Whole-slide images (WSI) glomerulus segmentation is essential for accurately diagnosing kidney diseases. In this work, we propose a general and practical pipeline for glomerulus segmentation that effectively enhances both patch-level and WSI-level segmentation tasks. Our approach leverages stitching on overlapping patches, increasing the detection coverage, especially when glomeruli are located near patch image borders. In addition, we conduct comprehensive evaluations from different segmentation models across two large and diverse datasets with over 30K glomerulus annotations. Experimental results demonstrate that models using our pipeline outperform the previous state-of-the-art method, achieving superior results across both datasets and setting a new benchmark for glomerulus segmentation in WSIs. The code and pre-trained models are available at https://github.com/huuquan1994/wsi_glomerulus_seg.

Magnetic Resonance Imaging (MRI) plays an important role in identifying clinically significant prostate cancer (csPCa), yet automated methods face challenges such as data imbalance, variable tumor sizes, and a lack of annotated data. This study introduces Anomaly-Driven U-Net (adU-Net), which incorporates anomaly maps derived from biparametric MRI sequences into a deep learning-based segmentation framework to improve csPCa identification. We conduct a comparative analysis of anomaly detection methods and evaluate the integration of anomaly maps into the segmentation pipeline. Anomaly maps, generated using Fixed-Point GAN reconstruction, highlight deviations from normal prostate tissue, guiding the segmentation model to potential cancerous regions. We compare the performance by using the average score, computed as the mean of the AUROC and Average Precision (AP). On the external test set, adU-Net achieves the best average score of 0.618, outperforming the baseline nnU-Net model (0.605). The results demonstrate that incorporating anomaly detection into segmentation improves generalization and performance, particularly with ADC-based anomaly maps, offering a promising direction for automated csPCa identification.

ABSTRACTObjectiveTo investigate hippocampal subfield atrophy and biological aging across the mild cognitive impairment (MCI) continuum, we used data from the Alzheimer's Disease Neuroimaging Initiative (ADNI).MethodsA cohort of 49 participants, categorized as cognitively normal (CN, n = 16), early MCI (EMCI, n = 16), or late MCI (LMCI, n = 17), underwent comprehensive neuroimaging, neuropsychological, and genetic assessments. High‐resolution 3D T1‐weighted MRI scans were processed using the volBrain platform and hippocampal subfield segmentation (HIPS) pipeline to quantify hippocampal subfield volumes and estimate biological age. Statistical analyses, including ANCOVA and stepwise regression, were employed to evaluate group differences and identify predictors of hippocampal biological age.ResultsThe results revealed significant volumetric reductions in LMCI, particularly within the CA1, CA4/dentate gyrus (DG), and stratum radiatum/lacunosum/moleculare (SRLM) subfields, with pronounced lateralized effects. Clinical and demographic covariates attenuated group differences in biological age, but volumetric adjustments highlighted a significant distinction between EMCI and LMCI, with EMCI exhibiting a higher biological age. Cognitive performance, as measured by the Montreal Cognitive Assessment (MoCA), emerged as a consistent predictor of biological age, while APOE ε4 carrier status was significantly elevated in LMCI patients. Regression analyses identified divergent contributions of CA2/3 (positively associated) and CA4/DG (negatively associated) volumes to biological age, underscoring the subfield‐specific pathophysiological mechanisms. Asymmetry indices, although variably expressed across groups, offered limited predictive utility, with CA2/3 and CA4/DG asymmetries modestly influencing biological age.ConclusionThese findings support the integration of subfield‐specific hippocampal volumetry and cognitive assessments in early diagnostic frameworks while highlighting the need for longitudinal studies to elucidate causal pathways linking subfield atrophy, biological aging, and cognitive decline.

