Int J Med Robot. 2024 Aug;20(4):e2664. doi: 10.1002/rcs.2664.

ABSTRACT

BACKGROUND: This study aimed to develop a novel deep convolutional neural network called Dual-path Double Attention Transformer (DDA-Transformer) designed to achieve precise and fast knee joint CT image segmentation and to validate it in robotic-assisted total knee arthroplasty (TKA).

METHODS: The femoral, tibial, patellar, and fibular segmentation performance and speed were evaluated and the accuracy of component sizing, bone resection and alignment of the robotic-assisted TKA system constructed using this deep learning network was clinically validated.

RESULTS: Overall, DDA-Transformer outperformed six other networks in terms of the Dice coefficient, intersection over union, average surface distance, and Hausdorff distance. DDA-Transformer exhibited significantly faster segmentation speeds than nnUnet, TransUnet and 3D-Unet (p < 0.01). Furthermore, the robotic-assisted TKA system outperforms the manual group in surgical accuracy.

CONCLUSIONS: DDA-Transformer exhibited significantly improved accuracy and robustness in knee joint segmentation, and this convenient and stable knee joint CT image segmentation network significantly improved the accuracy of the TKA procedure.

PMID:38994900 | DOI:10.1002/rcs.2664

World J Gastrointest Oncol. 2024 Jun 15;16(6):2404-2418. doi: 10.4251/wjgo.v16.i6.2404.

ABSTRACT

BACKGROUND: Research on gastrointestinal mucosal adenocarcinoma (GMA) is limited and controversial, and there is no reference tool for predicting postoperative survival.

AIM: To investigate the prognosis of GMA and develop predictive model.

METHODS: From the Surveillance, Epidemiology, and End Results database, we collected clinical information on patients with GMA. After random sampling, the patients were divided into the discovery (70% of the total, for model training), validation (20%, for model evaluation), and completely blind test cohorts (10%, for further model evaluation). The main assessment metric was the area under the receiver operating characteristic curve (AUC). All collected clinical features were used for Cox proportional hazard regression analysis to determine factors influencing GMA's prognosis.

RESULTS: This model had an AUC of 0.7433 [95% confidence intervals (95%CI): 0.7424-0.7442] in the discovery cohort, 0.7244 (GMA: 0.7234-0.7254) in the validation cohort, and 0.7388 (95%CI: 0.7378-0.7398) in the test cohort. We packaged it into Windows software for doctors' use and uploaded it. Mucinous gastric adenocarcinoma had the worst prognosis, and these were protective factors of GMA: Regional nodes examined [hazard ratio (HR): 0.98, 95%CI: 0.97-0.98, P < 0.001)] and chemotherapy (HR: 0.62, 95%CI: 0.58-0.66, P < 0.001).

CONCLUSION: The deep learning-based tool developed can accurately predict the overall survival of patients with GMA postoperatively. Combining surgery, chemotherapy, and adequate lymph node dissection during surgery can improve patient outcomes.

PMID:38994138 | PMC:PMC11236227 | DOI:10.4251/wjgo.v16.i6.2404

Radiol Bras. 2024 Jun 26;57:e20230096en. doi: 10.1590/0100-3984.2023.0096-en. eCollection 2024 Jan-Dec.

ABSTRACT

OBJECTIVE: To develop a natural language processing application capable of automatically identifying benign gallbladder diseases that require surgery, from radiology reports.

MATERIALS AND METHODS: We developed a text classifier to classify reports as describing benign diseases of the gallbladder that do or do not require surgery. We randomly selected 1,200 reports describing the gallbladder from our database, including different modalities. Four radiologists classified the reports as describing benign disease that should or should not be treated surgically. Two deep learning architectures were trained for classification: a convolutional neural network (CNN) and a bidirectional long short-term memory (BiLSTM) network. In order to represent words in vector form, the models included a Word2Vec representation, with dimensions of 300 or 1,000. The models were trained and evaluated by dividing the dataset into training, validation, and subsets (80/10/10).

RESULTS: The CNN and BiLSTM performed well in both dimensional spaces. For the 300- and 1,000-dimensional spaces, respectively, the F1-scores were 0.95945 and 0.95302 for the CNN model, compared with 0.96732 and 0.96732 for the BiLSTM model.

CONCLUSION: Our models achieved high performance, regardless of the architecture and dimensional space employed.

