system, and capable of improving the care of patients. Tests for specific medical conditions were requested. Opinions differed with regard to the ideal distribution of POC tests in the local health system. Conclusion POC tests are commonly used and greatly appreciated in this study setting. However, there are dissatisfactions with current POC tests and their use. To maximize benefit, stakeholders need to include end-user perspectives in the development and implementation of POC tests. Insights from this study will influence our ongoing efforts to develop POC tests that will be particularly usable in low-income settings. © 2017 Rasti et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Rasti, R. ,
Dehnavi, A.M. ,
Rabbani, H. ,
Hajizadeh, F. Iranian Conference on Machine Vision and Image Processing, MVIP (21666776) 2017pp. 192-196
This paper presents a new fully automatic algorithm for classification of 3D Optical Coherence Tomography (OCT) images as Age-related Macular Degeneration (AMD), Diabetic Macular Edema (DME), and healthy people. The proposed algorithm does not need to any retinal layer alignment and also segmentation processes (e.g., segmentation of intra-retinal layers and lesion structures). The algorithm utilizes a new Wavelet-based Convolutional Mixture of Experts (WCME) model as an adaptive feature extraction and classification method. The WCME benefits from spatial-frequency decomposition and also an ensemble of convolutional neural networks (CNNs) to build a high-level representation of OCT data. In this study, a retinal OCT data set constituted of 148 cases is used for evaluation of the method based on unbiased cross-validation approach. The dataset consists of 50 normal, 50 DME, and 48 AMD OCT acquisitions from Heidelberg device. With the proposed WCME model, the overall algorithm accurately classified the OCT data with a precision rate of 95.21% and an area under the ROC curve (AUC) of 0.986. © 2017 IEEE.
Journal of Isfahan Medical School (10277595) 34(404)pp. 1256-1261
Background: When two eyes are faced with two completely different sights, corresponding areas of the two eyes retina triggered with entirely two different stimulus and instead of receiving a fixed and stable image of two stimulus, brain perceive the alternation of the two images. This phenomenon is known as an intelligent tool to study cognitive processes. Methods: Taking a network of excitatory and inhibitory Hodgkin-Huxley type neuron in network architecture of visual cortex, the behavior of coupled spiking neuron in cortical brain that have a direct effect on the emergence of this phenomenon was modeled. To evaluate the performance of this model, direction lines as the two stimuli were applied to the eyes. Findings: Rebuilding the predominance time of each eye, and independence between the two successive dominance times were reported by this model. In addition, the effect of changes in the stimulus applied to each of the two eyes was studied and it was figured out that applied changing stimulus strength to one eye was affected only on dominance time of the other eye. Besides, increasing stimulus strength of both eyes decreased mean dominance time. Conclusion: Obtained results of this model were consistent with known experimental result for this phenomenon and many of dynamics features of this phenomenon reconstructed. Discussed theories in this field were also confirmed. © 2016, Isfahan University of Medical Sciences(IUMS). All rights reserved.