Hypoxic-ischemic encephalopathy (HIE) is caused by oxygen deprivation to the brain around the time of birth. Therapeutic hypothermia (TH) remains the only validated treatment, and current drug development is hindered by time-consuming manual cell counting and analysis. The field currently lacks robust automated methodologies capable of accurately quantifying cortical and subcortical neuronal cells.This study proposes a two-step cell analysis pipeline for segmentation of neurons and classification of their morphology in fetal sheep brains, with immediate utility in pre-clinical HIE drug discovery. A new dataset containing 180 images and 44,000 cells across 3 treatment groups (including sham, ischemia only, and ischemia treated with hypothermia) from 6 brain regions (CA1, CA3, CA4, DG hippocampal regions and PS1, PS2 parasagittal cortical regions) was manually annotated and used to train a generalized Mask R-CNN, achieving an average precision of 88.3%±1.9% (IoU threshold=0.5). A subsequent dataset (n=1500) of healthy, intermediate, and pyknotic cells was created and used to train a custom, lightweight CNN architecture to classify cells, achieving an overall accuracy of 93.0±0.5% with no healthy-pyknotic misclassifications. The final pipeline outputs novel indicators of HIE-impacted neuronal damage through proportion of healthy neurons. The pipeline's predicted count for number of healthy cells is correlated with original manual quantifications (Spearman's R=0.822, 0.869, 0.919 across treatment groups), validating the model's clinical performance.Clinical relevance - The proposed pipeline can accelerate preclinical drug development for HI in fetal sheep models by minimizing observer bias, reducing labor and time costs, and enabling a more detailed evaluation of neuronal damage.

Various deep learning applications have been developed to aid in the detection of bone abnormalities, such as tumors or osteolytic lesions. We propose a method for the automatic detection and classification of bones in CT scans to determine the anatomical location of these abnormalities. Our approach utilizes three 2D YOLOv5l models to predict the location and anatomical name of bones across the axial, coronal, and sagittal planes. We benchmarked this method against nnU-Net, a widely used 3D segmentation pipeline, using the TotalSegmentator dataset with synthetically generated osteolytic lesions and the RibFrac dataset, which contains annotated rib fractures. Our results show that our method outperforms nnU-Net in identifying lesions that are not located in the vertebrae or ribs, while nnU-Net excels in vertebra-level and rib-fracture localization. However, when predicting a range of possible rib or vertebra levels, rather than the exact levels, our method demonstrates highly accurate performance, outperforming nnU-Net. Overall, depending on the specific application, our work highlights that this multiplanar bone detection approach is a competitive alternative to 3D segmentation models for identifying bone abnormalities in CT scans.Clinical Relevance-In this paper, we have presented a reliable method for the localization and identification of bone tissue in CT scans to automatically provide anatomical information on bone abnormalities, streamlining reporting, and providing a framework for research and epidemiological studies.

Optimize deep learning-based vertebrae segmentation in longitudinal CT scans of multiple myeloma patients using structural uncertainty analysis. Retrospective CT scans from 474 multiple myeloma patients were divided into train (179 patients, 349 scans, 2005-2011) and test cohort (295 patients, 671 scans, 2012-2020). An enhanced segmentation pipeline was developed on the train cohort. It integrated vertebrae segmentation using an open-source deep learning method (Payer's) with a post-hoc structural uncertainty analysis. This analysis identified inconsistencies, automatically correcting them or flagging uncertain regions for human review. Segmentation quality was assessed through vertebral shape analysis using topology. Metrics included 'identification rate', 'longitudinal vertebral match rate', 'success rate' and 'series success rate' and evaluated across age/sex subgroups. Statistical analysis included McNemar and Wilcoxon signed-rank tests, with p<0.05 indicating significant improvement. Payer's method achieved an identification rate of 95.8% and success rate of 86.7%. The proposed pipeline automatically improved these metrics to 98.8% and 96.0%, respectively (p<0.001). Additionally, 3.6% of scans were marked for human inspection, increasing the success rate from 96.0% to 98.8% (p<0.001). The vertebral match rate increased from 97.0% to 99.7% (p<0.001), and the series success rate from 80.0% to 95.4% (p<0.001). Subgroup analysis showed more consistent performance across age and sex groups. The proposed pipeline significantly outperforms Payer's method, enhancing segmentation accuracy and reducing longitudinal matching errors while minimizing evaluation workload. Its uncertainty analysis ensures robust performance, making it a valuable tool for longitudinal studies in multiple myeloma.

