Comput Biol Med. 2024 Feb 23;171:108192. doi: 10.1016/j.compbiomed.2024.108192. Online ahead of print.

ABSTRACT

Doppler echocardiography is a widely utilised non-invasive imaging modality for assessing the functionality of heart valves, including the mitral valve. Manual assessments of Doppler traces by clinicians introduce variability, prompting the need for automated solutions. This study introduces an innovative deep learning model for automated detection of peak velocity measurements from mitral inflow Doppler images, independent from Electrocardiogram information. A dataset of Doppler images annotated by multiple expert cardiologists was established, serving as a robust benchmark. The model leverages heatmap regression networks, achieving 96% detection accuracy. The model discrepancy with the expert consensus falls comfortably within the range of inter- and intra-observer variability in measuring Doppler peak velocities. The dataset and models are open-source, fostering further research and clinical application.

PMID:38417384 | DOI:10.1016/j.compbiomed.2024.108192

J Environ Manage. 2024 Feb 27;354:120497. doi: 10.1016/j.jenvman.2024.120497. Online ahead of print.

ABSTRACT

By identifying Earth heritage sites, UNESCO Global Geoparks (UGGps) have promoted geo-tourism and regional economic prosperity. However, commercial and tourism development has altered the natural contexts of these geoparks, diminishing their initial value. Before implementing land use policies, spatial landscape parameters should be monitored in multiple dimensions and in real time. This study aims to develop Bilateral Segmentation Network (BiSeNet) models employing an upgraded U-structured neural network in order to monitor land use/cover changes and landscape indicators in a Vietnamese UGGp. This network has proven effective at preserving input image data and restricting the loss of spatial information in decoding data. To demonstrate the utility of deep learning, eight trained BiSeNet models were evaluated against Maximum Likelihood, Support Vector Machine, and Random Forest. The trained BSN-Nadam model (128x128), with a precision of 94% and an information loss of 0.1, can become a valuable instrument for analyzing and monitoring monthly changes in land uses/covers once tourism activities have been rapidly expanded. Three tourist routes and 41 locations in the Dak Nong UGGp were monitored for 30 years using three landscape indices: Disjunct Core Area Density (DCAD), Total Edge Contrast Index (TECI), Shannon's Diversity Index (SHDI), based on the results of the model. As a result, 18 identified geo-sites in the Daknong Geopark have been influenced significantly by agricultural and tourist activities since 2010, making these sites less uniform and unsustainable management. It promptly alerts UNESCO management to the deterioration of geological sites caused by urbanization and tourist development.

PMID:38417365 | DOI:10.1016/j.jenvman.2024.120497

J Am Vet Med Assoc. 2024 Feb 27:1-8. doi: 10.2460/javma.23.07.0372. Online ahead of print.

ABSTRACT

OBJECTIVE: To compare pedigree documentation and genetic test results to evaluate whether user-provided photographs influence the breed ancestry predictions of direct-to-consumer (DTC) genetic tests for dogs.

ANIMALS: 12 registered purebred pet dogs representing 12 different breeds.

METHODS: Each dog owner submitted 6 buccal swabs, 1 to each of 6 DTC genetic testing companies. Experimenters registered each sample per manufacturer instructions. For half of the dogs, the registration included a photograph of the DNA donor. For the other half of the dogs, photographs were swapped between dogs. DNA analysis and breed ancestry prediction were conducted by each company. The effect of condition (ie, matching vs shuffled photograph) was evaluated for each company's breed predictions. As a positive control, a convolutional neural network was also used to predict breed based solely on the photograph.

RESULTS: Results from 5 of the 6 tests always included the dog's registered breed. One test and the convolutional neural network were unlikely to identify the registered breed and frequently returned results that were more similar to the photograph than the DNA. Additionally, differences in the predictions made across all tests underscored the challenge of identifying breed ancestry, even in purebred dogs.

CLINICAL RELEVANCE: Veterinarians are likely to encounter patients who have conducted DTC genetic testing and may be asked to explain the results of genetic tests they did not order. This systematic comparison of commercially available tests provides context for interpreting results from consumer-grade DTC genetic testing kits.

