IEEE J Biomed Health Inform. 2024 Jun 27;PP. doi: 10.1109/JBHI.2024.3417014. Online ahead of print.

ABSTRACT

Deep learning approaches have demonstrated remarkable potential in predicting cancer drug responses (CDRs), using cell line and drug features. However, existing methods predominantly rely on single-omics data of cell lines, potentially overlooking the complex biological mechanisms governing cell line responses. This paper introduces DeepFusionCDR, a novel approach employing unsupervised contrastive learning to amalgamate multi-omics features, including mutation, transcriptome, methylome, and copy number variation data, from cell lines. Furthermore, we incorporate molecular SMILES-specific transformers to derive drug features from their chemical structures. The unified multi-omics and drug signatures are combined, and a multi-layer perceptron (MLP) is applied to predict IC50 values for cell line-drug pairs. Moreover, this MLP can discern whether a cell line is resistant or sensitive to a particular drug. We assessed DeepFusionCDR's performance on the GDSC dataset and juxtaposed it against cutting-edge methods, demonstrating its superior performance in regression and classification tasks. We also conducted ablation studies and case analyses to exhibit the effectiveness and versatility of our proposed approach. Our results underscore the potential of DeepFusionCDR to enhance CDR predictions by harnessing the power of multi-omics fusion and molecular-specific transformers. The prediction of DeepFusionCDR on TCGA patient data and case study highlight the practical application scenarios of DeepFusionCDR in real-world environments. Source code and datasets can be available on https://github.com/altriavin/DeepFusionCDR.

PMID:38935469 | DOI:10.1109/JBHI.2024.3417014

Acta Crystallogr D Struct Biol. 2024 Jul 1. doi: 10.1107/S2059798324005217. Online ahead of print.

ABSTRACT

When solving a structure of a protein from single-wavelength anomalous diffraction X-ray data, the initial phases obtained by phasing from an anomalously scattering substructure usually need to be improved by an iterated electron-density modification. In this manuscript, the use of convolutional neural networks (CNNs) for segmentation of the initial experimental phasing electron-density maps is proposed. The results reported demonstrate that a CNN with U-net architecture, trained on several thousands of electron-density maps generated mainly using X-ray data from the Protein Data Bank in a supervised learning, can improve current density-modification methods.

PMID:38935341 | DOI:10.1107/S2059798324005217

Med Phys. 2024 Jun 27. doi: 10.1002/mp.17276. Online ahead of print.

ABSTRACT

BACKGROUND: Gastrointestinal stromal tumors (GISTs) are clinically heterogeneous with various malignant potential in different individuals. It is crucial to explore a reliable method for preoperative risk stratification of gastric GISTs noninvasively.

PURPOSE: To establish and evaluate a machine learning model using the combination of computed tomography (CT) morphology, radiomics, and deep learning features to predict the risk stratification of primary gastric GISTs preoperatively.

METHODS: The 193 gastric GISTs lesions were randomly divided into training set, validation set, and test set in a ratio of 6:2:2. The qualitative and quantitative CT morphological features were assessed by two radiologists. The tumors were segmented manually, and then radiomic features were extracted using PyRadiomics and the deep learning features were extracted using pre-trained Resnet50 from arterial phase and venous phase CT images, respectively. Pearson correlation analysis and recursive feature elimination were used for feature selection. Support vector machines were employed to build a classifier for predicting the risk stratification of GISTs. This study compared the performance of models using different pre-trained convolutional neural networks (CNNs) to extract deep features for classification, as well as the performance of modeling features from single-phase and dual-phase images. The arterial phase, venous phase and dual-phase machine learning models were built, respectively, and the morphological features were added to the dual-phase machine learning model to construct a combined model. Receiver operating characteristic (ROC) curves were used to evaluate the efficacy of each model. The clinical application value of the combined model was determined through the decision curve analysis (DCA) and the net reclassification index (NRI) was analyzed.