The choroid plexus (ChP) is crucial for most cerebrospinal fluid (CSF) secretion. However, few studies have explored the morphology of ChP in a natural upright posture. This study investigated the positional, areal, and volumetric changes of the ChP in response to gravity when comparing supine and upright scanning postures. Thirty-one healthy volunteers underwent MRI scans using an innovative 1.5T rotatable MRI at pitch angles of 0 and 90 employing a 3D magnetization-prepared rapid gradient echo (MP-RAGE) sequence. Based on an in-house ChP segmentation pipeline, gap lengths between the ChP and lateral ventricles (LVEN), ChP surface area, volumes, and signal intensities in different directions were measured. The ChP exhibited a bottom gap decrease (-1.30 mm, P < 0.001), posterior gap increase (1.28 mm, P < 0.001), and bilateral gap growth (0.16 mm, P = 0.015) as well as downward centroidal shift (-1.31 mm, P < 0.001) relative to the LVEN when transitioning from lying to standing positions. Significant distance deviations were noted along the direction of gravity. Along with these positional changes, decreases in surface area (-5.30%, P = 0.021) and volume (-7.53%, P = 0.002) and an increase (11.46%, P < 0.001) in signal intensity of the ChP were observed from lying to standing. This study reveals the positional and volumetric change of the ChP within the LVEN with postural changes and demonstrates its morphology in a typical standing condition. These anatomic changes could provide additional evidence of CSF circulation and intracranial pressure in different postures.

Meningiomas are the most common primary intracranial tumors in adults. For most cases, surgical resection is effective in mitigating recurrence risk. Accurate visualization of meningiomas helps radiologists assess the distribution and volume of the tumor within the brain while assisting neurosurgeons in preoperative planning. This paper introduces an innovative realistic 3D medical visualization system, namely SegR3D. It incorporates a 3D medical image segmentation pipeline, which preprocesses the data via semi-supervised learning-based multi-target segmentation to generate masks of the lesion areas. Subsequently, both the original medical images and segmentation masks are utilized as non-scalar volume data inputs into the realistic rendering pipeline. We propose a novel importance transfer function, assigning varying degrees of importance to different mask values to emphasize the areas of interest. Our rendering pipeline integrates physically based rendering with advanced illumination techniques to enhance the depiction of the structural characteristics and shapes of lesion areas. We conducted a user study involving medical practitioners to evaluate the effectiveness of SegR3D. Our experimental results indicate that SegR3D demonstrates superior efficacy in the visual analysis of meningiomas compared to conventional visualization methods.

Abstract Fetal MRI is increasingly being employed in the diagnosis of fetal lung anomalies and segmentation-derived total fetal lung volumes are used as one of the parameters for prediction of neonatal outcomes. However, in clinical practice, segmentation is performed manually in 2D motion-corrupted stacks with thick slices which is time consuming and can lead to variations in estimated volumes. Furthermore, there is a known lack of consensus regarding a universal lung parcellation protocol and expected normal total lung volume formulas. The lungs are also segmented as one label without parcellation into lobes. In terms of automation, to the best of our knowledge, there have been no reported works on multi-lobe segmentation for fetal lung MRI. This work introduces the first automated deep learning segmentation pipeline for multi-regional lung segmentation for 3D motion-corrected T2w fetal body images for normal anatomy and congenital diaphragmatic hernia cases. The protocol for parcellation into 5 standard lobes was defined in the population-averaged 3D atlas. It was then used to generate a multi-label training dataset including 104 normal anatomy controls and 45 congenital diaphragmatic hernia cases from 0.55T, 1.5T and 3T acquisition protocols. The performance of 3D Attention UNet network was evaluated on 18 cases and showed good results for normal lung anatomy with expectedly lower Dice values for the ipsilateral lung. In addition, we also produced normal lung volumetry growth charts from 290 0.55T and 3T controls. This is the first step towards automated multi-regional fetal lung analysis for 3D fetal MRI.