PMID:38993952 | PMC:PMC11235066 | DOI:10.1590/0100-3984.2023.0096-en

Front Plant Sci. 2024 Jun 26;15:1349209. doi: 10.3389/fpls.2024.1349209. eCollection 2024.

ABSTRACT

Counting nematodes is a labor-intensive and time-consuming task, yet it is a pivotal step in various quantitative nematological studies; preparation of initial population densities and final population densities in pot, micro-plot and field trials for different objectives related to management including sampling and location of nematode infestation foci. Nematologists have long battled with the complexities of nematode counting, leading to several research initiatives aimed at automating this process. However, these research endeavors have primarily focused on identifying single-class objects within individual images. To enhance the practicality of this technology, there's a pressing need for an algorithm that cannot only detect but also classify multiple classes of objects concurrently. This study endeavors to tackle this challenge by developing a user-friendly Graphical User Interface (GUI) that comprises multiple deep learning algorithms, allowing simultaneous recognition and categorization of nematode eggs and second stage juveniles of Meloidogyne spp. In total of 650 images for eggs and 1339 images for juveniles were generated using two distinct imaging systems, resulting in 8655 eggs and 4742 Meloidogyne juveniles annotated using bounding box and segmentation, respectively. The deep-learning models were developed by leveraging the Convolutional Neural Networks (CNNs) machine learning architecture known as YOLOv8x. Our results showed that the models correctly identified eggs as eggs and Meloidogyne juveniles as Meloidogyne juveniles in 94% and 93% of instances, respectively. The model demonstrated higher than 0.70 coefficient correlation between model predictions and observations on unseen images. Our study has showcased the potential utility of these models in practical applications for the future. The GUI is made freely available to the public through the author's GitHub repository (https://github.com/bresilla/nematode_counting). While this study currently focuses on one genus, there are plans to expand the GUI's capabilities to include other economically significant genera of plant parasitic nematodes. Achieving these objectives, including enhancing the models' accuracy on different imaging systems, may necessitate collaboration among multiple nematology teams and laboratories, rather than being the work of a single entity. With the increasing interest among nematologists in harnessing machine learning, the authors are confident in the potential development of a universal automated nematode counting system accessible to all. This paper aims to serve as a framework and catalyst for initiating global collaboration toward this important goal.

PMID:38993936 | PMC:PMC11238600 | DOI:10.3389/fpls.2024.1349209

Front Transplant. 2023 Jun 23;2:1206085. doi: 10.3389/frtra.2023.1206085. eCollection 2023.

ABSTRACT

An accurate estimation of liver fat content is necessary to predict how a donated liver will function after transplantation. Currently, a pathologist needs to be available at all hours of the day, even at remote hospitals, when an organ donor is procured. Even among expert pathologists, the estimation of liver fat content is operator-dependent. Here we describe the development of a low-cost, end-to-end artificial intelligence platform to evaluate liver fat content on a donor liver biopsy slide in real-time. The hardware includes a high-resolution camera, display, and GPU to acquire and process donor liver biopsy slides. A deep learning model was trained to label and quantify fat globules in liver tissue. The algorithm was deployed on the device to enable real-time quantification and characterization of fat content for transplant decision-making. This information is displayed on the device and can also be sent to a cloud platform for further analysis.

PMID:38993883 | PMC:PMC11235317 | DOI:10.3389/frtra.2023.1206085

Arthroplast Today. 2024 Jun 18;28:101398. doi: 10.1016/j.artd.2024.101398. eCollection 2024 Aug.

ABSTRACT

BACKGROUND: Hip dysplasia is considered one of the leading etiologies contributing to hip degeneration and the eventual need for total hip arthroplasty (THA). We validated a deep learning (DL) algorithm to measure angles relevant to hip dysplasia and applied this algorithm to determine the prevalence of dysplasia in a large population based on incremental radiographic cutoffs.

METHODS: Patients from the Osteoarthritis Initiative with anteroposterior pelvis radiographs and without previous THAs were included. A DL algorithm automated 3 angles associated with hip dysplasia: modified lateral center-edge angle (LCEA), Tönnis angle, and modified Sharp angle. The algorithm was validated against manual measurements, and all angles were measured in a cohort of 3869 patients (61.2 ± 9.2 years, 57.1% female). The percentile distributions and prevalence of dysplastic hips were analyzed using each angle.