RESULTS: The area under the curve (AUC) of the dual-phase machine learning model was 0.876, which was higher than that of the arterial phase model or venous phase model (0.813, 0.838, respectively). The combined model had best predictive performance than the above models with an AUC of 0.941 (95% CI: 0.887-0.974) (p = 0.012, Delong test). DCA demonstrated that the combined model had good clinical application value with an NRI of 0.575 (95% CI: 0.357-0.891).

CONCLUSION: In this study, we established a combined model that incorporated dual-phase morphology, radiomics, and deep learning characteristics, which can be used to predict the preoperative risk stratification of gastric GISTs.

PMID:38935330 | DOI:10.1002/mp.17276

Insights Imaging. 2024 Jun 27;15(1):164. doi: 10.1186/s13244-024-01724-6.

ABSTRACT

OBJECTIVES: To develop and validate a deep learning (DL) model for automated segmentation of hepatic and portal veins, and apply the model in blood-free future liver remnant (FLR) assessments via CT before major hepatectomy.

METHODS: 3-dimensional 3D U-Net models were developed for the automatic segmentation of hepatic veins and portal veins on contrast-enhanced CT images. A total of 170 patients treated from January 2018 to March 2019 were included. 3D U-Net models were trained and tested under various liver conditions. The Dice similarity coefficient (DSC) and volumetric similarity (VS) were used to evaluate the segmentation accuracy. The use of quantitative volumetry for evaluating resection was compared between blood-filled and blood-free settings and between manual and automated segmentation.

RESULTS: The DSC values in the test dataset for hepatic veins and portal veins were 0.66 ± 0.08 (95% CI: (0.65, 0.68)) and 0.67 ± 0.07 (95% CI: (0.66, 0.69)), the VS values were 0.80 ± 0.10 (95% CI: (0.79, 0.84)) and 0.74 ± 0.08 (95% CI: (0.73, 0.76)), respectively No significant differences in FLR, FLR% assessments, or the percentage of major hepatectomy patients were noted between the blood-filled and blood-free settings (p = 0.67, 0.59 and 0.99 for manual methods, p = 0.66, 0.99 and 0.99 for automated methods, respectively) according to the use of manual and automated segmentation methods.

CONCLUSION: Fully automated segmentation of hepatic veins and portal veins and FLR assessment via blood-free CT before major hepatectomy are accurate and applicable in clinical cases involving the use of DL.

CRITICAL RELEVANCE STATEMENT: Our fully automatic models could segment hepatic veins, portal veins, and future liver remnant in blood-free setting on CT images before major hepatectomy with reliable outcomes.

KEY POINTS: Fully automatic segmentation of hepatic veins and portal veins was feasible in clinical practice. Fully automatic volumetry of future liver remnant (FLR)% in a blood-free setting was robust. No significant differences in FLR% assessments were noted between the blood-filled and blood-free settings.

PMID:38935177 | DOI:10.1186/s13244-024-01724-6

Optom Vis Sci. 2024 Jun 28. doi: 10.1097/OPX.0000000000002158. Online ahead of print.

ABSTRACT

SIGNIFICANCE: Our retinal image-based deep learning (DL) cardiac biological age (BioAge) model could facilitate fast, accurate, noninvasive screening for cardiovascular disease (CVD) in novel community settings and thus improve outcome with those with limited access to health care services.

PURPOSE: This study aimed to determine whether the results issued by our DL cardiac BioAge model are consistent with the known trends of CVD risk and the biomarker leukocyte telomere length (LTL), in a cohort of individuals from the UK Biobank.

METHODS: A cross-sectional cohort study was conducted using those individuals in the UK Biobank who had LTL data. These individuals were divided by sex, ranked by LTL, and then grouped into deciles. The retinal images were then presented to the DL model, and individual's cardiac BioAge was determined. Individuals within each LTL decile were then ranked by cardiac BioAge, and the mean of the CVD risk biomarkers in the top and bottom quartiles was compared. The relationship between an individual's cardiac BioAge, the CVD biomarkers, and LTL was determined using traditional correlation statistics.