PMID:38417257 | DOI:10.2460/javma.23.07.0372

BMC Med Imaging. 2024 Feb 28;24(1):51. doi: 10.1186/s12880-024-01227-2.

ABSTRACT

Pulmonary diseases are various pathological conditions that affect respiratory tissues and organs, making the exchange of gas challenging for animals inhaling and exhaling. It varies from gentle and self-limiting such as the common cold and catarrh, to life-threatening ones, such as viral pneumonia (VP), bacterial pneumonia (BP), and tuberculosis, as well as a severe acute respiratory syndrome, such as the coronavirus 2019 (COVID-19). The cost of diagnosis and treatment of pulmonary infections is on the high side, most especially in developing countries, and since radiography images (X-ray and computed tomography (CT) scan images) have proven beneficial in detecting various pulmonary infections, many machine learning (ML) models and image processing procedures have been utilized to identify these infections. The need for timely and accurate detection can be lifesaving, especially during a pandemic. This paper, therefore, suggested a deep convolutional neural network (DCNN) founded image detection model, optimized with image augmentation technique, to detect three (3) different pulmonary diseases (COVID-19, bacterial pneumonia, and viral pneumonia). The dataset containing four (4) different classes (healthy (10,325), COVID-19 (3,749), BP (883), and VP (1,478)) was utilized as training/testing data for the suggested model. The model's performance indicates high potential in detecting the three (3) classes of pulmonary diseases. The model recorded average detection accuracy of 94%, 95.4%, 99.4%, and 98.30%, and training/detection time of about 60/50 s. This result indicates the proficiency of the suggested approach when likened to the traditional texture descriptors technique of pulmonary disease recognition utilizing X-ray and CT scan images. This study introduces an innovative deep convolutional neural network model to enhance the detection of pulmonary diseases like COVID-19 and pneumonia using radiography. This model, notable for its accuracy and efficiency, promises significant advancements in medical diagnostics, particularly beneficial in developing countries due to its potential to surpass traditional diagnostic methods.

PMID:38418987 | DOI:10.1186/s12880-024-01227-2

BMC Bioinformatics. 2024 Feb 28;25(1):87. doi: 10.1186/s12859-024-05708-7.

ABSTRACT

Several experimental evidences have shown that the human endogenous hormones can interact with drugs in many ways and affect drug efficacy. The hormone drug interactions (HDI) are essential for drug treatment and precision medicine; therefore, it is essential to understand the hormone-drug associations. Here, we present HormoNet to predict the HDI pairs and their risk level by integrating features derived from hormone and drug target proteins. To the best of our knowledge, this is one of the first attempts to employ deep learning approach for prediction of HDI prediction. Amino acid composition and pseudo amino acid composition were applied to represent target information using 30 physicochemical and conformational properties of the proteins. To handle the imbalance problem in the data, we applied synthetic minority over-sampling technique technique. Additionally, we constructed novel datasets for HDI prediction and the risk level of their interaction. HormoNet achieved high performance on our constructed hormone-drug benchmark datasets. The results provide insights into the understanding of the relationship between hormone and a drug, and indicate the potential benefit of reducing risk levels of interactions in designing more effective therapies for patients in drug treatments. Our benchmark datasets and the source codes for HormoNet are available in: https://github.com/EmamiNeda/HormoNet .

PMID:38418979 | DOI:10.1186/s12859-024-05708-7

Respir Res. 2024 Feb 28;25(1):103. doi: 10.1186/s12931-024-02712-6.

ABSTRACT

BACKGROUND: The prognostic role of changes in body fat in patients with idiopathic pulmonary fibrosis (IPF) remains underexplored. We investigated the association between changes in body fat during the first year post-diagnosis and outcomes in patients with IPF.

METHODS: This single-center, retrospective study included IPF patients with chest CT scan and pulmonary function test (PFT) at diagnosis and a one-year follow-up between January 2010 and December 2020. The fat area (cm2, sum of subcutaneous and visceral fat) and muscle area (cm2) at the T12-L1 level were obtained from chest CT images using a fully automatic deep learning-based software. Changes in the body composition were dichotomized using thresholds dividing the lowest quartile and others, respectively (fat area: -52.3 cm2, muscle area: -7.4 cm2). Multivariable Cox regression analyses adjusted for PFT result and IPF extent on CT images and the log-rank test were performed to assess the association between the fat area change during the first year post-diagnosis and the composite outcome of death or lung transplantation.

