Your browser doesn't support javascript.
loading
Mostrar: 20 | 50 | 100
Resultados 1 - 20 de 143.876
Filtrar
1.
Data Brief ; 54: 110486, 2024 Jun.
Artigo em Inglês | MEDLINE | ID: mdl-38770039

RESUMO

Deep learning has been studied in recent years to identify periapical lesions- a significant indicator of periapical periodontitis in radiographs. An accurate dataset is essential for constructing an efficient learning model for detecting periapical lesions. In order to achieve this goal, we gathered and created a database of panoramic radiographs containing periapical lesions from the High-quality Dental Treatment Centre, School of Dentistry, Hanoi Medical University, between January 2016 and March 2021. Out of 16,519 radiographs, three experienced dentists identified 3,926 images of periapical lesions and annotated those lesions based on the Periapical Lesions Classification. By applying well-known data processing techniques (e.g. scaling, mirroring, and flipping), the amount of data is increased to 17,004 images through generating additional images for machine learning. The dataset has three folders: one for the original photos, one for the post-augmentation images, and the rest for the annotation of periapical lesions. The information could assist researchers in developing a predictive machine model for detecting periapical lesions in radiographs.

2.
Cureus ; 16(5): e60675, 2024 May.
Artigo em Inglês | MEDLINE | ID: mdl-38770053

RESUMO

The performance of two artificial intelligence (AI) platforms, ChatGPT 3.5 (OpenAI, California, United States) and Gemini (Google AI, California, United States) was assessed by answering 200 questions of microbiology drawn from validated sources. The questions were selected from topics such as General Microbiology, Immunology, and Microbiology Applied to Infectious Diseases. The study was conducted from December 2023 to March 2024, and the responses of the different AI platforms were compared with an answer key. Statistical analysis was performed to assess accuracy. ChatGPT 3.5 and Gemini had comparable accuracy with correct response scores of 71% and 70.5%, respectively. Their performance varied across different sections. Gemini performed better in General Microbiology and Immunology, and ChatGPT 3.5 had a better score in the Applied Microbiology section. The study's findings highlight that AI platforms such as ChatGPT and Gemini can be utilized in microbiology and medical education. The evolution and continuous updating of AI platforms are required to improve their performance.

3.
Front Cell Dev Biol ; 12: 1401945, 2024.
Artigo em Inglês | MEDLINE | ID: mdl-38770150

RESUMO

Background: Cutaneous melanoma is a highly heterogeneous cancer, and understanding the role of inflammation-related genes in its progression is crucial. Methods: The cohorts used include the TCGA cohort from TCGA database, and GSE115978, GSE19234, GSE22153 cohort, and GSE65904 cohort from GEO database. Weighted Gene Coexpression Network Analysis (WGCNA) identified key inflammatory modules. Machine learning techniques were employed to construct prognostic models, which were validated across multiple cohorts, including the TCGA cohort, GSE19234, GSE22153, and GSE65904. Immune cell infiltration, tumor mutation load, and immunotherapy response were assessed. The hub gene STAT1 was validated through cellular experiments. Results: Single-cell analysis revealed heterogeneity in inflammation-related genes, with NK cells, T cells, and macrophages showing elevated inflammation-related scores. WGCNA identified a module highly associated with inflammation. Machine learning yielded a CoxBoost + GBM prognostic model. The model effectively stratified patients into high-risk and low-risk groups in multiple cohorts. A nomogram and Receiver Operating Characteristic (ROC) curves confirmed the model's accuracy. Low-risk patients exhibited increased immune cell infiltration, higher Tumor Mutational Burden (TMB), and potentially better immunotherapy response. Cellular experiments validated the functional role of STAT1 in melanoma progression. Conclusion: Inflammation-related genes play a critical role in cutaneous melanoma progression. The developed prognostic model, nomogram, and validation experiments highlight the potential clinical relevance of these genes and provide a basis for further investigation into personalized treatment strategies for melanoma patients.