RESULTS: The DL cardiac BioAge model was able to accurately stratify individuals by the traditional CVD risk biomarkers, and for both males and females, those issued with a cardiac BioAge in the top quartile of their chronological peer group had a significantly higher mean systolic blood pressure, hemoglobin A1c, and 10-year Pooled Cohort Equation CVD risk scores compared with those individuals in the bottom quartile (p<0.001). Cardiac BioAge was associated with LTL shortening for both males and females (males: -0.22, r2 = 0.04; females: -0.18, r2 = 0.03).

CONCLUSIONS: In this cross-sectional cohort study, increasing CVD risk whether assessed by traditional biomarkers, CVD risk scoring, or our DL cardiac BioAge, CVD risk model, was inversely related to LTL. At a population level, our data support the growing body of evidence that suggests LTL shortening is a surrogate marker for increasing CVD risk and that this risk can be captured by our novel DL cardiac BioAge model.

PMID:38935034 | DOI:10.1097/OPX.0000000000002158

Mol Biol Evol. 2024 Jun 27:msae136. doi: 10.1093/molbev/msae136. Online ahead of print.

ABSTRACT

Most algorithms that are used to predict the effects of variants rely on evolutionary conservation. However, a majority of such techniques compute evolutionary conservation by solely using the alignment of multiple sequences while overlooking the evolutionary context of substitution events. We had introduced PHACT, a scoring-based pathogenicity predictor for missense mutations that can leverage phylogenetic trees, in our previous study. By building on this foundation, we now propose PHACTboost, a gradient boosting tree-based classifier that combines PHACT scores with information from multiple sequence alignments, phylogenetic trees, and ancestral reconstruction. By learning from data PHACTboost outperforms PHACT. Furthermore, the results of comprehensive experiments on carefully constructed sets of variants demonstrated that PHACTboost can outperform 40 prevalent pathogenicity predictors reported in the dbNSFP, including conventional tools, meta-predictors, and deep learning-based approaches as well as more recent tools such as, AlphaMissense, EVE, and CPT-1. The superiority of PHACTboost over these methods was particularly evident in case of hard variants for which different pathogenicity predictors offered conflicting results. We provide predictions of 215 million amino acid alterations over 20,191 proteins. PHACTboost is available at https://github.com/CompGenomeLab/PHACTboost. PHACTboost can improve our understanding of genetic diseases and facilitate more accurate diagnoses.

PMID:38934805 | DOI:10.1093/molbev/msae136

ACS Sens. 2024 Jun 27. doi: 10.1021/acssensors.4c00149. Online ahead of print.

ABSTRACT

Raman spectroscopy has become an important single-cell analysis tool for monitoring biochemical changes at the cellular level. However, Raman spectral data, typically presented as continuous data with high-dimensional characteristics, is distinct from discrete sequences, which limits the application of deep learning-based algorithms in data analysis due to the lack of discretization. Herein, a model called fragment-fusion transformer is proposed, which integrates the discrete fragmentation of continuous spectra based on their intrinsic characteristics with the extraction of intrafragment features and the fusion of interfragment features. The model integrates the intrinsic feature-based fragmentation of spectra with transformer, constructing the fragment transformer block for feature extraction within fragments. Interfragment information is combined through the pyramid design structure to improve the model's receptive field and fully exploit the spectral properties. During the pyramidal fusion process, the information gain of the final extracted features in the spectrum has been enhanced by a factor of 9.24 compared to the feature extraction stage within the fragment, and the information entropy has been enhanced by a factor of 13. The fragment-fusion transformer achieved a spectral recognition accuracy of 94.5%, which is 4% higher compared to the method without fragmentation and fusion processes on the test set of cell Raman spectroscopy identification experiments. In comparison to common spectral classification models such as KNN, SVM, logistic regression, and CNN, fragment-fusion transformer has achieved 4.4% higher accuracy than the best-performing CNN model. Fragment-fusion transformer method has the potential to serve as a general framework for discretization in the field of continuous spectral data analysis and as a research tool for analyzing the intrinsic information within spectra.

PMID:38934798 | DOI:10.1021/acssensors.4c00149

Laryngoscope. 2024 Jun 27. doi: 10.1002/lary.31609. Online ahead of print.