PurposeTo develop an artificial intelligence (AI) optic cup and disc segmentation pipeline for obtaining optic nerve head (ONH) measurements such as vertical cup-to-disc ratio (VCDR) from fundus images and externally validate performance against optical coherence tomography (OCT) measurements.MethodsThis diagnostic study used a retrospectively collected dataset of 27,252 fundus images associated with 12,477 OCT reports and 21,714 expert assessments of VCDR from electronic health records (EHRs) for 4289 patients inclusive of glaucoma suspects, primary and secondary glaucoma. The AI pipeline was trained on nine public glaucoma datasets and externally validated on a private hospital dataset and a publicly available dataset.ResultsAI VCDR predictions against OCT yielded mean absolute error (MAE), Pearson’s R, and concordance correlation coefficient (CCC) values of 0.097 (95% confidence interval [CI], 0.095–0.099), 0.80 (95% CI, 0.79–0.81), and 0.66 (95% CI, 0.64–0.67), respectively. EHR VCDRs against OCT had MAE, Pearson’s R, and CCC values of 0.086 (95% CI, 0.084–0.087), 0.77 (95% CI, 0.76–0.78), and 0.74 (95% CI, 0.73–0.75), respectively. The coefficient of variation (CV) of the AI pipeline on same-day images was 2.79%.ConclusionsThe proposed AI pipeline had strong correlation with OCT measurements and performed comparably to EHR assessments, with high repeatability. Increased diversity and cardinality of training data improved performance and generalizability to unseen datasets.Translational RelevanceAI pipelines for fundus images can provide ONH measurements such as VCDR near expert level in new patient populations without the need for additional model training.

Abstract BackgroundCT-based radio-biomarkers could provide non-invasive insights into tumour biology to risk-stratify patients. One of the limitations is laborious manual segmentation of regions-of-interest (ROI). We present a deep learning auto-segmentation pipeline for radiomic analysis.Patients and Methods153 patients with resected stage 2A-3B non-small cell lung cancer (NSCLCs) had tumours segmented using nnU-Net with review by two clinicians. The nnU-Net was pretrained with anatomical priors in non-cancerous lungs and finetuned on NSCLCs. Three ROIs were segmented: intra-tumoural, peri-tumoural, and whole lung. 1967 features were extracted using PyRadiomics. Feature reproducibility was tested using segmentation perturbations. Features were selected using minimum-redundancy-maximum-relevance with Random Forest-recursive feature elimination nested in 500 bootstraps.ResultsAuto-segmentation time was ∼36 seconds/series. Mean volumetric and surface Dice-Sørensen coefficient (DSC) scores were 0.84 (±0.28), and 0.79 (±0.34) respectively. DSC were significantly correlated with tumour shape (sphericity, diameter) and location (worse with chest wall adherence), but not batch effects (e.g. contrast, reconstruction kernel). 6.5% cases had ‘missed’ segmentations; 6.5% required major changes. Pre-training on anatomical priors resulted in better segmentations compared to training on tumour-labels alone (p&lt;0.001) and tumour with anatomical labels (p&lt;0.001).Most radiomic features were not reproducible following perturbations and resampling. Adding radiomic features, however, did not significantly improve the clinical model in predicting 2-year disease-free survival: AUCs 0.67 (95%CI 0.59-0.75) vs 0.63 (95%CI 0.54-0.71) respectively (p=0.28).ConclusionOur study demonstrates that integrating auto-segmentation into radio-biomarker discovery is feasible with high efficiency and accuracy. Whilst radiomic analysis show limited reproducibility, our auto-segmentation may allow more robust radio-biomarker analysis using deep learning features.