4.
J Inflamm Res ; 17: 3043-3055, 2024.
Artigo em Inglês | MEDLINE | ID: mdl-38770175

RESUMO

Background: With the aging of the population and the increasing incidence of neurological diseases, amnestic mild cognitive impairment (aMCI) has attracted attention. Hyperbaric oxygen (HBO) has gradually shown the potential in the treatment of aMCI as an emerging treatment method in recent times. This study is to observe the effect of HBO on the long-term learning memory of aMCI rats, and investigate the associated mechanisms. Methods: Seventy-two male rats (4-month-old) were randomly divided into control (CON) group, aMCI group, HBO group, 24 rats in each group. Each group was randomly divided into CON1, CON7, CON28; aMCI1, aMCI7, aMCI28; HBO1, HBO7, HBO28, 8 rats in each group. The aMCI model rats were established in aMCI and HBO groups. HBO group was treated with HBO for 7 days. The ethological and cytopathology which include Morris water maze (MWM) test, HE staining, TUNEL staining and the expression of Fas/FasL on neuron membrane were conducted to evaluate the effects of HBO on day 1, day 7 and day 28 after HBO treatment. Results: MWM test showed that the spatial learning and memory ability of the rats decreased in aMCI group, and recovered in HBO group; Compared with aMCI group, the pathological damage of hippocampal nerve cells was alleviated, the number of apoptotic cells was significantly reduced (P < 0.05), and the expression of Fas/FasL on the surface of nerve cell membrane was significantly weakened in HBO group (P < 0.05). There were no significant changes in the spatial learning and memory ability, pathological damage of hippocampal neurons, the number of apoptotic cells, and the changes of Fas/FasL on the surface of hippocampal neurons in HBO1, HBO7, and HBO28 groups (P > 0.05). However, in aMCI1, aMCI7, and aMCI28 groups gradually aggravated (P < 0.05). Conclusion: 1. HBO can improve the long-term learning and memory impairment by attenuating neuronal apoptosis in aMCI rats. 2. Fas/FasL mediated cell receptor death pathway is involved in the apoptosis of hippocampal neurons.

5.
Int J Public Health ; 69: 1606855, 2024.
Artigo em Inglês | MEDLINE | ID: mdl-38770181

RESUMO

Objectives: Suicide risk is elevated in lesbian, gay, bisexual, and transgender (LGBT) individuals. Limited data on LGBT status in healthcare systems hinder our understanding of this risk. This study used natural language processing to extract LGBT status and a deep neural network (DNN) to examine suicidal death risk factors among US Veterans. Methods: Data on 8.8 million veterans with visits between 2010 and 2017 was used. A case-control study was performed, and suicide death risk was analyzed by a DNN. Feature impacts and interactions on the outcome were evaluated. Results: The crude suicide mortality rate was higher in LGBT patients. However, after adjusting for over 200 risk and protective factors, known LGBT status was associated with reduced risk compared to LGBT-Unknown status. Among LGBT patients, black, female, married, and older Veterans have a higher risk, while Veterans of various religions have a lower risk. Conclusion: Our results suggest that disclosed LGBT status is not directly associated with an increase suicide death risk, however, other factors (e.g., depression and anxiety caused by stigma) are associated with suicide death risks.


Assuntos
Inteligência Artificial , Minorias Sexuais e de Gênero , Suicídio , Veteranos , Humanos , Masculino , Feminino , Minorias Sexuais e de Gênero/estatística & dados numéricos , Minorias Sexuais e de Gênero/psicologia , Pessoa de Meia-Idade , Estudos de Casos e Controles , Suicídio/estatística & dados numéricos , Veteranos/psicologia , Veteranos/estatística & dados numéricos , Estados Unidos/epidemiologia , Adulto , Fatores de Risco , Idoso , Processamento de Linguagem Natural
6.
Heliyon ; 10(10): e30741, 2024 May 30.
Artigo em Inglês | MEDLINE | ID: mdl-38770284