ABSTRACT

OBJECTIVES: To develop and validate machine learning (ML) and deep learning (DL) models using drug-induced sleep endoscopy (DISE) images to predict the therapeutic efficacy of hypoglossal nerve stimulator (HGNS) implantation.

METHODS: Patients who underwent DISE and subsequent HGNS implantation at a tertiary care referral center were included. Six DL models and five ML algorithms were trained on images from the base of tongue (BOT) and velopharynx (VP) from patients classified as responders or non-responders as defined by Sher's criteria (50% reduction in apnea-hypopnea index (AHI) and AHI < 15 events/h). Precision, recall, F1 score, and overall accuracy were evaluated as measures of performance.

RESULTS: In total, 25,040 images from 127 patients were included, of which 16,515 (69.3%) were from responders and 8,262 (30.7%) from non-responders. Models trained on the VP dataset had greater overall accuracy when compared to BOT alone and combined VP and BOT image sets, suggesting that VP images contain discriminative features for identifying therapeutic efficacy. The VCG-16 DL model had the best overall performance on the VP image set with high training accuracy (0.833), F1 score (0.78), and recall (0.883). Among ML models, the logistic regression model had the greatest accuracy (0.685) and F1 score (0.813).

CONCLUSION: Deep neural networks have potential to predict HGNS therapeutic efficacy using images from DISE, facilitating better patient selection for implantation. Development of multi-institutional data and image sets will allow for development of generalizable predictive models.

LEVEL OF EVIDENCE: N/A Laryngoscope, 2024.

PMID:38934474 | DOI:10.1002/lary.31609

Network. 2024 Jun 27:1-22. doi: 10.1080/0954898X.2024.2369137. Online ahead of print.

ABSTRACT

Cloud services are one of the most quickly developing technologies. Furthermore, load balancing is recognized as a fundamental challenge for achieving energy efficiency. The primary function of load balancing is to deliver optimal services by releasing the load over multiple resources. Fault tolerance is being used to improve the reliability and accessibility of the network. In this paper, a hybrid Deep Learning-based load balancing algorithm is developed. Initially, tasks are allocated to all VMs in a round-robin method. Furthermore, the Deep Embedding Cluster (DEC) utilizes the Central Processing Unit (CPU), bandwidth, memory, processing elements, and frequency scaling factors while determining if a VM is overloaded or underloaded. The task performed on the overloaded VM is valued and the tasks accomplished on the overloaded VM are assigned to the underloaded VM for cloud load balancing. In addition, the Deep Q Recurrent Neural Network (DQRNN) is proposed to balance the load based on numerous factors such as supply, demand, capacity, load, resource utilization, and fault tolerance. Furthermore, the effectiveness of this model is assessed by load, capacity, resource consumption, and success rate, with ideal values of 0.147, 0.726, 0.527, and 0.895 are achieved.

PMID:38934441 | DOI:10.1080/0954898X.2024.2369137

Magn Reson Med. 2024 Jun 27. doi: 10.1002/mrm.30197. Online ahead of print.

ABSTRACT

PURPOSE: To develop a fast denoising framework for high-dimensional MRI data based on a self-supervised learning scheme, which does not require ground truth clean image.

THEORY AND METHODS: Quantitative MRI faces limitations in SNR, because the variation of signal amplitude in a large set of images is the key mechanism for quantification. In addition, the complex non-linear signal models make the fitting process vulnerable to noise. To address these issues, we propose a fast deep-learning framework for denoising, which efficiently exploits the redundancy in multidimensional MRI data. A self-supervised model was designed to use only noisy images for training, bypassing the challenge of clean data paucity in clinical practice. For validation, we used two different datasets of simulated magnetization transfer contrast MR fingerprinting (MTC-MRF) dataset and in vivo DWI image dataset to show the generalizability.

RESULTS: The proposed method drastically improved denoising performance in the presence of mild-to-severe noise regardless of noise distributions compared to previous methods of the BM3D, tMPPCA, and Patch2self. The improvements were even pronounced in the following quantification results from the denoised images.