Human brain fiber tractography using diffusion magnetic resonance imaging is a crucial stage in mapping brain white matter structures, pre-surgical planning, and extracting connectivity patterns. Accurate and reliable tractography, by providing detailed geometric information about the position of neural pathways, minimizes the risk of damage during neurosurgical procedures. Both tractography itself and its post-processing steps such as bundle segmentation are usually used in these contexts. Many approaches have been put forward in the past decades and recently, multiple data-driven tractography algorithms and automatic segmentation pipelines have been proposed to address the limitations of traditional methods. Several of these recent methods are based on learning algorithms that have demonstrated promising results. In this study, in addition to introducing diffusion MRI datasets, we review learning-based algorithms such as conventional machine learning, deep learning, reinforcement learning and dictionary learning methods that have been used for white matter tract, nerve and pathway recognition as well as whole brain streamlines or whole brain tractogram creation. The contribution is to discuss both tractography and tract recognition methods, in addition to extending previous related reviews with most recent methods, covering architectures as well as network details, assess the efficiency of learning-based methods through a comprehensive comparison in this field, and finally demonstrate the important role of learning-based methods in tractography.

We introduce NeuroVerse, a framework for exploring 3D nanometric-scale reconstructions of neural and glial cellular processes in the central nervous system. Using image stacks from volume electron microscopy, NeuroVerse generates 3D mesh models through a SAM2-based segmentation pipeline and integrates absorption signals for deployment in a Metaverse environment. The framework includes a SAM2 adapter optimized for biological microscopy imaging, adapted with feature enhancement blocks and dual decoders to improve the segmentation of complex cellular structures. An interactive virtual AI agent, powered by Heygen and OpenAI models with domain-specific knowledge, provides semi-real-time assistance. NeuroVerse supports education and collaborative analysis for neuroanatomy and neuroscience. It includes a pipeline for the creation of 3D models, automated segmentation, mesh reconstruction, and heatmap computation, optimized for the Spatial.io ecosystem. Contributions include a virtual anatomy lab for neuroanatomy education and collaborative sessions on spatial morphology correlation and neuroenergetic absorption models. Evaluations show that the SAM2 adapter preserves fine cellular details and manages irregular boundaries. Preliminary sessions indicate potential to enhance neuroscience education, improve remote collaboration among scientists, and provide access to advanced neuroscientific data and tools. Evaluation of the virtual AI agent confirms its ability to provide context-aware support, interpret complex cellular structures, and facilitate understanding through semi-real-time assistance for students analyzing neural and glial reconstructions. NeuroVerse combines imaging, segmentation, and AI technologies within an immersive Metaverse platform for neuroscience education and research. • Development of digital twin of the Human Anatomy Institute of the University of Turin. • AI-Based pipeline for EM image segmentation and Data interpretation. • Case study for multiple usage for Data Analysis and Education in the Metaverse.

The impact of training image quality with a novel protocol on artificial intelligence-based LGE-MRI image segmentation for potential atrial fibrillation management.

Automated leaf segmentation pipelines must balance accuracy, scalability, and usability to be readily adopted in plant research. We present an end-to-end deep learning pipeline designed for practical use in plant phenotyping, which we developed and evaluated during a real-world plant growth experiment using Atriplex lentiformis. The pipeline integrates a fine-tuned Mask Region-based Convolutional Neural Network (Mask R-CNN) segmentation model trained on 176 plant images and achieves high performance despite the small training data set (Dice coefficient = 0.781). We quantitatively compare the fine-tuned Mask R-CNN model to Meta AI’s Segment Anything Model (SAM) and evaluate natural language prompts using Grounded SAM and the Leaf-Only SAM post-processing pipeline for refining segmentation outputs. Our findings highlight that transfer learning on a specialized data set can still outperform a large foundation model in domain-specific tasks. In addition, we integrate QR codes for automated sample identification and benchmark multiple QR code decoding libraries, evaluating their robustness under real-world imaging conditions like distortion and lighting variation. To ensure accessibility, we deploy the pipeline as a user-friendly Streamlit web application, allowing researchers to analyze images without deep learning expertise. By focusing on practical deployment in addition to model performance, this study provides an open-source, scalable framework for plant science applications and addresses real-world challenges in automation and usability by the end-researcher.