RESUMO

The supracondylar fracture of the child is a common fracture. Its physiology, physiopathology and treatment use periosteum. As far as we know, there is no 3D printed model of this typical fracture in children with periosteum. The purposes of the research are to present the concept of an educational 3D printed supra condylar model with periosteum of the child and its expert validation. Materials and methods: The basis for the paediatric elbow model was a 3D scan of a four-year-old girl. Once the data had been extracted, the models were constructed using 3D Slicer®, Autodesk fusion 360® and Ultimaker Cura® software's. The Creality 3D Ender 6 SE Printer® used PLA filaments to print bone and TPU for periosteum. Printing took place at the University Hospital and the steps were modelled by hand. 3D printed bones and 3D printed periosteum were manually assembled. Participants: Expert validation with twelve paediatric orthopaedic surgeons took place in three University hospitals of the North of France. Results: Four Lagrange and Rigault 3D printed models of supracondylar fractures with periosteum were obtained with 200 h of design, printing and manual assembly based on a four-year-old elbow. According to the paediatric orthopaedic surgery experts, the size of the model is very good, but the model itself is of little interest compared to the information provided by the reconstruction of a 3D scanner. In total, with 9 out of 12 questions scoring higher than 8/10, the model was considered to be a good model for informing parents and teaching students. Conclusions: This study details the design of the first 3D-printed supra condylar fracture model in children with a full-size physeal and periosteum. The model has been validated by paediatric orthopaedic surgery experts.

7.
Heliyon ; 10(10): e31010, 2024 May 30.
Artigo em Inglês | MEDLINE | ID: mdl-38770294

RESUMO

Purpose: To evaluate the feasibility of rib fracture detection in low-dose computed tomography (CT) images with a RetinaNet-based approach and to evaluate the potential of lowdose CT for rib fracture detection compared with regular-dose CT images. Materials and methods: The RetinaNet-based deep learning model was trained using 7300 scans with 50,410 rib fractures that were used as internal training datasets from four multicenter. The external test datasets consisted of both regular-dose and low-dose chest-abdomen CT images of rib fractures; the MICCAI 2020 RibFrac Challenge Dataset was used as the public dataset. Radiologists' interpretations were used as reference standards. The performance of the model in rib fracture detection was compared with the radiologists' interpretation. Results: In total, 728 traumatic rib fractures of 100 patients [60 men (60 %); mean age, 53.45 ± 11.19 (standard deviation (SD)); range, 18-77 years] were assessed in the external datasets. In these patients, the regular-dose group had a mean CT dose index volume (CTDIvol) of 7.18 mGy (SD: 2.22) and a mean dose length product (DLP) of 305.38 mGy cm (SD: 95.31); the low-dose group had a mean CTDIvol of 2.79 mGy (SD: 1.11) and a mean DLP of 131.52 mGy cm (SD: 55.58). The sensitivity of the RetinaNet-based model and that of the radiologists was 0.859 and 0.721 in the low-dose CT images and 0.886 and 0.794 in the regular-dose CT images, respectively. Conclusions: These findings indicate that the RetinaNet-based model can detect rib fractures in low-dose CT images with a robust performance, indicating its feasibility in assisting radiologists with rib fracture diagnosis.

