Med Biol Eng Comput. 2024 May 18. doi: 10.1007/s11517-024-03122-y. Online ahead of print.

ABSTRACT

Medical image denoising has been a subject of extensive research, with various techniques employed to enhance image quality and facilitate more accurate diagnostics. The evolution of denoising methods has highlighted impressive results but struggled to strike equilibrium between noise reduction and edge preservation which limits its applicability in various domains. This paper manifests the novel methodology that integrates an adaptive masking strategy, transformer-based U-Net Prior generator, edge enhancement module, and modified non-local block (MNLB) for denoising brain MRI clinical images. The adaptive masking strategy maintains the vital information through dynamic mask generation while the prior generator by capturing hierarchical features regenerates the high-quality prior MRI images. Finally, these images are fed to the edge enhancement module to boost structural information by maintaining crucial edge details, and the MNLB produces the denoised output by deriving non-local contextual information. The comprehensive experimental assessment is performed by employing two datasets namely the brain tumor MRI dataset and Alzheimer's dataset for diverse metrics and compared with conventional denoising approaches. The proposed denoising methodology achieves a PSNR of 40.965 and SSIM of 0.938 on the Alzheimer's dataset and also achieves a PSNR of 40.002 and SSIM of 0.926 on the brain tumor MRI dataset at a noise level of 50% revealing its supremacy in noise minimization. Furthermore, the impact of different masking ratios on denoising performance is analyzed which reveals that the proposed method showed PSNR of 40.965, SSIM of 0.938, MAE of 5.847, and MSE of 3.672 at the masking ratio of 60%. Moreover, the findings pave the way for the advancement of clinical image processing, facilitating precise detection of tumors in clinical MRI images.

PMID:38761289 | DOI:10.1007/s11517-024-03122-y

Aging (Albany NY). 2024 May 16;16. doi: 10.18632/aging.205835. Online ahead of print.

ABSTRACT

BACKGROUND: Valvular heart disease (VHD) is becoming increasingly important to manage the risk of future complications. Electrocardiographic (ECG) changes may be related to multiple VHDs, and (AI)-enabled ECG has been able to detect some VHDs. We aimed to develop five deep learning models (DLMs) to identify aortic stenosis, aortic regurgitation, pulmonary regurgitation, tricuspid regurgitation, and mitral regurgitation.

METHODS: Between 2010 and 2021, 77,047 patients with echocardiography and 12-lead ECG performed within 7 days were identified from an academic medical center to provide DLM development (122,728 ECGs), and internal validation (7,637 ECGs). Additional 11,800 patients from a community hospital were identified to external validation. The ECGs were classified as with or without moderate-to-severe VHDs according to transthoracic echocardiography (TTE) records, and we also collected the other echocardiographic data and follow-up TTE records to identify new-onset valvular heart diseases.

RESULTS: AI-ECG adjusted for age and sex achieved areas under the curves (AUCs) of >0.84, >0.80, >0.77, >0.83, and >0.81 for detecting aortic stenosis, aortic regurgitation, pulmonary regurgitation, tricuspid regurgitation, and mitral regurgitation, respectively. Since predictions of each DLM shared similar components of ECG rhythms, the positive findings of each DLM were highly correlated with other valvular heart diseases. Of note, a total of 37.5-51.7% of false-positive predictions had at least one significant echocardiographic finding, which may lead to a significantly higher risk of future moderate-to-severe VHDs in patients with initially minimal-to-mild VHDs.

CONCLUSION: AI-ECG may be used as a large-scale screening tool for detecting VHDs and a basis to undergo an echocardiography.

PMID:38761181 | DOI:10.18632/aging.205835

Acta Ophthalmol. 2024 May 18. doi: 10.1111/aos.16710. Online ahead of print.

ABSTRACT

PURPOSE: Age-related macular degeneration (AMD) is a complex eye disorder affecting millions worldwide. This article uses deep learning techniques to investigate the relationship between AMD, genetics and optical coherence tomography (OCT) scans.