RESULTS: The algorithm had no significant difference (P > .05) in measurements (paired difference: 0.3°-0.7°) against readers and had excellent agreement for dysplasia classification (kappa = 0.78-0.88). In 140 minutes, 23,214 measurements were automated for 3869 patients. LCEA and Sharp angles were higher and the Tönnis angle was lower (P < .01) in females. The dysplastic hip prevalence varied from 2.5% to 20% utilizing the following cutoffs: 17.3°-25.5° (LCEA), 9.4°-15.6° (Tönnis), and 41.3°-45.9° (Sharp).

CONCLUSIONS: A DL algorithm was developed to measure and classify hips with mild hip dysplasia. The reported prevalence of dysplasia in a large patient cohort was dependent on both the measurement and threshold, with 12.4% of patients having dysplasia radiographic indices indicative of higher THA risk.

PMID:38993836 | PMC:PMC11237356 | DOI:10.1016/j.artd.2024.101398

Digit Discov. 2024 Jun 10;3(7):1410-1420. doi: 10.1039/d4dd00116h. eCollection 2024 Jul 10.

ABSTRACT

This work presents a recommendation system for metal-organic frameworks (MOFs) inspired by online content platforms. By leveraging the unsupervised Doc2Vec model trained on document-structured intrinsic MOF characteristics, the model embeds MOFs into a high-dimensional chemical space and suggests a pool of promising materials for specific applications based on user-endorsed MOFs with similarity analysis. This proposed approach significantly reduces the need for exhaustive labeling of every material in the database, focusing instead on a select fraction for in-depth investigation. Ranging from methane storage and carbon capture to quantum properties, this study illustrates the system's adaptability to various applications.

PMID:38993728 | PMC:PMC11235176 | DOI:10.1039/d4dd00116h

J Med Imaging (Bellingham). 2024 Jul;11(4):044501. doi: 10.1117/1.JMI.11.4.044501. Epub 2024 Jul 9.

ABSTRACT

PURPOSE: Medical imaging-based machine learning (ML) for computer-aided diagnosis of in vivo lesions consists of two basic components or modules of (i) feature extraction from non-invasively acquired medical images and (ii) feature classification for prediction of malignancy of lesions detected or localized in the medical images. This study investigates their individual performances for diagnosis of low-dose computed tomography (CT) screening-detected lesions of pulmonary nodules and colorectal polyps.

APPROACH: Three feature extraction methods were investigated. One uses the mathematical descriptor of gray-level co-occurrence image texture measure to extract the Haralick image texture features (HFs). One uses the convolutional neural network (CNN) architecture to extract deep learning (DL) image abstractive features (DFs). The third one uses the interactions between lesion tissues and X-ray energy of CT to extract tissue-energy specific characteristic features (TFs). All the above three categories of extracted features were classified by the random forest (RF) classifier with comparison to the DL-CNN method, which reads the images, extracts the DFs, and classifies the DFs in an end-to-end manner. The ML diagnosis of lesions or prediction of lesion malignancy was measured by the area under the receiver operating characteristic curve (AUC). Three lesion image datasets were used. The lesions' tissue pathological reports were used as the learning labels.

RESULTS: Experiments on the three datasets produced AUC values of 0.724 to 0.878 for the HFs, 0.652 to 0.965 for the DFs, and 0.985 to 0.996 for the TFs, compared to the DL-CNN of 0.694 to 0.964. These experimental outcomes indicate that the RF classifier performed comparably to the DL-CNN classification module and the extraction of tissue-energy specific characteristic features dramatically improved AUC value.

CONCLUSIONS: The feature extraction module is more important than the feature classification module. Extraction of tissue-energy specific characteristic features is more important than extraction of image abstractive and characteristic features.

PMID:38993628 | PMC:PMC11234229 | DOI:10.1117/1.JMI.11.4.044501

Proc Mach Learn Res. 2024;227:1406-1422.