RESULTS: In total, 307 IPF patients (69.3 ± 8.1 years; 238 men) were included. During the first year post-diagnosis, fat area, muscle area, and body mass index (BMI) changed by -15.4 cm2, -1 cm2, and - 0.4 kg/m2, respectively. During a median follow-up of 47 months, 146 patients had the composite outcome (47.6%). In Cox regression analyses, a change in the fat area < -52.3 cm2 was associated with composite outcome incidence in models adjusted with baseline clinical variables (hazard ratio [HR], 1.566, P = .022; HR, 1.503, P = .036 in a model including gender, age, and physiology [GAP] index). This prognostic value was consistent when adjusted with one-year changes in clinical variables (HR, 1.495; P = .030). However, the change in BMI during the first year was not a significant prognostic factor (P = .941). Patients with a change in fat area exceeding this threshold experienced the composite outcome more frequently than their counterparts (58.4% vs. 43.9%; P = .007).

CONCLUSION: A ≥ 52.3 cm2 decrease in fat area, automatically measured using deep learning technique, at T12-L1 in one year post-diagnosis was an independent poor prognostic factor in IPF patients.

PMID:38418966 | DOI:10.1186/s12931-024-02712-6

BMC Bioinformatics. 2024 Feb 29;25(1):88. doi: 10.1186/s12859-024-05716-7.

ABSTRACT

BACKGROUND: Predicting outcome of breast cancer is important for selecting appropriate treatments and prolonging the survival periods of patients. Recently, different deep learning-based methods have been carefully designed for cancer outcome prediction. However, the application of these methods is still challenged by interpretability. In this study, we proposed a novel multitask deep neural network called UISNet to predict the outcome of breast cancer. The UISNet is able to interpret the importance of features for the prediction model via an uncertainty-based integrated gradients algorithm. UISNet improved the prediction by introducing prior biological pathway knowledge and utilizing patient heterogeneity information.

RESULTS: The model was tested in seven public datasets of breast cancer, and showed better performance (average C-index = 0.691) than the state-of-the-art methods (average C-index = 0.650, ranged from 0.619 to 0.677). Importantly, the UISNet identified 20 genes as associated with breast cancer, among which 11 have been proven to be associated with breast cancer by previous studies, and others are novel findings of this study.

CONCLUSIONS: Our proposed method is accurate and robust in predicting breast cancer outcomes, and it is an effective way to identify breast cancer-associated genes. The method codes are available at: https://github.com/chh171/UISNet .

PMID:38418940 | DOI:10.1186/s12859-024-05716-7

Sci Rep. 2024 Feb 28;14(1):4852. doi: 10.1038/s41598-024-54536-5.

ABSTRACT

Assessing infant carrying and holding (C/H), or physical infant-caregiver interaction, is important for a wide range of contexts in development research. An automated detection and quantification of infant C/H is particularly needed in long term at-home studies where development of infants' neurobehavior is measured using wearable devices. Here, we first developed a phenomenological categorization for physical infant-caregiver interactions to support five different definitions of C/H behaviors. Then, we trained and assessed deep learning-based classifiers for their automatic detection from multi-sensor wearable recordings that were originally used for mobile assessment of infants' motor development. Our results show that an automated C/H detection is feasible at few-second temporal accuracy. With the best C/H definition, the automated detector shows 96% accuracy and 0.56 kappa, which is slightly less than the video-based inter-rater agreement between trained human experts (98% accuracy, 0.77 kappa). The classifier performance varies with C/H definition reflecting the extent to which infants' movements are present in each C/H variant. A systematic benchmarking experiment shows that the widely used actigraphy-based method ignores the normally occurring C/H behaviors. Finally, we show proof-of-concept for the utility of the novel classifier in studying C/H behavior across infant development. Particularly, we show that matching the C/H detections to individuals' gross motor ability discloses novel insights to infant-parent interaction.

PMID:38418850 | DOI:10.1038/s41598-024-54536-5

Commun Med (Lond). 2024 Feb 28;4(1):31. doi: 10.1038/s43856-024-00464-4.