METHODS: The cohort consisted of 332 patients, of which 235 were diagnosed with AMD and 97 were controls with no signs of AMD. The genome-wide association studies summary statistics utilized to establish the polygenic risk score (PRS) in relation to AMD were derived from the GERA European study. A PRS estimation based on OCT volumes for both eyes was performed using a proprietary convolutional neural network (CNN) model supported by machine learning models. The method's performance was assessed using numerical evaluation metrics, and the Grad-CAM technique was used to evaluate the results by visualizing the features learned by the model.

RESULTS: The best results were obtained with the CNN and the Extra Tree regressor (MAE = 0.55, MSE = 0.49, RMSE = 0.70, R2 = 0.34). Extending the feature vector with additional information on AMD diagnosis, age and smoking history improved the results slightly, with mainly AMD diagnosis used by the model (MAE = 0.54, MSE = 0.44, RMSE = 0.66, R2 = 0.42). Grad-CAM heatmap evaluation showed that the model decisions rely on retinal morphology factors relevant to AMD diagnosis.

CONCLUSION: The developed method allows an efficient PRS estimation from OCT images. A new technique for analysing the association of OCT images with PRS of AMD, using a deep learning approach, may provide an opportunity to discover new associations between genotype-based AMD risk and retinal morphology.

PMID:38761033 | DOI:10.1111/aos.16710

BMC Oral Health. 2024 May 17;24(1):574. doi: 10.1186/s12903-024-04235-4.

ABSTRACT

BACKGROUND: To develop and validate a deep learning model for automated assessment of endodontic case difficulty from periapical radiographs.

METHODS: A dataset of 1,386 periapical radiographs was compiled from two clinical sites. Two dentists and two endodontists annotated the radiographs for difficulty using the "simple assessment" criteria from the American Association of Endodontists' case difficulty assessment form in the Endocase application. A classification task labeled cases as "easy" or "hard", while regression predicted overall difficulty scores. Convolutional neural networks (i.e. VGG16, ResNet18, ResNet50, ResNext50, and Inception v2) were used, with a baseline model trained via transfer learning from ImageNet weights. Other models was pre-trained using self-supervised contrastive learning (i.e. BYOL, SimCLR, MoCo, and DINO) on 20,295 unlabeled dental radiographs to learn representation without manual labels. Both models were evaluated using 10-fold cross-validation, with performance compared to seven human examiners (three general dentists and four endodontists) on a hold-out test set.

RESULTS: The baseline VGG16 model attained 87.62% accuracy in classifying difficulty. Self-supervised pretraining did not improve performance. Regression predicted scores with ± 3.21 score error. All models outperformed human raters, with poor inter-examiner reliability.

CONCLUSION: This pilot study demonstrated the feasibility of automated endodontic difficulty assessment via deep learning models.

PMID:38760686 | DOI:10.1186/s12903-024-04235-4

Nat Methods. 2024 May 17. doi: 10.1038/s41592-024-02295-6. Online ahead of print.

NO ABSTRACT

PMID:38760611 | DOI:10.1038/s41592-024-02295-6

Nat Med. 2024 May 17. doi: 10.1038/s41591-024-02995-8. Online ahead of print.

ABSTRACT

Precision in the diagnosis of diverse central nervous system (CNS) tumor types is crucial for optimal treatment. DNA methylation profiles, which capture the methylation status of thousands of individual CpG sites, are state-of-the-art data-driven means to enhance diagnostic accuracy but are also time consuming and not widely available. Here, to address these limitations, we developed Deep lEarning from histoPathoLOgy and methYlation (DEPLOY), a deep learning model that classifies CNS tumors to ten major categories from histopathology. DEPLOY integrates three distinct components: the first classifies CNS tumors directly from slide images ('direct model'), the second initially generates predictions for DNA methylation beta values, which are subsequently used for tumor classification ('indirect model'), and the third classifies tumor types directly from routinely available patient demographics. First, we find that DEPLOY accurately predicts beta values from histopathology images. Second, using a ten-class model trained on an internal dataset of 1,796 patients, we predict the tumor categories in three independent external test datasets including 2,156 patients, achieving an overall accuracy of 95% and balanced accuracy of 91% on samples that are predicted with high confidence. These results showcase the potential future use of DEPLOY to assist pathologists in diagnosing CNS tumors within a clinically relevant short time frame.