ABSTRACT

Multiplex immunofluorescence (MxIF) is an advanced molecular imaging technique that can simultaneously provide biologists with multiple (i.e., more than 20) molecular markers on a single histological tissue section. Unfortunately, due to imaging restrictions, the more routinely used hematoxylin and eosin (H&E) stain is typically unavailable with MxIF on the same tissue section. As biological H&E staining is not feasible, previous efforts have been made to obtain H&E whole slide image (WSI) from MxIF via deep learning empowered virtual staining. However, the tiling effect is a long-lasting problem in high-resolution WSI-wise synthesis. The MxIF to H&E synthesis is no exception. Limited by computational resources, the cross-stain image synthesis is typically performed at the patch-level. Thus, discontinuous intensities might be visually identified along with the patch boundaries assembling all individual patches back to a WSI. In this work, we propose a deep learning based unpaired high-resolution image synthesis method to obtain virtual H&E WSIs from MxIF WSIs (each with 27 markers/stains) with reduced tiling effects. Briefly, we first extend the CycleGAN framework by adding simultaneous nuclei and mucin segmentation supervision as spatial constraints. Then, we introduce a random walk sliding window shifting strategy during the optimized inference stage, to alleviate the tiling effects. The validation results show that our spatially constrained synthesis method achieves a 56% performance gain for the downstream cell segmentation task. The proposed inference method reduces the tiling effects by using 50% fewer computation resources without compromising performance. The proposed random sliding window inference method is a plug-and-play module, which can be generalized for other high-resolution WSI image synthesis applications. The source code with our proposed model are available at https://github.com/MASILab/RandomWalkSlidingWindow.git.

PMID:38993526 | PMC:PMC11238901

Mayo Clin Proc Digit Health. 2024 Jun;2(2):221-230. doi: 10.1016/j.mcpdig.2024.03.005. Epub 2024 Apr 15.

ABSTRACT

OBJECTIVE: To validate deep learning models' ability to predict post-transplantation major adverse cardiovascular events (MACE) in patients undergoing liver transplantation (LT).

PATIENTS AND METHODS: We used data from Optum's de-identified Clinformatics Data Mart Database to identify liver transplant recipients between January 2007 and March 2020. To predict post-transplantation MACE risk, we considered patients' demographics characteristics, diagnoses, medications, and procedural data recorded back to 3 years before the LT procedure date (index date). MACE is predicted using the bidirectional gated recurrent units (BiGRU) deep learning model in different prediction interval lengths up to 5 years after the index date. In total, 18,304 liver transplant recipients (mean age, 57.4 years [SD, 12.76]; 7158 [39.1%] women) were used to develop and test the deep learning model's performance against other baseline machine learning models. Models were optimized using 5-fold cross-validation on 80% of the cohort, and model performance was evaluated on the remaining 20% using the area under the receiver operating characteristic curve (AUC-ROC) and the area under the precision-recall curve (AUC-PR).

RESULTS: Using different prediction intervals after the index date, the top-performing model was the deep learning model, BiGRU, and achieved an AUC-ROC of 0.841 (95% CI, 0.822-0.862) and AUC-PR of 0.578 (95% CI, 0.537-0.621) for a 30-day prediction interval after LT.

CONCLUSION: Using longitudinal claims data, deep learning models can efficiently predict MACE after LT, assisting clinicians in identifying high-risk candidates for further risk stratification or other management strategies to improve transplant outcomes based on important features identified by the model.

PMID:38993485 | PMC:PMC11238640 | DOI:10.1016/j.mcpdig.2024.03.005

J Med Signals Sens. 2024 Apr 8;14:9. doi: 10.4103/jmss.jmss_13_23. eCollection 2024.

ABSTRACT

BACKGROUND: Brain tumor segmentation is highly contributive in diagnosing and treatment planning. Manual brain tumor delineation is a time-consuming and tedious task and varies depending on the radiologist's skill. Automated brain tumor segmentation is of high importance and does not depend on either inter- or intra-observation. The objective of this study is to automate the delineation of brain tumors from the Fluid-attenuated inversion recovery (FLAIR), T1-weighted (T1W), T2-weighted (T2W), and T1W contrast-enhanced (T1ce) magnetic resonance (MR) sequences through a deep learning approach, with a focus on determining which MR sequence alone or which combination thereof would lead to the highest accuracy therein.

METHODS: The BraTS-2020 challenge dataset, containing 370 subjects with four MR sequences and manually delineated tumor masks, is applied to train a residual neural network. This network is trained and assessed separately for each one of the MR sequences (single-channel input) and any combination thereof (dual- or multi-channel input).

RESULTS: The quantitative assessment of the single-channel models reveals that the FLAIR sequence would yield higher segmentation accuracy compared to its counterparts with a 0.77 ± 0.10 Dice index. As to considering the dual-channel models, the model with FLAIR and T2W inputs yields a 0.80 ± 0.10 Dice index, exhibiting higher performance. The joint tumor segmentation on the entire four MR sequences yields the highest overall segmentation accuracy with a 0.82 ± 0.09 Dice index.