ABSTRACT

BACKGROUND: Long-term monitoring of Electrocardiogram (ECG) recordings is crucial to diagnose arrhythmias. Clinicians can find it challenging to diagnose arrhythmias, and this is a particular issue in more remote and underdeveloped areas. The development of digital ECG and AI methods could assist clinicians who need to diagnose arrhythmias outside of the hospital setting.

METHODS: We constructed a large-scale Chinese ECG benchmark dataset using data from 272,753 patients collected from January 2017 to December 2021. The dataset contains ECG recordings from all common arrhythmias present in the Chinese population. Several experienced cardiologists from Shanghai First People's Hospital labeled the dataset. We then developed a deep learning-based multi-label interpretable diagnostic model from the ECG recordings. We utilized Accuracy, F1 score and AUC-ROC to compare the performance of our model with that of the cardiologists, as well as with six comparison models, using testing and hidden data sets.

RESULTS: The results show that our approach achieves an F1 score of 83.51%, an average AUC ROC score of 0.977, and 93.74% mean accuracy for 6 common arrhythmias. Results from the hidden dataset demonstrate the performance of our approach exceeds that of cardiologists. Our approach also highlights the diagnostic process.

CONCLUSIONS: Our diagnosis system has superior diagnostic performance over that of clinicians. It also has the potential to help clinicians rapidly identify abnormal regions on ECG recordings, thus improving efficiency and accuracy of clinical ECG diagnosis in China. This approach could therefore potentially improve the productivity of out-of-hospital ECG diagnosis and provides a promising prospect for telemedicine.

PMID:38418628 | DOI:10.1038/s43856-024-00464-4

Nat Commun. 2024 Feb 28;15(1):1834. doi: 10.1038/s41467-024-45323-x.

ABSTRACT

Predicting the evolution of systems with spatio-temporal dynamics in response to external stimuli is essential for scientific progress. Traditional equations-based approaches leverage first principles through the numerical approximation of differential equations, thus demanding extensive computational resources. In contrast, data-driven approaches leverage deep learning algorithms to describe system evolution in low-dimensional spaces. We introduce an architecture, termed Latent Dynamics Network, capable of uncovering low-dimensional intrinsic dynamics in potentially non-Markovian systems. Latent Dynamics Networks automatically discover a low-dimensional manifold while learning the system dynamics, eliminating the need for training an auto-encoder and avoiding operations in the high-dimensional space. They predict the evolution, even in time-extrapolation scenarios, of space-dependent fields without relying on predetermined grids, thus enabling weight-sharing across query-points. Lightweight and easy-to-train, Latent Dynamics Networks demonstrate superior accuracy (normalized error 5 times smaller) in highly-nonlinear problems with significantly fewer trainable parameters (more than 10 times fewer) compared to state-of-the-art methods.

PMID:38418469 | DOI:10.1038/s41467-024-45323-x

Am J Orthod Dentofacial Orthop. 2024 Mar;165(3):369-371. doi: 10.1016/j.ajodo.2023.12.003.

NO ABSTRACT

PMID:38418035 | DOI:10.1016/j.ajodo.2023.12.003

Med Eng Phys. 2024 Feb;124:104101. doi: 10.1016/j.medengphy.2024.104101. Epub 2024 Jan 9.

ABSTRACT

With the advancement of deep learning technology, computer-aided diagnosis (CAD) is playing an increasing role in the field of medical diagnosis. In particular, the emergence of Transformer-based models has led to a wider application of computer vision technology in the field of medical image processing. In the diagnosis of thyroid diseases, the diagnosis of benign and malignant thyroid nodules based on the TI-RADS classification is greatly influenced by the subjective judgment of ultrasonographers, and at the same time, it also brings an extremely heavy workload to ultrasonographers. To address this, we propose Swin-Residual Transformer (SRT) in this paper, which incorporates residual blocks and triplet loss into Swin Transformer (SwinT). It improves the sensitivity to global and localized features of thyroid nodules and better distinguishes small feature differences. In our exploratory experiments, SRT model achieves an accuracy of 0.8832 with an AUC of 0.8660, outperforming state-of-the-art convolutional neural network (CNN) and Transformer models. Also, ablation experiments have demonstrated the improved performance in the thyroid nodule classification task after introducing residual blocks and triple loss. These results validate the potential of the proposed SRT model to improve the diagnosis of thyroid nodules' ultrasound images. It also provides a feasible guarantee to avoid excessive puncture sampling of thyroid nodules in future clinical diagnosis.