PMID:38760587 | DOI:10.1038/s41591-024-02995-8

J Assoc Res Otolaryngol. 2024 May 17. doi: 10.1007/s10162-024-00948-z. Online ahead of print.

ABSTRACT

INTRODUCTION: The stria vascularis (SV) may have a significant role in various otologic pathologies. Currently, researchers manually segment and analyze the stria vascularis to measure structural atrophy. Our group developed a tool, SVPath, that uses deep learning to extract and analyze the stria vascularis and its associated capillary bed from whole temporal bone histopathology slides (TBS).

METHODS: This study used an internal dataset of 203 digitized hematoxylin and eosin-stained sections from a normal macaque ear and a separate external validation set of 10 sections from another normal macaque ear. SVPath employed deep learning methods YOLOv8 and nnUnet to detect and segment the SV features from TBS, respectively. The results from this process were analyzed with the SV Analysis Tool (SVAT) to measure SV capillaries and features related to SV morphology, including width, area, and cell count. Once the model was developed, both YOLOv8 and nnUnet were validated on external and internal datasets.

RESULTS: YOLOv8 implementation achieved over 90% accuracy for cochlea and SV detection. nnUnet SV segmentation achieved a DICE score of 0.84-0.95; the capillary bed DICE score was 0.75-0.88. SVAT was applied to compare both the ears used in the study. There was no statistical difference in SV width, SV area, and average area of capillary between the two ears. There was a statistical difference between the two ears for the cell count per SV.

CONCLUSION: The proposed method accurately and efficiently analyzes the SV from temporal histopathology bone slides, creating a platform for researchers to understand the function of the SV further.

PMID:38760547 | DOI:10.1007/s10162-024-00948-z

Commun Med (Lond). 2024 May 17;4(1):90. doi: 10.1038/s43856-024-00518-7.

ABSTRACT

BACKGROUND: Though deep learning has consistently demonstrated advantages in the automatic interpretation of breast ultrasound images, its black-box nature hinders potential interactions with radiologists, posing obstacles for clinical deployment.

METHODS: We proposed a domain knowledge-based interpretable deep learning system for improving breast cancer risk prediction via paired multimodal ultrasound images. The deep learning system was developed on 4320 multimodal breast ultrasound images of 1440 biopsy-confirmed lesions from 1348 prospectively enrolled patients across two hospitals between August 2019 and December 2022. The lesions were allocated to 70% training cohort, 10% validation cohort, and 20% test cohort based on case recruitment date.

RESULTS: Here, we show that the interpretable deep learning system can predict breast cancer risk as accurately as experienced radiologists, with an area under the receiver operating characteristic curve of 0.902 (95% confidence interval = 0.882 - 0.921), sensitivity of 75.2%, and specificity of 91.8% on the test cohort. With the aid of the deep learning system, particularly its inherent explainable features, junior radiologists tend to achieve better clinical outcomes, while senior radiologists experience increased confidence levels. Multimodal ultrasound images augmented with domain knowledge-based reasoning cues enable an effective human-machine collaboration at a high level of prediction performance.

CONCLUSIONS: Such a clinically applicable deep learning system may be incorporated into future breast cancer screening and support assisted or second-read workflows.

PMID:38760506 | DOI:10.1038/s43856-024-00518-7

Nat Commun. 2024 May 17;15(1):4199. doi: 10.1038/s41467-024-48434-7.