Traditional computed tomography (CT) methods for 3D reconstruction face resolution limitations and require time-consuming post-processing workflows. While deep learning techniques improve the accuracy of segmentation, traditional voxel-based segmentation and surface reconstruction pipelines tend to introduce artifacts such as disconnected regions, topological inconsistencies, and stepped distortions. To overcome these challenges, we propose ToPoMesh, an end-to-end 3D mesh reconstruction deep learning framework for direct reconstruction of high-fidelity surface meshes from CT volume data. To address the existing problems, our approach introduces three core innovations: (1) accurate local and global shape modeling by preserving and enhancing local feature information through residual connectivity and self-attention mechanisms in graph convolutional networks; (2) an adaptive variant density (Avd) mesh de-pooling strategy, which dynamically optimizes the vertex distribution; (3) a topology modification module that iteratively prunes the error surfaces and boundary smoothing via variable regularity terms to obtain finer mesh surfaces. Experiments on the LiTS, MSD pancreas tumor, MSD hippocampus, and MSD spleen datasets demonstrate that ToPoMesh outperforms state-of-the-art methods. Quantitative evaluations demonstrate a 57.4% reduction in Chamfer distance (liver) and a 0.47% improvement in F-score compared to end-to-end 3D reconstruction methods, while qualitative results confirm enhanced fidelity for thin structures and complex anatomical topologies versus segmentation frameworks. Importantly, our method eliminates the need for manual post-processing, realizes the ability to reconstruct 3D meshes from images, and can provide precise guidance for surgical planning and diagnosis.

ABSTRACT Translating quantitative skeletal muscle MRI biomarkers into clinics requires efficient automatic segmentation methods. The purpose of this work is to investigate a simple yet effective iterative methodology for building a high‐quality automatic segmentation model while minimizing the manual annotation effort. We used a retrospective database of quantitative MRI examinations ( n = 70) of healthy and pathological thighs for training a nnU‐Net segmentation model. Healthy volunteers and patients with various neuromuscular diseases, broadly categorized as dystrophic, inflammatory, neurogenic, and unlabeled NMDs. We designed an iterative procedure, progressively adding cases to the training set and using a simple visual five‐level rating scale to judge the validity of generated segmentations for clinical use. On an independent test set ( n = 20), we assessed the quality of the segmentation in 13 individual thigh muscles using standard segmentation metrics—dice coefficient (DICE) and 95% Hausdorff distance (HD95)—and quantitative biomarkers—cross‐sectional area (CSA), fat fraction (FF), and water‐T1/T2. We obtained high‐quality segmentations (DICE = 0.88 ± 0.15/0.86 ± 0.14, HD95 = 6.35 ± 12.33/6.74 ± 11.57 mm), comparable to recent works, although with a smaller training set ( n = 30). Inter‐rater agreement on the five‐level scale was fair to moderate but showed progressive improvement of the segmentation model along with the iterations. We observed limited differences from manually delineated segmentations on the quantitative outcomes (MAD: CSA = 65.2 mm 2 , FF = 1%, water‐T1 = 8.4 ms, water‐T2 = 0.35 ms), with variability comparable to manual delineations.