PMID:38418029 | DOI:10.1016/j.medengphy.2024.104101

Med Eng Phys. 2024 Feb;124:104104. doi: 10.1016/j.medengphy.2024.104104. Epub 2024 Jan 15.

ABSTRACT

In recent years, research has highlighted the association between increased adipose tissue surrounding the human heart and elevated susceptibility to cardiovascular diseases such as atrial fibrillation and coronary heart disease. However, the manual segmentation of these fat deposits has not been widely implemented in clinical practice due to the substantial workload it entails for medical professionals and the associated costs. Consequently, the demand for more precise and time-efficient quantitative analysis has driven the emergence of novel computational methods for fat segmentation. This study presents a novel deep learning-based methodology that offers autonomous segmentation and quantification of two distinct types of cardiac fat deposits. The proposed approach leverages the pix2pix network, a generative conditional adversarial network primarily designed for image-to-image translation tasks. By applying this network architecture, we aim to investigate its efficacy in tackling the specific challenge of cardiac fat segmentation, despite not being originally tailored for this purpose. The two types of fat deposits of interest in this study are referred to as epicardial and mediastinal fats, which are spatially separated by the pericardium. The experimental results demonstrated an average accuracy of 99.08% and f1-score 98.73 for the segmentation of the epicardial fat and 97.90% of accuracy and f1-score of 98.40 for the mediastinal fat. These findings represent the high precision and overlap agreement achieved by the proposed methodology. In comparison to existing studies, our approach exhibited superior performance in terms of f1-score and run time, enabling the images to be segmented in real time.

PMID:38418017 | DOI:10.1016/j.medengphy.2024.104104

Phys Med Biol. 2024 Feb 28. doi: 10.1088/1361-6560/ad2e6f. Online ahead of print.

ABSTRACT

Positron Emission Tomography (PET) is essential for non-invasive imaging of metabolic processes in healthcare applications. However, the use of radiolabeled tracers exposes patients to ionizing radiation, raising concerns about carcinogenic potential, and warranting efforts to minimize doses without sacrificing diagnostic quality. In this work, we present a novel neural network architecture, PETformer, designed for denoising ultra-low-dose PET images without requiring structural priors such as CT or MRI. The architecture utilizes a U-net backbone, synergistically combining multi-headed transposed attention (MDTA) blocks with kernel-basis attention (KBA) and channel attention (CA) mechanisms for both short- and long-range dependencies and enhanced feature extraction. PETformer is trained and validated on a dataset of 317 patients imaged on a total-body uEXPLORER PET/CT scanner. Quantitative evaluations using structural similarity index measure (SSIM) and liver SNR showed PETformer's significant superiority over other established denoising algorithms across different dose-reduction factors. Its ability to identify and recover intrinsic anatomical details from background noise with dose reductions as low as 2% and its capacity in maintaining high target-to-background ratios while preserving the integrity of uptake values of small lesions enables PET-only fast and accurate disease diagnosis. Furthermore, PETformer exhibits computational efficiency with only 37M trainable parameters, making it well-suited for commercial integration.&#xD.

PMID:38417180 | DOI:10.1088/1361-6560/ad2e6f

Phys Med Biol. 2024 Feb 28. doi: 10.1088/1361-6560/ad2e6b. Online ahead of print.