CONCLUSION: The FLAIR MR sequence is considered the best choice for tumor segmentation on a single MR sequence, while the joint segmentation on the entire four MR sequences would yield higher tumor delineation accuracy.

PMID:38993203 | PMC:PMC11111160 | DOI:10.4103/jmss.jmss_13_23

Comput Methods Programs Biomed. 2024 Jul 4;254:108315. doi: 10.1016/j.cmpb.2024.108315. Online ahead of print.

ABSTRACT

BACKGROUND AND OBJECTIVE: Deep learning usually achieves good performance in the supervised way, which requires a large amount of labeled data. However, manual labeling of electrocardiograms (ECGs) is laborious that requires much medical knowledge. Semi-supervised learning (SSL) provides an effective way of leveraging unlabeled data to improve model performance, providing insight for solving this problem. The objective of this study is to improve the performance of cardiovascular disease (CVD) detection by fully utilizing unlabeled ECG.

METHODS: A novel SSL algorithm fusing consistency regularization and pseudo-labeling techniques (CPSS) is proposed. CPSS consists of supervised learning and unsupervised learning. For supervised learning, the labeled ECGs are mapped into prediction vectors by the classifier. The cross-entropy loss function is used to optimize the classifier. For unsupervised learning, the unlabeled ECGs are weakly and strongly augmented, and a consistency loss is used to minimize the difference between the classifier's predictions for the two augmentations. Pseudo-labeling techniques include positive pseudo-labeling (PL) and ranking-based negative pseudo-labeling (RNL). PL introduces pseudo-labels for data with high prediction confidence. RNL assigns negative pseudo-labels to the lower-ranked categories in the prediction vectors to leverage data with low prediction confidence. In this study, VGGNet and ResNet are used as classifiers, which are jointly optimized by labeled and unlabeled ECGs.

RESULTS: CPSS has been validated on several databases. With the same number of labeled ECGs (10%), it improves the accuracies over pure supervised learning by 13.59%, 4.60%, and 5.38% in the CPSC2018, PTB-XL, and Chapman databases, respectively. CPSS achieves comparable results to the fully supervised method with only 10% of labeled ECGs, which reduces the labeling workload by 90%. In addition, to verify the practicality of CPSS, a cardiovascular disease monitoring system is designed by heterogeneously deploying the trained classifiers on an SoC (system-on-a-chip), which can detect CVD in real time.

CONCLUSION: The results of this study indicate that the proposed CPSS can significantly improve the performance of CVD detection using unlabeled ECG, which reduces the burden of ECG labeling in deep learning. In addition, the designed monitoring system makes the proposed CPSS promising for real-world applications.

PMID:38991373 | DOI:10.1016/j.cmpb.2024.108315

J Neurol Sci. 2024 Jun 10;463:123089. doi: 10.1016/j.jns.2024.123089. Online ahead of print.

ABSTRACT

BACKGROUND: The core clinical sign of Parkinson's disease (PD) is bradykinesia, for which a standard test is finger tapping: the clinician observes a person repetitively tap finger and thumb together. That requires an expert eye, a scarce resource, and even experts show variability and inaccuracy. Existing applications of technology to finger tapping reduce the tapping signal to one-dimensional measures, with researcher-defined features derived from those measures.

OBJECTIVES: (1) To apply a deep learning neural network directly to video of finger tapping, without human-defined measures/features, and determine classification accuracy for idiopathic PD versus controls. (2) To visualise the features learned by the model.

METHODS: 152 smartphone videos of 10s finger tapping were collected from 40 people with PD and 37 controls. We down-sampled pixel dimensions and videos were split into 1 s clips. A 3D convolutional neural network was trained on these clips.

RESULTS: For discriminating PD from controls, our model showed training accuracy 0.91, and test accuracy 0.69, with test precision 0.73, test recall 0.76 and test AUROC 0.76. We also report class activation maps for the five most predictive features. These show the spatial and temporal sections of video upon which the network focuses attention to make a prediction, including an apparent dropping thumb movement distinct for the PD group.

CONCLUSIONS: A deep learning neural network can be applied directly to standard video of finger tapping, to distinguish PD from controls, without a requirement to extract a one-dimensional signal from the video, or pre-define tapping features.

PMID:38991323 | DOI:10.1016/j.jns.2024.123089

Comput Biol Med. 2024 Jul 10;179:108810. doi: 10.1016/j.compbiomed.2024.108810. Online ahead of print.