8.
Heliyon ; 10(10): e30763, 2024 May 30.
Artigo em Inglês | MEDLINE | ID: mdl-38770315

RESUMO

Accurate delineation of Gross Tumor Volume (GTV) is crucial for radiotherapy. Deep learning-driven GTV segmentation technologies excel in rapidly and accurately delineating GTV, providing a basis for radiologists in formulating radiation plans. The existing 2D and 3D segmentation models of GTV based on deep learning are limited by the loss of spatial features and anisotropy respectively, and are both affected by the variability of tumor characteristics, blurred boundaries, and background interference. All these factors seriously affect the segmentation performance. To address the above issues, a Layer-Volume Parallel Attention (LVPA)-UNet model based on 2D-3D architecture has been proposed in this study, in which three strategies are introduced. Firstly, 2D and 3D workflows are introduced in the LVPA-UNet. They work in parallel and can guide each other. Both the fine features of each slice of 2D MRI and the 3D anatomical structure and spatial features of the tumor can be extracted by them. Secondly, parallel multi-branch depth-wise strip convolutions adapt the model to tumors of varying shapes and sizes within slices and volumetric spaces, and achieve refined processing of blurred boundaries. Lastly, a Layer-Channel Attention mechanism is proposed to adaptively adjust the weights of slices and channels according to their different tumor information, and then to highlight slices and channels with tumor. The experiments by LVPA-UNet on 1010 nasopharyngeal carcinoma (NPC) MRI datasets from three centers show a DSC of 0.7907, precision of 0.7929, recall of 0.8025, and HD95 of 1.8702 mm, outperforming eight typical models. Compared to the baseline model, it improves DSC by 2.14 %, precision by 2.96 %, and recall by 1.01 %, while reducing HD95 by 0.5434 mm. Consequently, while ensuring the efficiency of segmentation through deep learning, LVPA-UNet is able to provide superior GTV delineation results for radiotherapy and offer technical support for precision medicine.

9.
Heliyon ; 10(10): e30866, 2024 May 30.
Artigo em Inglês | MEDLINE | ID: mdl-38770317

RESUMO

The nuclear reactor control unit employs human factor engineering to ensure efficient operations and prevent any catastrophic incidents. This sector is of utmost importance for public safety. This study focuses on simulated analysis of specific areas of nuclear reactor control, specifically administration, operation, and maintenance, using artificial intelligence software. The investigation yields effective artificial intelligence algorithms that capture the essential and non-essential components of numerous parameters to be monitored in nuclear reactor control. The investigation further examines the interdependencies between various parameters and validates the statistical outputs of the model through attribution analysis. Furthermore, a Multivariant ANOVA analysis is conducted to identify the interactive plots and mean plots of crucial parameters interactions. The artificial intelligence algorithms demonstrate the correlation between the number of vacant staff jobs and both the frequency of license event reports each year and the ratio of contract employees to regular employees in the administrative domain. An AI method uncovers the relationships between the operator failing rate (OFR), operator processed errors (OEE), and operations at limited time frames (OLC). The AI algorithm reveals the interdependence between equipment in the out of service (EOS), progressive maintenance schedule (PRMR), and preventive maintenance schedules (PMRC). Effective machine learning neural network models are derived from generative adversarial network (GAN) algorithms and proposed for administrative, operational and maintenance loops of nuclear reactor control unit.

10.
Front Hum Neurosci ; 18: 1348154, 2024.
Artigo em Inglês | MEDLINE | ID: mdl-38770396

RESUMO

Introduction: The primary objective of this research is to examine acrophobia, a widely prevalent and highly severe phobia characterized by an overwhelming dread of heights, which has a substantial impact on a significant proportion of individuals worldwide. The objective of our study was to develop a real-time and precise instrument for evaluating levels of acrophobia by utilizing electroencephalogram (EEG) signals. Methods: EEG data was gathered from a sample of 18 individuals diagnosed with acrophobia. Subsequently, a range of classifiers, namely Support Vector Classifier (SVC), K-nearest Neighbors (KNN), Random Forest (RF), Decision Tree (DT), Adaboost, Linear Discriminant Analysis (LDA), Convolutional Neural Network (CNN), and Artificial Neural Network (ANN), were employed in the analysis. These methodologies encompass both machine learning (ML) and deep learning (DL) techniques. Results: The Convolutional Neural Network (CNN) and Artificial Neural Network (ANN) models demonstrated notable efficacy. The Convolutional Neural Network (CNN) model demonstrated a training accuracy of 96% and a testing accuracy of 99%, whereas the Artificial Neural Network (ANN) model attained a training accuracy of 96% and a testing accuracy of 97%. The findings of this study highlight the effectiveness of the proposed methodology in accurately categorizing real-time degrees of acrophobia using EEG data. Further investigation using correlation matrices for each level of acrophobia showed substantial EEG frequency band connections. Beta and Gamma mean values correlated strongly, suggesting cognitive arousal and acrophobic involvement could synchronize activity. Beta and Gamma activity correlated strongly with acrophobia, especially at higher levels. Discussion: The results underscore the promise of this innovative approach as a dependable and sophisticated method for evaluating acrophobia. This methodology has the potential to make a substantial contribution toward the comprehension and assessment of acrophobia, hence facilitating the development of more individualized and efficacious therapeutic interventions.