CONCLUSION: The proposed MD-S2S (Multidimensional-Self2Self) denoising technique could be further applied to various multi-dimensional MRI data and improve the quantification accuracy of tissue parameter maps.

PMID:38934408 | DOI:10.1002/mrm.30197

J Am Med Inform Assoc. 2024 Jun 27:ocae144. doi: 10.1093/jamia/ocae144. Online ahead of print.

ABSTRACT

OBJECTIVES: The surge in patient portal messages (PPMs) with increasing needs and workloads for efficient PPM triage in healthcare settings has spurred the exploration of AI-driven solutions to streamline the healthcare workflow processes, ensuring timely responses to patients to satisfy their healthcare needs. However, there has been less focus on isolating and understanding patient primary concerns in PPMs-a practice which holds the potential to yield more nuanced insights and enhances the quality of healthcare delivery and patient-centered care.

MATERIALS AND METHODS: We propose a fusion framework to leverage pretrained language models (LMs) with different language advantages via a Convolution Neural Network for precise identification of patient primary concerns via multi-class classification. We examined 3 traditional machine learning models, 9 BERT-based language models, 6 fusion models, and 2 ensemble models.

RESULTS: The outcomes of our experimentation underscore the superior performance achieved by BERT-based models in comparison to traditional machine learning models. Remarkably, our fusion model emerges as the top-performing solution, delivering a notably improved accuracy score of 77.67 ± 2.74% and an F1 score of 74.37 ± 3.70% in macro-average.

DISCUSSION: This study highlights the feasibility and effectiveness of multi-class classification for patient primary concern detection and the proposed fusion framework for enhancing primary concern detection.

CONCLUSIONS: The use of multi-class classification enhanced by a fusion of multiple pretrained LMs not only improves the accuracy and efficiency of patient primary concern identification in PPMs but also aids in managing the rising volume of PPMs in healthcare, ensuring critical patient communications are addressed promptly and accurately.

PMID:38934289 | DOI:10.1093/jamia/ocae144

Front Hum Neurosci. 2024 Jun 12;18:1394107. doi: 10.3389/fnhum.2024.1394107. eCollection 2024.

ABSTRACT

BACKGROUND: Error-related potentials (ErrPs) are electrophysiological responses that naturally occur when humans perceive wrongdoing or encounter unexpected events. It offers a distinctive means of comprehending the error-processing mechanisms within the brain. A method for detecting ErrPs with high accuracy holds significant importance for various ErrPs-based applications, such as human-in-the-loop Brain-Computer Interface (BCI) systems. Nevertheless, current methods fail to fulfill the generalization requirements for detecting such ErrPs due to the high non-stationarity of EEG signals across different tasks and the limited availability of ErrPs datasets.

METHODS: This study introduces a deep learning-based model that integrates convolutional layers and transformer encoders for the classification of ErrPs. Subsequently, a model training strategy, grounded in transfer learning, is proposed for the effective training of the model. The datasets utilized in this study are available for download from the publicly accessible databases.

RESULTS: In cross-task classification, an average accuracy of about 78% was achieved, exceeding the baseline. Furthermore, in the leave-one-subject-out, within-session, and cross-session classification scenarios, the proposed model outperformed the existing techniques with an average accuracy of 71.81, 78.74, and 77.01%, respectively.

CONCLUSIONS: Our approach contributes to mitigating the challenge posed by limited datasets in the ErrPs field, achieving this by reducing the requirement for extensive training data for specific target tasks. This may serve as inspiration for future studies that concentrate on ErrPs and their applications.

PMID:38933146 | PMC:PMC11199896 | DOI:10.3389/fnhum.2024.1394107

Telemed J E Health. 2024 Jun 27. doi: 10.1089/tmj.2023.0703. Online ahead of print.