ABSTRACT

&#xD;Objective: Honeycomb lung is a rare but severe disease characterized by honeycomb-like imaging features and distinct radiological characteristics. Therefore, this study aims to develop a deep-learning model capable of segmenting honeycomb lung lesions from CT scans to address the efficacy issue of honeycomb lung segmentation.&#xD;Methods: This study proposes a Sparse Mapping-based Graph Representation Segmentation Network (SM-GRSNet). SM-GRSNet integrates an attention affinity mechanism to effectively filter redundant features at a coarse-grained region level. The attention encoder generated by this mechanism specifically focuses on the lesion area. Additionally, we introduce a graph representation module based on sparse links in SM-GRSNet. Subsequently, graph representation operations are performed on the sparse graph, yielding detailed lesion segmentation results. Finally, we construct a pyramid-structured cascaded decoder in SM-GRSNet, which combines features from the sparse link-based graph representation modules and attention encoders to generate the final segmentation mask.&#xD;Results: Experimental results demonstrate that the proposed SM-GRSNet achieves state-of-the-art performance on a dataset comprising 7170 honeycomb lung CT images. Our model attains the highest IOU (87.62%), Dice(93.41%). Furthermore, our model also achieves the lowest HD95 (6.95) and ASD (2.47). &#xD;Significance: The SM-GRSNet method proposed in this paper can be used for automatic segmentation of honeycomb lung CT images, which enhances the segmentation performance of Honeycomb lung lesions under small sample datasets. It will help doctors with early screening, accurate diagnosis ,and customized treatment. This method maintains a high correlation and consistency between the automatic segmentation results and the expert manual segmentation results. Accurate automatic segmentation of the honeycomb lung lesion area is clinically important.

PMID:38417177 | DOI:10.1088/1361-6560/ad2e6b

Biomed Phys Eng Express. 2024 Feb 28. doi: 10.1088/2057-1976/ad2e36. Online ahead of print.

ABSTRACT

Steady-state visual evoked potential (SSVEP) is a key technique of electroencephalography (EEG)-based brain-computer interfaces (BCI), which has been widely applied to neurological function assessment and postoperative rehabilitation. However, accurate decoding of the user's intended based on the SSVEP-EEG signals is challenging due to the low signal-to-noise ratio and large individual variability of the signals. To address these issues, we proposed a parallel multi-band fusion convolutional neural network (PMF-CNN). Multi frequency band signals were served as the input of PMF-CNN to fully utilize the time-frequency information of EEG. Three parallel modules, spatial self-attention (SAM), temporal self-attention (TAM), and squeeze-excitation (SEM), were proposed to automatically extract multi-dimensional features from spatial, temporal, and frequency domains, respectively. A novel spatial-temporal-frequency representation were designed to capture the correlation of electrode channels, time intervals, and different sub-harmonics by using SAM, TAM, and SEM, respectively. The three parallel modules operate independently and simultaneously. A four layers CNN classification module was designed to fuse parallel multi-dimensional features and achieve the accurate classification of SSVEP-EEG signals. The PMF-CNN was further interpreted by using brain functional connectivity analysis. The proposed method was validated using two large publicly available datasets. After trained using our proposed dual-stage training pattern, the classification accuracies were 99.37% and 93.96%, respectively, which are superior to the current state-of-the-art SSVEP-EEG classification algorithms. The algorithm exhibits high classification accuracy and good robustness, which has the potential to be applied to postoperative rehabilitation.

PMID:38417170 | DOI:10.1088/2057-1976/ad2e36

J Breast Imaging. 2022 Oct 10;4(5):451-459. doi: 10.1093/jbi/wbac052.

ABSTRACT

Breast cancer screening has evolved substantially over the past few decades because of advancements in new image acquisition systems and novel artificial intelligence (AI) algorithms. This review provides a brief overview of the history, current state, and future of AI in breast cancer screening and diagnosis along with challenges involved in the development of AI systems. Although AI has been developing for interpretation tasks associated with breast cancer screening for decades, its potential to combat the subjective nature and improve the efficiency of human image interpretation is always expanding. The rapid advancement of computational power and deep learning has increased greatly in AI research, with promising performance in detection and classification tasks across imaging modalities. Most AI systems, based on human-engineered or deep learning methods, serve as concurrent or secondary readers, that is, as aids to radiologists for a specific, well-defined task. In the future, AI may be able to perform multiple integrated tasks, making decisions at the level of or surpassing the ability of humans. Artificial intelligence may also serve as a partial primary reader to streamline ancillary tasks, triaging cases or ruling out obvious normal cases. However, before AI is used as an independent, autonomous reader, various challenges need to be addressed, including explainability and interpretability, in addition to repeatability and generalizability, to ensure that AI will provide a significant clinical benefit to breast cancer screening across all populations.