11.
Front Genet ; 15: 1393406, 2024.
Artigo em Inglês | MEDLINE | ID: mdl-38770419

RESUMO

Motivation: In recent years, there have been significant advances in various chromatin conformation capture techniques, and annotating the topological structure from Hi-C contact maps has become crucial for studying the three-dimensional structure of chromosomes. However, the structure and function of chromatin loops are highly dynamic and diverse, influenced by multiple factors. Therefore, obtaining the three-dimensional structure of the genome remains a challenging task. Among many chromatin loop prediction methods, it is difficult to fully extract features from the contact map and make accurate predictions at low sequencing depths. Results: In this study, we put forward a deep learning framework based on the diffusion model called CD-Loop for predicting accurate chromatin loops. First, by pre-training the input data, we obtain prior probabilities for predicting the classification of the Hi-C contact map. Then, by combining the denoising process based on the diffusion model and the prior probability obtained by pre-training, candidate loops were predicted from the input Hi-C contact map. Finally, CD-Loop uses a density-based clustering algorithm to cluster the candidate chromatin loops and predict the final chromatin loops. We compared CD-Loop with the currently popular methods, such as Peakachu, Chromosight, and Mustache, and found that in different cell types, species, and sequencing depths, CD-Loop outperforms other methods in loop annotation. We conclude that CD-Loop can accurately predict chromatin loops and reveal cell-type specificity. The code is available at https://github.com/wangyang199897/CD-Loop.

13.
Cureus ; 16(4): e58639, 2024 Apr.
Artigo em Inglês | MEDLINE | ID: mdl-38770467

RESUMO

Objective This study evaluated the potential of Chat Generative Pre-trained Transformer (ChatGPT) as an educational tool for neurosurgery residents preparing for the American Board of Neurological Surgery (ABNS) primary examination. Methods Non-imaging questions from the Congress of Neurological Surgeons (CNS) Self-Assessment in Neurological Surgery (SANS) online question bank were input into ChatGPT. Accuracy was evaluated and compared to human performance across subcategories. To quantify ChatGPT's educational potential, the concordance and insight of explanations were assessed by multiple neurosurgical faculty. Associations among these metrics as well as question length were evaluated. Results ChatGPT had an accuracy of 50.4% (1,068/2,120), with the highest and lowest accuracies in the pharmacology (81.2%, 13/16) and vascular (32.9%, 91/277) subcategories, respectively. ChatGPT performed worse than humans overall, as well as in the functional, other, peripheral, radiology, spine, trauma, tumor, and vascular subcategories. There were no subjects in which ChatGPT performed better than humans and its accuracy was below that required to pass the exam. The mean concordance was 93.4% (198/212) and the mean insight score was 2.7. Accuracy was negatively associated with question length (R2=0.29, p=0.03) but positively associated with both concordance (p<0.001, q<0.001) and insight (p<0.001, q<0.001). Conclusions The current study provides the largest and most comprehensive assessment of the accuracy and explanatory quality of ChatGPT in answering ABNS primary exam questions. The findings demonstrate shortcomings regarding ChatGPT's ability to pass, let alone teach, the neurosurgical boards.