ABSTRACT

Background: Blurry images in teledermatology and consultation increased the diagnostic difficulty for both deep learning models and physicians. We aim to determine the extent of restoration in diagnostic accuracy after blurry images are deblurred by deep learning models. Methods: We used 19,191 skin images from a public skin image dataset that includes 23 skin disease categories, 54 skin images from a public dataset of blurry skin images, and 53 blurry dermatology consultation photos in a medical center to compare the diagnosis accuracy of trained diagnostic deep learning models and subjective sharpness between blurry and deblurred images. We evaluated five different deblurring models, including models for motion blur, Gaussian blur, Bokeh blur, mixed slight blur, and mixed strong blur. Main Outcomes and Measures: Diagnostic accuracy was measured as sensitivity and precision of correct model prediction of the skin disease category. Sharpness rating was performed by board-certified dermatologists on a 4-point scale, with 4 being the highest image clarity. Results: The sensitivity of diagnostic models dropped 0.15 and 0.22 on slightly and strongly blurred images, respectively, and deblurring models restored 0.14 and 0.17 for each group. The sharpness ratings perceived by dermatologists improved from 1.87 to 2.51 after deblurring. Activation maps showed the focus of diagnostic models was compromised by the blurriness but was restored after deblurring. Conclusions: Deep learning models can restore the diagnostic accuracy of diagnostic models for blurry images and increase image sharpness perceived by dermatologists. The model can be incorporated into teledermatology to help the diagnosis of blurry images.

PMID:38934135 | DOI:10.1089/tmj.2023.0703

Front Endocrinol (Lausanne). 2024 Jun 11;15:1429115. doi: 10.3389/fendo.2024.1429115. eCollection 2024.

ABSTRACT

OBJECTIVES: The growing incidence of differentiated thyroid cancer (DTC) have been linked to insulin resistance and metabolic syndrome. The imperative need for developing effective diagnostic imaging tools to predict the non-iodine-avid status of lung metastasis (LMs) in differentiated thyroid cancer (DTC) patients is underscored to prevent unnecessary radioactive iodine treatment (RAI).

METHODS: Primary cohort consisted 1962 pretreated LMs of 496 consecutive DTC patients with pretreated initially diagnosed LMs who underwent chest CT and subsequent post-treatment radioiodine SPECT. After automatic lesion segmentation by SE V-Net, SE Net deep learning was trained to predict non-iodine-avid status of LMs. External validation cohort contained 123 pretreated LMs of 24 consecutive patients from other two hospitals. Stepwise validation was further performed according to the nodule's largest diameter.

RESULTS: The SE-Net deep learning network yielded area under the receiver operating characteristic curve (AUC) values of 0.879 (95% confidence interval: 0.852-0.906) and 0.713 (95% confidence interval: 0.613-0.813) for internal and external validation. With the LM diameter decreasing from ≥10mm to ≤4mm, the AUCs remained relatively stable, for smallest nodules (≤4mm), the model yielded an AUC of 0.783. Decision curve analysis showed that most patients benefited using deep learning to decide radioactive I131 treatment.

CONCLUSION: This study presents a noninvasive, less radioactive and fully automatic approach that can facilitate suitable DTC patient selection for RAI therapy of LMs. Further prospective multicenter studies with larger study cohorts and related metabolic factors should address the possibility of comprehensive clinical transformation.

PMID:38933823 | PMC:PMC11201526 | DOI:10.3389/fendo.2024.1429115

Front Neurosci. 2024 Jun 12;18:1344114. doi: 10.3389/fnins.2024.1344114. eCollection 2024.

ABSTRACT

One-shot learning, the ability to learn a new concept from a single instance, is a distinctive brain function that has garnered substantial interest in machine learning. While modeling physiological mechanisms poses challenges, advancements in artificial neural networks have led to performances in specific tasks that rival human capabilities. Proposing one-shot learning methods with these advancements, especially those involving simple mechanisms, not only enhance technological development but also contribute to neuroscience by proposing functionally valid hypotheses. Among the simplest methods for one-shot class addition with deep learning image classifiers is "weight imprinting," which uses neural activity from a new class image data as the corresponding new synaptic weights. Despite its simplicity, its relevance to neuroscience is ambiguous, and it often interferes with original image classification, which is a significant drawback in practical applications. This study introduces a novel interpretation where a part of the weight imprinting process aligns with the Hebbian rule. We show that a single Hebbian-like process enables pre-trained deep learning image classifiers to perform one-shot class addition without any modification to the original classifier's backbone. Using non-parametric normalization to mimic brain's fast Hebbian plasticity significantly reduces the interference observed in previous methods. Our method is one of the simplest and most practical for one-shot class addition tasks, and its reliance on a single fast Hebbian-like process contributes valuable insights to neuroscience hypotheses.