PMID:38416954 | DOI:10.1093/jbi/wbac052

J Breast Imaging. 2022 Oct 10;4(5):488-495. doi: 10.1093/jbi/wbac046.

ABSTRACT

OBJECTIVE: Artificial intelligence (AI)-based triage algorithms may improve cancer detection and expedite radiologist workflow. To this end, the performance of a commercial AI-based triage algorithm on screening mammograms was evaluated across breast densities and lesion types.

METHODS: This retrospective, IRB-exempt, multicenter, multivendor study examined 1255 screening 4-view mammograms (400 positive and 855 negative studies). Images were anonymized by providing institutions and analyzed by a commercially available AI algorithm (cmTriage, CureMetrix, La Jolla, CA) that performed retrospective triage at the study level by flagging exams as "suspicious" or not. Sensitivities and specificities with confidence intervals were derived from area under the curve (AUC) calculations.

RESULTS: The algorithm demonstrated an AUC of 0.95 (95% CI: 0.94-0.96) for case identification. Area under the curve held across densities (0.95) and lesion types (masses: 0.94 [95% CI: 0.92-0.96] or microcalcifications: 0.97 [95% CI: 0.96-0.99]). The algorithm has a default sensitivity of 93% (95% CI: 95.6%-90.5%) with specificity of 76.3% (95% CI: 79.2%-73.4%). To evaluate real-world performance, a sensitivity of 86.9% (95% CI: 83.6%-90.2%) was tested, as observed for practicing radiologists by the Breast Cancer Surveillance Consortium (BCSC) study. The resulting specificity was 88.5% (95% CI: 86.4%-90.7%), similar to the BCSC specificity of 88.9%, indicating performance comparable to real-world results.

CONCLUSION: When tested for lesion detection, an AI-based triage software can perform at the level of practicing radiologists. Drawing attention to suspicious exams may improve reader specificity and help streamline radiologist workflow, enabling faster turnaround times and improving care.

PMID:38416951 | DOI:10.1093/jbi/wbac046

PLoS One. 2024 Feb 28;19(2):e0298739. doi: 10.1371/journal.pone.0298739. eCollection 2024.

ABSTRACT

With the rapid development of ocean observation technology, underwater object detection has begun to occupy an essential position in the fields of aquaculture, environmental monitoring, marine science, etc. However, due to the problems unique to underwater images such as severe noise, blurred objects, and multi-scale, deep learning-based target detection algorithms lack sufficient capabilities to cope with these challenges. To address these issues, we improve DETR to make it well suited for underwater scenarios. First, a simple and effective learnable query recall mechanism is proposed to mitigate the effect of noise and can significantly improve the detection performance of the object. Second, for underwater small and irregular object detection, a lightweight adapter is designed to provide multi-scale features for the encoding and decoding stages. Third, the regression mechanism of the bounding box is optimized using the combination loss of smooth L1 and CIoU. Finally, we validate the designed network against other state-of-the-art methods on the RUOD dataset. The experimental results show that the proposed method is effective.

PMID:38416764 | DOI:10.1371/journal.pone.0298739

PLoS Comput Biol. 2024 Feb 28;20(2):e1011892. doi: 10.1371/journal.pcbi.1011892. eCollection 2024 Feb.

ABSTRACT

In proteomics, a crucial aspect is to identify peptide sequences. De novo sequencing methods have been widely employed to identify peptide sequences, and numerous tools have been proposed over the past two decades. Recently, deep learning approaches have been introduced for de novo sequencing. Previous methods focused on encoding tandem mass spectra and predicting peptide sequences from the first amino acid onwards. However, when predicting peptides using tandem mass spectra, the peptide sequence can be predicted not only from the first amino acid but also from the last amino acid due to the coexistence of b-ion (or a- or c-ion) and y-ion (or x- or z-ion) fragments in the tandem mass spectra. Therefore, it is essential to predict peptide sequences bidirectionally. Our approach, called NovoB, utilizes a Transformer model to predict peptide sequences bidirectionally, starting with both the first and last amino acids. In comparison to Casanovo, our method achieved an improvement of the average peptide-level accuracy rate of approximately 9.8% across all species.

PMID:38416757 | DOI:10.1371/journal.pcbi.1011892