14.
Proc Natl Acad Sci U S A ; 121(22): e2313496121, 2024 May 28.
Artigo em Inglês | MEDLINE | ID: mdl-38771874

RESUMO

Closing the achievement gap for minority students in higher education requires addressing the lack of belonging these students experience. This paper introduces a psychological intervention that strategically targets key elements within the learning environment to foster the success of minority students. The intervention sought to enhance Palestinian minority student's sense of belonging by increasing the presence of their native language. We tested the effectiveness of the intervention in two field experiments in Israel (n > 20,000), at the height of the COVID-19 pandemic when all classes were held via Zoom. Lecturers in the experimental condition added a transcript of their names in Arabic to their default display (English/Hebrew only). Our findings revealed a substantial and positive impact on Palestinian student's sense of belonging, class participation, and overall grades. In experiment 1, Palestinian student's average grade increased by 10 points. In experiment 2, there was an average increase of 4 points among Palestinian students' semester grade. Our intervention demonstrates that small institutional changes when carefully crafted can have a significant impact on minority populations. These results have significant implications for addressing educational disparities and fostering inclusive learning environment.


Assuntos
Árabes , COVID-19 , Grupos Minoritários , Estudantes , Humanos , Israel , Grupos Minoritários/educação , Grupos Minoritários/psicologia , Estudantes/psicologia , COVID-19/epidemiologia , Feminino , Árabes/psicologia , Masculino , Aprendizagem , Educação a Distância/métodos , SARS-CoV-2
15.
Proc Natl Acad Sci U S A ; 121(22): e2316300121, 2024 May 28.
Artigo em Inglês | MEDLINE | ID: mdl-38771876

RESUMO

The transition to remote learning in the context of COVID-19 led to dramatic setbacks in education. Is the return to in-person classes sufficient to eliminate these losses eventually? We study this question using data from the universe of secondary students in São Paulo State, Brazil. We estimate the causal medium-run impacts of the length of exposure to remote learning during the pandemic through a triple-differences strategy, which contrasts changes in educational outcomes across municipalities and grades that resumed in-person classes earlier (already by Q4/2020) or only in 2021. We find that relative learning losses from longer exposure to remote learning did not fade out over time-attesting that school reopening was at the same time key but not enough to mitigate accumulated learning losses in face of persistence. Using observational and experimental variation in local responses across 645 municipalities, we further document that remedial educational policies in the aftermath of the pandemic boosted learning recovery.


Assuntos
COVID-19 , Educação a Distância , Brasil/epidemiologia , COVID-19/epidemiologia , Humanos , Educação a Distância/métodos , Instituições Acadêmicas , SARS-CoV-2 , Pandemias , Estudantes , Aprendizagem , Adolescente
16.
Artigo em Inglês | MEDLINE | ID: mdl-38771915

RESUMO

INTRODUCTION: Autoimmune disorders affect 4.5% to 9.4% of children, significantly reducing their quality of life. The diagnosis and prognosis of autoimmune diseases are uncertain because of the variety of onset and development. Machine learning can identify clinically relevant patterns from vast amounts of data. Hence, its introduction has been beneficial in the diagnosis and management of patients. AREAS COVERED: This narrative review was conducted through searching various electronic databases, including PubMed, Scopus, and Web of Science This study thoroughly explores the current knowledge and identifies the remaining gaps in the applications of machine learning specifically in the context of pediatric autoimmune and related diseases. EXPERT OPINION: Machine learning algorithms have the potential to completely change how pediatric autoimmune disorders are identified, treated, and managed. Machine learning can assist physicians in making more precise and fast judgments, identifying new biomarkers and therapeutic targets, and personalizing treatment strategies for each patient by utilizing massive datasets and powerful analytics.