PMID:38933813 | PMC:PMC11202076 | DOI:10.3389/fnins.2024.1344114

Plant Phenomics. 2024 Jun 26;6:0188. doi: 10.34133/plantphenomics.0188. eCollection 2024.

ABSTRACT

The tassel state in maize hybridization fields not only reflects the growth stage of the maize but also reflects the performance of the detasseling operation. Existing tassel detection models are primarily used to identify mature tassels with obvious features, making it difficult to accurately identify small tassels or detasseled plants. This study presents a novel approach that utilizes unmanned aerial vehicles (UAVs) and deep learning techniques to accurately identify and assess tassel states, before and after manually detasseling in maize hybridization fields. The proposed method suggests that a specific tassel annotation and data augmentation strategy is valuable for substantial enhancing the quality of the tassel training data. This study also evaluates mainstream object detection models and proposes a series of highly accurate tassel detection models based on tassel categories with strong data adaptability. In addition, a strategy for blocking large UAV images, as well as improving tassel detection accuracy, is proposed to balance UAV image acquisition and computational cost. The experimental results demonstrate that the proposed method can accurately identify and classify tassels at various stages of detasseling. The tassel detection model optimized with the enhanced data achieves an average precision of 94.5% across all categories. An optimal model combination that uses blocking strategies for different development stages can improve the tassel detection accuracy to 98%. This could be useful in addressing the issue of missed tassel detections in maize hybridization fields. The data annotation strategy and image blocking strategy may also have broad applications in object detection and recognition in other agricultural scenarios.

PMID:38933805 | PMC:PMC11200267 | DOI:10.34133/plantphenomics.0188

IMA J Appl Math. 2024 May 23;89(1):143-174. doi: 10.1093/imamat/hxae011. eCollection 2024 Jan.

ABSTRACT

Partial differential equations (PDEs) play a fundamental role in the mathematical modelling of many processes and systems in physical, biological and other sciences. To simulate such processes and systems, the solutions of PDEs often need to be approximated numerically. The finite element method, for instance, is a usual standard methodology to do so. The recent success of deep neural networks at various approximation tasks has motivated their use in the numerical solution of PDEs. These so-called physics-informed neural networks and their variants have shown to be able to successfully approximate a large range of PDEs. So far, physics-informed neural networks and the finite element method have mainly been studied in isolation of each other. In this work, we compare the methodologies in a systematic computational study. Indeed, we employ both methods to numerically solve various linear and nonlinear PDEs: Poisson in 1D, 2D and 3D, Allen-Cahn in 1D, semilinear Schrödinger in 1D and 2D. We then compare computational costs and approximation accuracies. In terms of solution time and accuracy, physics-informed neural networks have not been able to outperform the finite element method in our study. In some experiments, they were faster at evaluating the solved PDE.

PMID:38933736 | PMC:PMC11197852 | DOI:10.1093/imamat/hxae011

Front Oncol. 2024 Jun 12;14:1379624. doi: 10.3389/fonc.2024.1379624. eCollection 2024.

ABSTRACT

OBJECTIVES: Precise segmentation of Odontogenic Cystic Lesions (OCLs) from dental Cone-Beam Computed Tomography (CBCT) is critical for effective dental diagnosis. Although supervised learning methods have shown practical diagnostic results in segmenting various diseases, their ability to segment OCLs covering different sub-class varieties has not been extensively investigated.

METHODS: In this study, we propose a new supervised learning method termed OCL-Net that combines a Multi-Scaled U-Net model, along with an Auto-Adapting mechanism trained with a combined supervised loss. Anonymous CBCT images were collected retrospectively from one hospital. To assess the ability of our model to improve the diagnostic efficiency of maxillofacial surgeons, we conducted a diagnostic assessment where 7 clinicians were included to perform the diagnostic process with and without the assistance of auto-segmentation masks.