ABSTRACT

Artificial intelligence (AI) is a field of computer science that involves acquiring information, developing rule bases, and mimicking human behaviour. The fundamental concept behind AI is to create intelligent computer systems that can operate with minimal human intervention or without any intervention at all. These rule-based systems are developed using various machine learning and deep learning models, enabling them to solve complex problems. AI is integrated with these models to learn, understand, and analyse provided data. The rapid advancement of Artificial Intelligence (AI) is reshaping numerous industries, with the pharmaceutical sector experiencing a notable transformation. AI is increasingly being employed to automate, optimize, and personalize various facets of the pharmaceutical industry, particularly in pharmacological research. Traditional drug development methods areknown for being time-consuming, expensive, and less efficient, often taking around a decade and costing billions of dollars. The integration of artificial intelligence (AI) techniques addresses these challenges by enabling the examination of compounds with desired properties from a vast pool of input drugs. Furthermore, it plays a crucial role in drug screening by predicting toxicity, bioactivity, ADME properties (absorption, distribution, metabolism, and excretion), physicochemical properties, and more. AI enhances the drug design process by improving the efficiency and accuracy of predicting drug behaviour, interactions, and properties. These approaches further significantly improve the precision of drug discovery processes and decrease clinical trial costs leading to the development of more effective drugs.

PMID:38991316 | DOI:10.1016/j.compbiomed.2024.108810

Comput Biol Med. 2024 Jul 10;179:108811. doi: 10.1016/j.compbiomed.2024.108811. Online ahead of print.

ABSTRACT

Brain atrophy measurements derived from magnetic resonance imaging (MRI) are a promising marker for the diagnosis and prognosis of neurodegenerative pathologies such as Alzheimer's disease or multiple sclerosis. However, its use in individualized assessments is currently discouraged due to a series of technical and biological issues. In this work, we present a deep learning pipeline for segmentation-based brain atrophy quantification that improves upon the automated labels of the reference method from which it learns. This goal is achieved through tissue similarity regularization that exploits the a priori knowledge that scans from the same subject made within a short interval must have similar tissue volumes. To train the presented pipeline, we use unlabeled pairs of T1-weighted MRI scans having a tissue similarity prior, and generate the target brain tissue segmentations in a fully automated manner using the fsl_anat pipeline implemented in the FMRIB Software Library (FSL). Tissue similarity regularization is enforced during training through a weighted loss term that penalizes tissue volume differences between short-interval scan pairs from the same subject. In inference, the pipeline performs end-to-end skull stripping and brain tissue segmentation from a single T1-weighted MRI scan in its native space, i.e., without performing image interpolation. For longitudinal evaluation, each image is independently segmented first, and then measures of change are computed. We evaluate the presented pipeline in two different MRI datasets, MIRIAD and ADNI1, which have longitudinal and short-interval imaging from healthy controls (HC) and Alzheimer's disease (AD) subjects. In short-interval scan pairs, tissue similarity regularization reduces the quantification error and improves the consistency of measured tissue volumes. In the longitudinal case, the proposed pipeline shows reduced variability of atrophy measures and higher effect sizes of differences in annualized rates between HC and AD subjects. Our pipeline obtains a Cohen's d effect size of d=2.07 on the MIRIAD dataset, an increase from the reference pipeline used to train it (d=1.01), and higher than that of SIENA (d=1.73), a well-known state-of-the-art approach. In the ADNI1 dataset, the proposed pipeline improves its effect size (d=1.37) with respect to the reference pipeline (d=0.80) and surpasses SIENA (d=1.33). The proposed data-driven deep learning regularization reduces the biases and systematic errors learned from the reference segmentation method, which is used to generate the training targets. Improving the accuracy and reliability of atrophy quantification methods is essential to unlock brain atrophy as a diagnostic and prognostic marker in neurodegenerative pathologies.

PMID:38991315 | DOI:10.1016/j.compbiomed.2024.108811

Eur J Radiol. 2024 Jul 3;178:111602. doi: 10.1016/j.ejrad.2024.111602. Online ahead of print.