17.
Med Teach ; : 1-8, 2024 May 21.
Artigo em Inglês | MEDLINE | ID: mdl-38771960

RESUMO

PURPOSE: The concept of Entrustable Professional Activities (EPA) is increasingly used to operationalize learning in the clinical workplace, yet little is known about the emotions of learners feeling the responsibility when carrying out professional tasks. METHODS: We explored the emotional experiences of medical students in their final clerkship year when performing clinical tasks. We used an online reflective diary. Text entries were analysed using inductive-deductive content analysis with reference to the EPA framework and the control-value theory of achievement emotions. RESULTS: Students described a wide range of emotions related to carrying out various clinical tasks. They reported positive-activating emotions, ranging from enjoyment to relaxation, and negative-deactivating emotions, ranging from anxiety to boredom. Emotions varied across individual students and were related to the characteristics of a task, an increasing level of autonomy, the students' perceived ability to perform a task and the level of supervision provided. DISCUSSION: Emotions are widely present and impact on the workplace learning of medical students which is related to key elements of the EPA framework. Supervisors play a key role in eliciting positive-activating emotions and the motivation to learn by providing a level of supervision and guidance appropriate to the students' perceived ability to perform the task.

18.
Med Phys ; 2024 May 21.
Artigo em Inglês | MEDLINE | ID: mdl-38772037

RESUMO

BACKGROUND: Deformable registration is required to generate a time-integrated activity (TIA) map which is essential for voxel-based dosimetry. The conventional iterative registration algorithm using anatomical images (e.g., computed tomography (CT)) could result in registration errors in functional images (e.g., single photon emission computed tomography (SPECT) or positron emission tomography (PET)). Various deep learning-based registration tools have been proposed, but studies specifically focused on the registration of serial hybrid images were not found. PURPOSE: In this study, we introduce CoRX-NET, a novel unsupervised deep learning network designed for deformable registration of hybrid medical images. The CoRX-NET structure is based on the Swin-transformer (ST), allowing for the representation of complex spatial connections in images. Its self-attention mechanism aids in the effective exchange and integration of information across diverse image regions. To augment the amalgamation of SPECT and CT features, cross-stitch layers have been integrated into the network. METHODS: Two different 177 Lu DOTATATE SPECT/CT datasets were acquired at different medical centers. 22 sets from Seoul National University and 14 sets from Sunway Medical Centre are used for training/internal validation and external validation respectively. The CoRX-NET architecture builds upon the ST, enabling the modeling of intricate spatial relationships within images. To further enhance the fusion of SPECT and CT features, cross-stitch layers have been incorporated within the network. The network takes a pair of SPECT/CT images (e.g., fixed and moving images) and generates a deformed SPECT/CT image. The performance of the network was compared with Elastix and TransMorph using L1 loss and structural similarity index measure (SSIM) of CT, SSIM of normalized SPECT, and local normalized cross correlation (LNCC) of SPECT as metrics. The voxel-wise root mean square errors (RMSE) of TIA were compared among the different methods. RESULTS: The ablation study revealed that cross-stitch layers improved SPECT/CT registration performance. The cross-stitch layers notably enhance SSIM (internal validation: 0.9614 vs. 0.9653, external validation: 0.9159 vs. 0.9189) and LNCC of normalized SPECT images (internal validation: 0.7512 vs. 0.7670, external validation: 0.8027 vs. 0.8027). CoRX-NET with the cross-stitch layer achieved superior performance metrics compared to Elastix and TransMorph, except for CT SSIM in the external dataset. When qualitatively analyzed for both internal and external validation cases, CoRX-NET consistently demonstrated superior SPECT registration results. In addition, CoRX-NET accomplished SPECT/CT image registration in less than 6 s, whereas Elastix required approximately 50 s using the same PC's CPU. When employing CoRX-NET, it was observed that the voxel-wise RMSE values for TIA were approximately 27% lower for the kidney and 33% lower for the tumor, compared to when Elastix was used. CONCLUSION: This study represents a major advancement in achieving precise SPECT/CT registration using an unsupervised deep learning network. It outperforms conventional methods like Elastix and TransMorph, reducing uncertainties in TIA maps for more accurate dose assessments.