ABSTRACT

Understanding how biodiversity has changed through time is a central goal of evolutionary biology. However, estimates of past biodiversity are challenged by the inherent incompleteness of the fossil record, even when state-of-the-art statistical methods are applied to adjust estimates while correcting for sampling biases. Here we develop an approach based on stochastic simulations of biodiversity and a deep learning model to infer richness at global or regional scales through time while incorporating spatial, temporal and taxonomic sampling variation. Our method outperforms alternative approaches across simulated datasets, especially at large spatial scales, providing robust palaeodiversity estimates under a wide range of preservation scenarios. We apply our method on two empirical datasets of different taxonomic and temporal scope: the Permian-Triassic record of marine animals and the Cenozoic evolution of proboscideans. Our estimates provide a revised quantitative assessment of two mass extinctions in the marine record and reveal rapid diversification of proboscideans following their expansion out of Africa and a >70% diversity drop in the Pleistocene.

PMID:38760390 | DOI:10.1038/s41467-024-48434-7

ACS Nano. 2024 May 17. doi: 10.1021/acsnano.4c02919. Online ahead of print.

ABSTRACT

Flexible sensing systems (FSSs) designed to measure plantar pressure can deliver instantaneous feedback on human movement and posture. This feedback is crucial not only for preventing and controlling diseases associated with abnormal plantar pressures but also for optimizing athletes' postures to minimize injuries. The development of an optimal plantar pressure sensor hinges on key metrics such as a wide sensing range, high sensitivity, and long-term stability. However, the effectiveness of current flexible sensors is impeded by numerous challenges, including limitations in structural deformability, mechanical incompatibility between multifunctional layers, and instability under complex stress conditions. Addressing these limitations, we have engineered an integrated pressure sensing system with high sensitivity and reliability for human plantar pressure and gait analysis. It features a high-modulus, porous laminated ionic fiber structure with robust self-bonded interfaces, utilizing a unified polyimide material system. This system showcases a high sensitivity (156.6 kPa-1), an extensive sensing range (up to 4000 kPa), and augmented interfacial toughness and durability (over 150,000 cycles). Additionally, our FSS is capable of real-time monitoring of plantar pressure distribution across various sports activities. Leveraging deep learning, the flexible sensing system achieves a high-precision, intelligent recognition of different plantar types with a 99.8% accuracy rate. This approach provides a strategic advancement in the field of flexible pressure sensors, ensuring prolonged stability and accuracy even amidst complex pressure dynamics and providing a feasible solution for long-term gait monitoring and analysis.

PMID:38760182 | DOI:10.1021/acsnano.4c02919

Drug Discov Today. 2024 May 15:104024. doi: 10.1016/j.drudis.2024.104024. Online ahead of print.

ABSTRACT

3D structure-based drug design (SBDD) is considered a challenging and rational way for innovative drug discovery. Geometric deep learning is a promising approach that solves the accurate model training of 3D SBDD through building neural network models to learn non-Euclidean data, such as 3D molecular graphs and manifold data. Here, we summarize geometric deep learning methods and applications that contain 3D molecular representations, equivariant graph neural networks (EGNNs), and six generative model methods [diffusion model, flow-based model, generative adversarial networks (GANs), variational autoencoder (VAE), autoregressive models, and energy-based models]. Our review provides insights into geometric deep learning methods and advanced applications of 3D SBDD that will be of relevance for the drug discovery community.

PMID:38759948 | DOI:10.1016/j.drudis.2024.104024

Int J Biol Macromol. 2024 May 15:132433. doi: 10.1016/j.ijbiomac.2024.132433. Online ahead of print.