RESULTS: We collected 300 anonymous CBCT images which were manually annotated for segmentation masks. Extensive experiments demonstrate the effectiveness of our OCL-Net for CBCT OCLs segmentation, achieving an overall Dice score of 88.84%, an IoU score of 81.23%, and an AUC score of 92.37%. Through our diagnostic assessment, we found that when clinicians were assisted with segmentation labels from OCL-Net, their average diagnostic accuracy increased from 53.21% to 55.71%, while the average time spent significantly decreased from 101s to 47s (P<0.05).

CONCLUSION: The findings demonstrate the potential of our approach as a robust auto-segmentation system on OCLs in CBCT images, while the segmented masks can be used to further improve OCLs dental diagnostic efficiency.

PMID:38933446 | PMC:PMC11199543 | DOI:10.3389/fonc.2024.1379624

Inverse Probl. 2024 Aug 1;40(8):085002. doi: 10.1088/1361-6420/ad4f0a. Epub 2024 Jun 25.

ABSTRACT

Supervised deep learning-based methods have inspired a new wave of image reconstruction methods that implicitly learn effective regularization strategies from a set of training data. While they hold potential for improving image quality, they have also raised concerns regarding their robustness. Instabilities can manifest when learned methods are applied to find approximate solutions to ill-posed image reconstruction problems for which a unique and stable inverse mapping does not exist, which is a typical use case. In this study, we investigate the performance of supervised deep learning-based image reconstruction in an alternate use case in which a stable inverse mapping is known to exist but is not yet analytically available in closed form. For such problems, a deep learning-based method can learn a stable approximation of the unknown inverse mapping that generalizes well to data that differ significantly from the training set. The learned approximation of the inverse mapping eliminates the need to employ an implicit (optimization-based) reconstruction method and can potentially yield insights into the unknown analytic inverse formula. The specific problem addressed is image reconstruction from a particular case of radially truncated circular Radon transform (CRT) data, referred to as 'half-time' measurement data. For the half-time image reconstruction problem, we develop and investigate a learned filtered backprojection method that employs a convolutional neural network to approximate the unknown filtering operation. We demonstrate that this method behaves stably and readily generalizes to data that differ significantly from training data. The developed method may find application to wave-based imaging modalities that include photoacoustic computed tomography.

PMID:38933410 | PMC:PMC11197394 | DOI:10.1088/1361-6420/ad4f0a

Front Comput Neurosci. 2024 Jun 12;18:1414462. doi: 10.3389/fncom.2024.1414462. eCollection 2024.

ABSTRACT

Parkinson's disease (PD) is a globally significant health challenge, necessitating accurate and timely diagnostic methods to facilitate effective treatment and intervention. In recent years, self-supervised deep representation pattern learning (SS-DRPL) has emerged as a promising approach for extracting valuable representations from data, offering the potential to enhance the efficiency of voice-based PD detection. This research study focuses on investigating the utilization of SS-DRPL in conjunction with deep learning algorithms for voice-based PD classification. This study encompasses a comprehensive evaluation aimed at assessing the accuracy of various predictive models, particularly deep learning methods when combined with SS-DRPL. Two deep learning architectures, namely hybrid Long Short-Term Memory and Recurrent Neural Networks (LSTM-RNN) and Deep Neural Networks (DNN), are employed and compared in terms of their ability to detect voice-based PD cases accurately. Additionally, several traditional machine learning models are also included to establish a baseline for comparison. The findings of the study reveal that the incorporation of SS-DRPL leads to improved model performance across all experimental setups. Notably, the LSTM-RNN architecture augmented with SS-DRPL achieves the highest F1-score of 0.94, indicating its superior ability to detect PD cases using voice-based data effectively. This outcome underscores the efficacy of SS-DRPL in enabling deep learning models to learn intricate patterns and correlations within the data, thereby facilitating more accurate PD classification.

PMID:38933392 | PMC:PMC11199684 | DOI:10.3389/fncom.2024.1414462