19.
Med Phys ; 2024 May 21.
Artigo em Inglês | MEDLINE | ID: mdl-38772044

RESUMO

BACKGROUND: Determining the optimal energy layer (EL) for each field, under considering both dose constraints and delivery efficiency, is crucial to promoting the development of proton arc therapy (PAT) technology. PURPOSE: This study aimed to explore the feasibility and potential clinical benefits of utilizing machine learning (ML) technique to automatically select EL for each field in PAT plans of lung cancer. METHODS: Proton Bragg peak position (BPP) was employed to characterize EL. The ground truth BPPs for each field were determined using the modified ELO-SPAT framework. Features in geometric, water-equivalent thicknesses (WET) and beamlet were defined and extracted. By analyzing the relationship between the extracted features and ground truth, a polynomial regression model with L2-norm regularization (Ridge regression) was constructed and trained. The performance of the regression model was reported as an error between the predictions and the ground truth. Besides, the predictions were used to make PAT plans (PAT_PRED). These plans were compared with those using the ground truth BPPs (PAT_TRUTH) and the mid-WET of the target volumes (PAT_MID) in terms of relative biological effectiveness-weighted dose (RWD) distributions. One hundred ten patients with lung cancer, a total of 7920 samples, were enrolled retrospectively, with 5940 cases randomly selected as the training set and the remaining 1980 cases as the testing set. Nine patients (648 samples) were collected additionally to evaluate the regression model in terms of plan quality and robustness. RESULTS: With regard to the prediction errors, the root mean squared errors and mean absolute errors between the ML-predicted and ground truth BPPs for the testing set were 9.165 and 6.572 mm, respectively, indicating differences of approximately two to three ELs. As for plan quality, the PAT_TRUTH and PAT_PRED plans performed similarly in terms of plan robustness, target coverage and organs at risk (OARs) protection, with differences smaller than 0.5 Gy(RBE). This trend was also observed for dose conformity and uniformity. The PAT_MID plans produced the lowest robustness index and lowest doses to OARs, along with the highest heterogeneity index, indicating better protection for OARs, improved plan robustness, but compromised dose homogeneity. Additionally, for relatively small tumor sizes, the PAT_MID plan demonstrated a notably poor dose conformity index. CONCLUSIONS: Within this cohort under investigation, our study demonstrated the feasibility of using ML technique to predict ELs for each field, offering a fast (within 2 s) and memory-efficient reduced way to select ELs for PAT plan.

20.
Comput Biol Med ; 176: 108609, 2024 May 14.
Artigo em Inglês | MEDLINE | ID: mdl-38772056

RESUMO

Semi-supervised medical image segmentation presents a compelling approach to streamline large-scale image analysis, alleviating annotation burdens while maintaining comparable performance. Despite recent strides in cross-supervised training paradigms, challenges persist in addressing sub-network disagreement and training efficiency and reliability. In response, our paper introduces a novel cross-supervised learning framework, Quality-driven Deep Cross-supervised Learning Network (QDC-Net). QDC-Net incorporates both an evidential sub-network and an vanilla sub-network, leveraging their complementary strengths to effectively handle disagreement. To enable the reliability and efficiency of semi-supervised training, we introduce a real-time quality estimation of the model's segmentation performance and propose a directional cross-training approach through the design of directional weights. We further design a truncated form of sample-wise loss weighting to mitigate the impact of inaccurate predictions and collapsed samples in semi-supervised training. Extensive experiments on LA and Pancreas-CT datasets demonstrate that QDC-Net surpasses other state-of-the-art methods in semi-supervised medical image segmentation. Code release is available at https://github.com/Medsemiseg.

SELEÇÃO DE REFERÊNCIAS
DETALHE DA PESQUISA
...