ABSTRACT

Nanopore direct RNA sequencing provided a promising solution for unraveling the landscapes of modifications on single RNA molecules. Here, we proposed NanoMUD, a computational framework for predicting the RNA pseudouridine modification (Ψ) and its methylated analog N1-methylpseudouridine (m1Ψ), which have critical application in mRNA vaccination, at single-base and single-molecule resolution from direct RNA sequencing data. Electric signal features were fed into a bidirectional LSTM neural network to achieve improved accuracy and predictive capabilities. Motif-specific models (NNUNN, N = A, C, U or G) were trained based on features extracted from designed dataset and achieved superior performance on molecule-level modification prediction (Ψ models: min AUC = 0.86, max AUC = 0.99; m1Ψ models: min AUC = 0.87, max AUC = 0.99). We then aggregated read-level predictions for site stoichiometry estimation. Given the observed sequence-dependent bias in model performance, we trained regression models based on the distribution of modification probabilities for sites with known stoichiometry. The distribution-based site stoichiometry estimation method allows unbiased comparison between different contexts. To demonstrate the feasibility of our work, three case studies on both in vitro and in vivo transcribed RNAs were presented. NanoMUD will make a powerful tool to facilitate the research on modified therapeutic IVT RNAs and provides useful insight to the landscape and stoichiometry of pseudouridine and N1-pseudouridine on in vivo transcribed RNA species.

PMID:38759861 | DOI:10.1016/j.ijbiomac.2024.132433

World Neurosurg. 2024 May 15:S1878-8750(24)00828-3. doi: 10.1016/j.wneu.2024.05.065. Online ahead of print.

ABSTRACT

BACKGROUND: Traumatic brain injury (TBI) has become a major source of disability worldwide, increasing the interest in algorithms that use Artificial Intelligence (AI) to optimize the interpretation of imaging studies, prognosis estimation, and critical care issues. In this study we present a bibliometric analysis and Mini Review on the main uses that have been developed for TBI in AI.

METHODS: The results informing this review come from a Scopus database search as of April 15, 2023. The bibliometric analysis was carried out via the mapping bibliographic metrics method. Knowledge mapping was made in the VOSviewer software (V1.6.18), analyzing the "link strength" of networks based on co-occurrence of keywords, countries co-authorship and co-cited authors. In the mini-review section, we highlight the main findings and contributions of the studies.

RESULTS: A total of 495 scientific publications were identified from 2000 to 2023, with 9262 citations published since 2013. Among the 160 journals identified, The Journal of Neurotrauma, Frontiers in Neurology, and Plos One where those with the greatest number of publications. The most frequently co-occurring keywords were: "machine learning", "deep learning", "magnetic resonance imaging", and "intracranial pressure". The United States accounted for more collaborations than any other country, followed by United Kingdom and China. Four co-citation author clusters were found, and the top 20 papers were divided into reviews and original articles.

CONCLUSION: AI has become a relevant research field in TBI during the last 20 years, demonstrating great potential in imaging, but a more modest performance for prognostic estimation and neuromonitoring.

PMID:38759786 | DOI:10.1016/j.wneu.2024.05.065

Int J Surg. 2024 May 17. doi: 10.1097/JS9.0000000000001593. Online ahead of print.

ABSTRACT

BACKGROUND: Clinical differentiation between pulmonary metastases and noncalcified pulmonary hamartomas (NCPH) often presents challenges, leading to potential misdiagnosis. However, the efficacy of a comprehensive model that integrates clinical features, radiomics, and deep learning (CRDL) for differential diagnosis of these two diseases remains uncertain.

OBJECTIVE: This study evaluated the diagnostic efficacy of a Clinical Features, Radiomics, and Deep Learning (CRDL) model in differentiating pulmonary metastases from noncalcified pulmonary hamartomas (NCPH).

METHODS: We retrospectively analyzed the clinical and imaging data of 256 patients from Hospital A and 85 patients from Hospital B, who were pathologically confirmed pulmonary hamartomas or pulmonary metastases after thoracic surgery. Employing Python 3.7 software suites, we extracted radiomic features and deep learning attributes from patient datasets. The cohort was divided into training set, internal validation set, and external validation set. The diagnostic performance of the constructed models was evaluated using receiver operating characteristic (ROC) curve analysis to determine their effectiveness in differentiating between pulmonary metastases and NCPH.