ABSTRACT

INTRODUCTION: The non-perfused volume divided by total fibroid load (NPV/TFL) is a predictive outcome parameter for MRI-guided high-intensity focused ultrasound (MR-HIFU) treatments of uterine fibroids, which is related to long-term symptom relief. In current clinical practice, the MR-HIFU outcome parameters are typically determined by visual inspection, so an automated computer-aided method could facilitate objective outcome quantification. The objective of this study was to develop and evaluate a deep learning-based segmentation algorithm for volume measurements of the uterus, uterine fibroids, and NPVs in MRI in order to automatically quantify the NPV/TFL.

MATERIALS AND METHODS: A segmentation pipeline was developed and evaluated using expert manual segmentations of MRI scans of 115 uterine fibroid patients, screened for and/or undergoing MR-HIFU treatment. The pipeline contained three separate neural networks, one per target structure. The first step in the pipeline was uterus segmentation from contrast-enhanced (CE)-T1w scans. This segmentation was subsequently used to remove non-uterus background tissue for NPV and fibroid segmentation. In the following step, NPVs were segmented from uterus-only CE-T1w scans. Finally, fibroids were segmented from uterus-only T2w scans. The segmentations were used to calculate the volume for each structure. Reliability and agreement between manual and automatic segmentations, volumes, and NPV/TFLs were assessed.

RESULTS: For treatment scans, the Dice similarity coefficients (DSC) between the manually and automatically obtained segmentations were 0.90 (uterus), 0.84 (NPV) and 0.74 (fibroid). Intraclass correlation coefficients (ICC) were 1.00 [0.99, 1.00] (uterus), 0.99 [0.98, 1.00] (NPV) and 0.98 [0.95, 0.99] (fibroid) between manually and automatically derived volumes. For manually and automatically derived NPV/TFLs, the mean difference was 5% [-41%, 51%] (ICC: 0.66 [0.32, 0.85]).

CONCLUSION: The algorithm presented in this study automatically calculates uterus volume, fibroid load, and NPVs, which could lead to more objective outcome quantification after MR-HIFU treatments of uterine fibroids in comparison to visual inspection. When robustness has been ascertained in a future study, this tool may eventually be employed in clinical practice to automatically measure the NPV/TFL after MR-HIFU procedures of uterine fibroids.

PMID:38991285 | DOI:10.1016/j.ejrad.2024.111602

Can J Cardiol. 2024 Jul 9:S0828-282X(24)00523-3. doi: 10.1016/j.cjca.2024.07.003. Online ahead of print.

ABSTRACT

Leveraging artificial intelligence (AI) for the analysis of electrocardiograms (ECG) has the potential to transform diagnosis and estimate the prognosis of not only cardiac but, increasingly, non-cardiac conditions. In this review, we summarize clinical studies and AI-enhanced ECG-based clinical applications in the early detection, diagnosis, and estimating prognosis of cardiovascular diseases (CVD) in the last five years (2019-2023). With advancements in deep learning and the rapid increased use of ECG technologies, a large number of clinical studies have been published. However, a majority of these studies are single-center, retrospective, proof-of-concept studies that lack external validation. Prospective studies that progress from development toward deployment in clinical settings account for <15% of the studies. Successful implementations of ECG-based AI applications that have received approval from the Food and Drug Administration (FDA) have been developed through commercial collaborations, with about half of them being for mobile or wearable devices. The field is in its early stages, and overcoming several obstacles is essential, such as prospective validation in multi-center large datasets, addressing technical issues, bias, privacy, data security, model generalizability, and global scalability. This review concludes with a discussion of these challenges and potential solutions. By providing a holistic view of the state of AI in ECG analysis, this review aims to set a foundation for future research directions, emphasizing the need for comprehensive, clinically integrated, and globally deployable AI solutions in CVD management.

PMID:38992812 | DOI:10.1016/j.cjca.2024.07.003

BMC Med Imaging. 2024 Jul 11;24(1):172. doi: 10.1186/s12880-024-01338-w.

ABSTRACT

OBJECTIVES: In the interpretation of panoramic radiographs (PRs), the identification and numbering of teeth is an important part of the correct diagnosis. This study evaluates the effectiveness of YOLO-v5 in the automatic detection, segmentation, and numbering of deciduous and permanent teeth in mixed dentition pediatric patients based on PRs.

METHODS: A total of 3854 mixed pediatric patients PRs were labelled for deciduous and permanent teeth using the CranioCatch labeling program. The dataset was divided into three subsets: training (n = 3093, 80% of the total), validation (n = 387, 10% of the total) and test (n = 385, 10% of the total). An artificial intelligence (AI) algorithm using YOLO-v5 models were developed.