Organoids replicate tissue architecture and function and offer a unique opportunity to explore the impact of external perturbations in vitro. However, conducting large-scale screening procedures to investigate the effects of various stresses on cellular morphology and topology in these systems poses important challenges, including limitations in high-resolution three-dimensional (3D) imaging and accessible 3D analysis platforms. In this study, we introduce an AI-based multilevel segmentation and cellular topology pipeline for screening morphology and topology modifications in 3D cell culture at both the nuclear and cytoplasmic levels, as well as at the whole-organoid scale. We demonstrate the versatility of our approach through proof-of-concept experiments, encompassing well-characterized conditions and poorly explored mechanical stressors such as microgravity. By offering a user-friendly interface named 3DCellScope and a comprehensive set of tools for discovery-like assays in screening 3D organoid models, our pipeline demonstrates wide-ranging potential for applications in biomedical research.

Background:Accurate analysis of neuronal primary cilia is essential for understanding developmental processing of neurons. But existing image segmentation methods struggle with staining variability and background noise. To address this, we developed a more robust segmentation and statistical analysis pipeline using an animal model small sample size and with known neuronal microstructure alterations.Methods:Maternal obesity was induced in mice via a high-fat/high-sucrose diet. Hippocampal tissue from 6-month-old offspring of obese and control dams was analyzed. We developed a MATLAB-based pipeline to segment neuronal cilia from z-stack images, applying mathematical transformations and using the Weibull distribution and Bayesian Information Criterion (BIC) to assess group differencesResults:The technique segmented cilia despite artifacts, revealing group-specific patterns. Statistical analysis confirmed significant differences, highlighting the method’s robustness over traditional tests, especially with small samples.Conclusion:Our method reliably segments neuronal primary cilia in immune-stained sections with thionin-counter staining and offers a sensitive, assumption-free alternative to traditional statistical tests, ideal for small-sample neurobiological studies

Quantitative cardiovascular PET/CT imaging is useful in the diagnosis of multiple cardiac perfusion and motion pathologies. The common approach for cardiac segmentation consists in using co-registered CT images, exploiting publicly available deep learning (DL)-based segmentation models. However, the mismatch between structural CT images and PET uptake limits the usefulness of these approaches. Besides, the performance of DL models is not consistent over low-dose or ultra-low-dose CT images commonly used in clinical PET/CT imaging. In this work, we developed a DL-based methodology to tackle this issue by segmenting directly cardiac PET images. This study included 406 cardiac PET images from 146 patients (43 18F-FDG, 329 13N-NH3, and 37 82Rb images). Using previously trained DL nnU-Net models in our group, we segmented the whole heart and the three main cardiac components, namely the left myocardium (LM), left ventricle cavity (LV), and right ventricle (RV) on co-registered CT images. The segmentation was resampled to PET resolution and edited through a combination of automated image processing and manual correction. The corrected segmentation masks and SUV PET images were fed to a nnU-Net V2 pipeline to be trained in fivefold data split strategy by defining two tasks: task #1 for whole cardiac segmentation and task #2 for segmentation of three cardiac components. Fifteen cardiac images were used as external validation set. The DL delineated masks were compared with standard of reference masks using Dice coefficient, Jaccard distance, mean surface distance, and segment volume relative error (%). Task #1 average Dice coefficient in internal validation fivefold was 0.932 ± 0.033. The average Dice on the 15 external cases were comparable with the fivefold Dice reaching an average of 0.941 ± 0.018. Task #2 average Dice in fivefold validation was 0.88 ± 0.063, 0.828 ± 0.091, and 0.876 ± 0.062 for LM, LV, and RV, respectively. There was no statistically significant difference among the Dice coefficients, neither between images acquired by three radiotracers nor between the different folds (P-values > > 0.05). The overall average volume prediction error in cardiac components segmentation was less than 2%. We developed an automated DL-based segmentation pipeline to segment the whole heart and cardiac components with acceptable accuracy and robust performance in the external test set and over three radiotracers used in nuclear cardiovascular imaging. The proposed methodology can overcome unreliable segmentations performed on CT images.