RESULTS: Clinical features such as white blood cell count (WBC), platelet count (PLT), history of cancer, carcinoembryonic antigen (CEA) level, tumor marker status, lesion margin characteristics (smooth or blurred) and maximum diameter were found to have diagnostic value in differentiating between the two diseases. In the domains of radiomics and deep learning. Of the 1,130 radiomics features and 512 deep learning features, 24 and 7, respectively, were selected for model development. The area under the ROC curve (AUC) values for the four groups were 0.980, 0.979, 0.999, and 0.985 in the training set, 0.947, 0.816, 0.934, and 0.952 in the internal validation set, and 0.890, 0.904, 0.923, and 0.938 in the external validation set. This demonstrated that the CRDL model showed the greatest efficacy.

CONCLUSIONS: The comprehensive model incorporating clinical features, radiomics, and deep learning shows promise for aiding in the differentiation between pulmonary metastases and hamartomas.

PMID:38759692 | DOI:10.1097/JS9.0000000000001593

Phys Med Biol. 2024 May 17. doi: 10.1088/1361-6560/ad4d4f. Online ahead of print.

ABSTRACT

OBJECTIVE: Deep learning algorithms have demonstrated impressive performance by leveraging large labeled data. However, acquiring pixel-level annotations for medical image analysis, especially in segmentation tasks, is both costly and time-consuming, posing challenges for supervised learning techniques. Existing semi-supervised methods tend to underutilize representations of unlabeled data and handle labeled and unlabeled data separately, neglecting their interdependencies.

APPROACH: To address this issue, we introduce the Data-Augmented Attention-Decoupled Contrastive model (DADC). This model incorporates an attention decoupling module and utilizes contrastive learning to effectively distinguish foreground and background, significantly improving segmentation accuracy. Our approach integrates an augmentation technique that merges information from both labeled and unlabeled data, notably boosting network performance, especially in scenarios with limited labeled data.

MAIN RESULTS: We conducted comprehensive experiments on the ABUS dataset and the results demonstrate that DADC outperforms existing segmentation methods in terms of segmentation performance.

PMID:38759677 | DOI:10.1088/1361-6560/ad4d4f

Comput Med Imaging Graph. 2024 May 12;115:102384. doi: 10.1016/j.compmedimag.2024.102384. Online ahead of print.

ABSTRACT

BACKGROUND: The KRAS, NRAS, and BRAF genotypes are critical for selecting targeted therapies for patients with metastatic colorectal cancer (mCRC). Here, we aimed to develop a deep learning model that utilizes pathologic whole-slide images (WSIs) to accurately predict the status of KRAS, NRAS, and BRAFV600E.

METHODS: 129 patients with left-sided colon cancer and rectal cancer from the Third Affiliated Hospital of Sun Yat-sen University were assigned to the training and testing cohorts. Utilizing three convolutional neural networks (ResNet18, ResNet50, and Inception v3), we extracted 206 pathological features from H&E-stained WSIs, serving as the foundation for constructing specific pathological models. A clinical feature model was then developed, with carcinoembryonic antigen (CEA) identified through comprehensive multiple regression analysis as the key biomarker. Subsequently, these two models were combined to create a clinical-pathological integrated model, resulting in a total of three genetic prediction models.

RESULT: 103 patients were evaluated in the training cohort (1782,302 image tiles), while the remaining 26 patients were enrolled in the testing cohort (489,481 image tiles). Compared with the clinical model and the pathology model, the combined model which incorporated CEA levels and pathological signatures, showed increased predictive ability, with an area under the curve (AUC) of 0.96 in the training and an AUC of 0.83 in the testing cohort, accompanied by a high positive predictive value (PPV 0.92).

CONCLUSION: The combined model demonstrated a considerable ability to accurately predict the status of KRAS, NRAS, and BRAFV600E in patients with left-sided colorectal cancer, with potential application to assist doctors in developing targeted treatment strategies for mCRC patients, and effectively identifying mutations and eliminating the need for confirmatory genetic testing.