RESULTS: The sensitivity, precision, F-1 score, and mean average precision-0.5 (mAP-0.5) values were 0.99, 0.99, 0.99, and 0.98 respectively, to teeth detection. The sensitivity, precision, F-1 score, and mAP-0.5 values were 0.98, 0.98, 0.98, and 0.98, respectively, to teeth segmentation.

CONCLUSIONS: YOLO-v5 based models can have the potential to detect and enable the accurate segmentation of deciduous and permanent teeth using PRs of pediatric patients with mixed dentition.

PMID:38992601 | DOI:10.1186/s12880-024-01338-w

Sci Rep. 2024 Jul 12;14(1):16077. doi: 10.1038/s41598-024-67125-3.

ABSTRACT

Traditionally, constructing training datasets for automatic muscle segmentation from medical images involved skilled operators, leading to high labor costs and limited scalability. To address this issue, we developed a tool that enables efficient annotation by non-experts and assessed its effectiveness for training an automatic segmentation network. Our system allows users to deform a template three-dimensional (3D) anatomical model to fit a target magnetic-resonance image using free-form deformation with independent control points for axial, sagittal, and coronal directions. This method simplifies the annotation process by allowing non-experts to intuitively adjust the model, enabling simultaneous annotation of all muscles in the template. We evaluated the quality of the tool-assisted segmentation performed by non-experts, which achieved a Dice coefficient greater than 0.75 compared to expert segmentation, without significant errors such as mislabeling adjacent muscles or omitting musculature. An automatic segmentation network trained with datasets created using this tool demonstrated performance comparable to or superior to that of networks trained with expert-generated datasets. This innovative tool significantly reduces the time and labor costs associated with dataset creation for automatic muscle segmentation, potentially revolutionizing medical image annotation and accelerating the development of deep learning-based segmentation networks in various clinical applications.

PMID:38992241 | DOI:10.1038/s41598-024-67125-3

Eur Radiol Exp. 2024 Jul 12;8(1):77. doi: 10.1186/s41747-024-00477-7.

ABSTRACT

BACKGROUND: Cardiac magnetic resonance (CMR) in the four-chamber plane offers comprehensive insight into the volumetrics of the heart. We aimed to develop an artificial intelligence (AI) model of time-resolved segmentation using the four-chamber cine.

METHODS: A fully automated deep learning algorithm was trained using retrospective multicentre and multivendor data of 814 subjects. Validation, reproducibility, and mortality prediction were evaluated on an independent cohort of 101 subjects.

RESULTS: The mean age of the validation cohort was 54 years, and 66 (65%) were males. Left and right heart parameters demonstrated strong correlations between automated and manual analysis, with a ρ of 0.91-0.98 and 0.89-0.98, respectively, with minimal bias. All AI four-chamber volumetrics in repeatability analysis demonstrated high correlation (ρ = 0.99-1.00) and no bias. Automated four-chamber analysis underestimated both left ventricular (LV) and right ventricular (RV) volumes compared to ground-truth short-axis cine analysis. Two correction factors for LV and RV four-chamber analysis were proposed based on systematic bias. After applying the correction factors, a strong correlation and minimal bias for LV volumetrics were observed. During a mean follow-up period of 6.75 years, 16 patients died. On stepwise multivariable analysis, left atrial ejection fraction demonstrated an independent association with death in both manual (hazard ratio (HR) = 0.96, p = 0.003) and AI analyses (HR = 0.96, p < 0.001).

CONCLUSION: Fully automated four-chamber CMR is feasible, reproducible, and has the same real-world prognostic value as manual analysis. LV volumes by four-chamber segmentation were comparable to short-axis volumetric assessment.

TRIALS REGISTRATION: ClinicalTrials.gov: NCT05114785.

RELEVANCE STATEMENT: Integrating fully automated AI in CMR promises to revolutionise clinical cardiac assessment, offering efficient, accurate, and prognostically valuable insights for improved patient care and outcomes.

KEY POINTS: • Four-chamber cine sequences remain one of the most informative acquisitions in CMR examination. • This deep learning-based, time-resolved, fully automated four-chamber volumetric, functional, and deformation analysis solution. • LV and RV were underestimated by four-chamber analysis compared to ground truth short-axis segmentation. • Correction bias for both LV and RV volumes by four-chamber segmentation, minimises the systematic bias.

PMID:38992116 | DOI:10.1186/s41747-024-00477-7