PMID:38759471 | DOI:10.1016/j.compmedimag.2024.102384

Talanta. 2024 May 6;276:126217. doi: 10.1016/j.talanta.2024.126217. Online ahead of print.

ABSTRACT

In this manuscript, a 3D-printed analytical device has been successfully developed to classify illicit drugs using smartphone-based colorimetry. Representative compounds of different families, including cocaine, 3,4-methylenedioxy-methamphetamine (MDMA), amphetamine and cathinone derivatives, pyrrolidine cathinones, and 3,4-methylenedioxy cathinones, have been analyzed and classified after appropriate reaction with Marquis, gallic acid, sulfuric acid, Simon and Scott reagents. A picture of the colored products was acquired using a smartphone, and the corrected RGB values were used as input data in the chemometric treatment. ANN using two active layers of nodes (6 nodes in layer 1 and 2 nodes in layer 2) with a sigmoidal transfer function and a minimum strict threshold of 0.50 identified illicit drug samples with a sensitivity higher than 83.4 % and a specificity of 100 % with limits of detection in the microgram range. The 3D printed device can operate connected to a rechargeable lithium-ion cell portable battery, is inexpensive, and requires minimal training. The analytical device has been able to discriminate the analyzed psychoactive substances from cutting and mixing agents, being a useful tool for law enforcement agents to use as a screening method.

PMID:38759361 | DOI:10.1016/j.talanta.2024.126217

Med Image Anal. 2024 May 13;95:103198. doi: 10.1016/j.media.2024.103198. Online ahead of print.

ABSTRACT

To mitigate systematic errors in magnetic resonance fingerprinting (MRF), the precomputed dictionary is usually computed with minimal granularity across the entire range of tissue parameters. However, the dictionary grows exponentially with the number of parameters increase, posing significant challenges to the computational efficiency and matching accuracy of pattern-matching algorithms. Existing works, primarily based on convolutional neural networks (CNN), focus solely on local information to reconstruct multiple parameter maps, lacking in-depth investigations on the MRF mechanism. These methods may not exploit long-distance redundancies and the contextual information within voxel fingerprints introduced by the Bloch equation dynamics, leading to limited reconstruction speed and accuracy. To overcome these limitations, we propose a novel end-to-end neural network called the Local and Global Vision Transformer (LG-ViT) for MRF parameter reconstruction. Our proposed LG-ViT employs a multi-stage architecture that effectively reduces the computational overhead associated with the high-dimensional MRF data and the transformer model. Specifically, a local Transformer encoder is proposed to capture contextual information embedded within voxel fingerprints and local correlations introduced by the interconnected human tissues. Additionally, a global Transformer encoder is proposed to leverage long-distance dependencies arising from shared characteristics among different tissues across various spatial regions. By incorporating MRF physics-based data priors and effectively capturing local and global correlations, our proposed LG-ViT can achieve fast and accurate MRF parameter reconstruction. Experiments on both simulation and in vivo data demonstrate that the proposed method enables faster and more accurate MRF parameter reconstruction compared to state-of-the-art deep learning-based methods.

PMID:38759259 | DOI:10.1016/j.media.2024.103198

BMC Med Imaging. 2024 May 17;24(1):113. doi: 10.1186/s12880-024-01297-2.

ABSTRACT

BACKGROUND: Recent Convolutional Neural Networks (CNNs) perform low-error reconstruction in fast Magnetic Resonance Imaging (MRI). Most of them convolve the image with kernels and successfully explore the local information. Nonetheless, the non-local image information, which is embedded among image patches relatively far from each other, may be lost due to the limitation of the receptive field of the convolution kernel. We aim to incorporate a graph to represent non-local information and improve the reconstructed images by using the Graph Convolutional Enhanced Self-Similarity (GCESS) network.

METHODS: First, the image is reconstructed into the graph to extract the non-local self-similarity in the image. Second, GCESS uses spatial convolution and graph convolution to process the information in the image, so that local and non-local information can be effectively utilized. The network strengthens the non-local similarity between similar image patches while reconstructing images, making the reconstruction of structure more reliable.

RESULTS: Experimental results on in vivo knee and brain data demonstrate that the proposed method achieves better artifact suppression and detail preservation than state-of-the-art methods, both visually and quantitatively. Under 1D Cartesian sampling with 4 × acceleration (AF = 4), the PSNR of knee data reached 34.19 dB, 1.05 dB higher than that of the compared methods; the SSIM achieved 0.8994, 2% higher than the compared methods. Similar results were obtained for the reconstructed images under other sampling templates as demonstrated in our experiment.

CONCLUSIONS: The proposed method successfully constructs a hybrid graph convolution and spatial convolution network to reconstruct images. This method, through its training process, amplifies the non-local self-similarities, significantly benefiting the structural integrity of the reconstructed images. Experiments demonstrate that the proposed method outperforms the state-of-the-art reconstruction method in suppressing artifacts, as well as in preserving image details.

PMID:38760778 | DOI:10.1186/s12880-024-01297-2

BMC Med Res Methodol. 2024 May 17;24(1):114. doi: 10.1186/s12874-024-02231-4.

ABSTRACT

BACKGROUND: Smoking is a critical risk factor responsible for over eight million annual deaths worldwide. It is essential to obtain information on smoking habits to advance research and implement preventive measures such as screening of high-risk individuals. In most countries, including Denmark, smoking habits are not systematically recorded and at best documented within unstructured free-text segments of electronic health records (EHRs). This would require researchers and clinicians to manually navigate through extensive amounts of unstructured data, which is one of the main reasons that smoking habits are rarely integrated into larger studies. Our aim is to develop machine learning models to classify patients' smoking status from their EHRs.

METHODS: This study proposes an efficient natural language processing (NLP) pipeline capable of classifying patients' smoking status and providing explanations for the decisions. The proposed NLP pipeline comprises four distinct components, which are; (1) considering preprocessing techniques to address abbreviations, punctuation, and other textual irregularities, (2) four cutting-edge feature extraction techniques, i.e. Embedding, BERT, Word2Vec, and Count Vectorizer, employed to extract the optimal features, (3) utilization of a Stacking-based Ensemble (SE) model and a Convolutional Long Short-Term Memory Neural Network (CNN-LSTM) for the identification of smoking status, and (4) application of a local interpretable model-agnostic explanation to explain the decisions rendered by the detection models. The EHRs of 23,132 patients with suspected lung cancer were collected from the Region of Southern Denmark during the period 1/1/2009-31/12/2018. A medical professional annotated the data into 'Smoker' and 'Non-Smoker' with further classifications as 'Active-Smoker', 'Former-Smoker', and 'Never-Smoker'. Subsequently, the annotated dataset was used for the development of binary and multiclass classification models. An extensive comparison was conducted of the detection performance across various model architectures.

RESULTS: The results of experimental validation confirm the consistency among the models. However, for binary classification, BERT method with CNN-LSTM architecture outperformed other models by achieving precision, recall, and F1-scores between 97% and 99% for both Never-Smokers and Active-Smokers. In multiclass classification, the Embedding technique with CNN-LSTM architecture yielded the most favorable results in class-specific evaluations, with equal performance measures of 97% for Never-Smoker and measures in the range of 86 to 89% for Active-Smoker and 91-92% for Never-Smoker.

CONCLUSION: Our proposed NLP pipeline achieved a high level of classification performance. In addition, we presented the explanation of the decision made by the best performing detection model. Future work will expand the model's capabilities to analyze longer notes and a broader range of categories to maximize its utility in further research and screening applications.

PMID:38760718 | DOI:10.1186/s12874-024-02231-4