Robust semi-automatic segmentation method: an expert assistant tool for muscles in CT and MR data

Background and objectivesBody composition (BC) assessment by computed tomography (CT) at the level of the third lumbar vertebra (L3) is considered the gold standard for evaluating muscle and adipose compartments. However, manual segmentation is time-consuming and impractical for large data sets. This study aimed to compare the automated method, Identificação Automatizada da Composição Corporal (IACC), with manual segmentation performed using 3D Slicer.MethodsThis retrospective cross-sectional study included 126 participants, each contributing one single axial CT slice at the L3 level, obtained from routine clinical CT scans acquired between November 2023 and January 2025. Manual segmentations were performed in 3D Slicer by a trained evaluator and validated by an experienced radiologist. The IACC automatically identified the L3 level and segmented skeletal muscle (MUSCLE), subcutaneous adipose tissue (SAT), and visceral adipose tissue (VAT). Analyses included intra- and inter-rater reproducibility (ICC, Dice coefficient), Pearson correlation, agreement (ICC), and Bland–Altman plots. ResultsThe mean analysis time was approximately 5 min per scan with IACC, compared with 25–30 min using the manual method. Intra-rater reproducibility was excellent (ICC > 0.99), and inter-rater agreement demonstrated high Dice coefficients for SAT (0.982), skeletal muscle (0.940), and VAT (0.932). Concordance between methods was high for SAT (ICC = 0.932) and VAT (ICC = 0.958), and good for skeletal muscle (ICC = 0.784), all with Pearson’s r > 0.97. Bland–Altman analysis showed minimal mean bias for VAT (− 0.35 cm2) and small variability for SAT, indicating excellent agreement for adipose tissue compartments. In contrast, skeletal muscle presented greater disagreement, with a larger negative bias and wider limits of agreement, reflecting higher variability in automated measurements.ConclusionsIACC demonstrated high accuracy and efficiency for automated segmentation of SAT and muscle, substantially reducing analysis time compared to manual segmentation. Although greater variability was observed for skeletal muscle measurements, the tool shows promise for use in large-scale population studies and potential clinical applications, particularly, where rapid and standardized analyses are required.

BackgroundThe evaluation of automatic segmentation algorithms is commonly performed using geometric metrics. An analysis based on dosimetric parameters might be more relevant in clinical practice but is often lacking in the literature. The aim of this study was to investigate the impact of state-of-the-art 3D U-Net-generated organ delineations on dose optimization in radiation therapy (RT) for prostate cancer patients.MethodsA database of 69 computed tomography images with prostate, bladder, and rectum delineations was used for single-label 3D U-Net training with dice similarity coefficient (DSC)-based loss. Volumetric modulated arc therapy (VMAT) plans have been generated for both manual and automatic segmentations with the same optimization settings. These were chosen to give consistent plans when applying perturbations to the manual segmentations. Contours were evaluated in terms of DSC, average and 95% Hausdorff distance (HD). Dose distributions were evaluated with the manual segmentation as reference using dose volume histogram (DVH) parameters and a 3%/3 mm gamma-criterion with 10% dose cut-off. A Pearson correlation coefficient between DSC and dosimetric metrics, i.e. gamma index and DVH parameters, has been calculated.Results3D U-Net-based segmentation achieved a DSC of 0.87 (0.03) for prostate, 0.97 (0.01) for bladder and 0.89 (0.04) for rectum. The mean and 95% HD were below 1.6 (0.4) and below 5 (4) mm, respectively. The DVH parameters, V_{60/65/70,{mathrm{Gy}}} for the bladder and V_{50/65/70,{mathrm{Gy}}} for the rectum, showed agreement between dose distributions within pm , 5% and pm ,2%, respectively. The D_{98/2%} and V_{95%}, for prostate and its 3 mm expansion (surrogate clinical target volume) showed agreement with the reference dose distribution within 2% and 3 Gy with the exception of one case. The average gamma pass-rate was 85%. The comparison between geometric and dosimetric metrics showed no strong statistically significant correlation.ConclusionsThe 3D U-Net developed for this work achieved state-of-the-art geometrical performance. Analysis based on clinically relevant DVH parameters of VMAT plans demonstrated neither excessive dose increase to OARs nor substantial under/over-dosage of the target in all but one case. Yet the gamma analysis indicated several cases with low pass rates. The study highlighted the importance of adding dosimetric analysis to the standard geometric evaluation.

To better understand the outcomes after stereotactic radiosurgery (SRS) for brain metastases, we have created a registry that archives MRI studies alongside clinical data in this population. To consider outcomes quantitatively, each metastatic lesion must be segmented to define a 3D volume. In large populations, lesion segmentation is time consuming (and expensive when this requires an experienced Radiation Oncologist or Neurosurgeon) to manually segment each lesion slice by slice. We sought to determine whether semi-automated segmentation using the Growth-Cut toolbox in 3DSlicer (Egger et al. 2013) provides contours equivalent to those created by physicians using manual segmentation methods. A subset of 40 lesions that had undergone manual segmentation for SRS planning were selected to compare with semi-automated segmentation methods. Using the 3DSlicer Growth -Cut toolbox, a novice member of the team partially outlined the lesion tumor and non-tumor tissues. The toolbox then propagated lesion boundaries and created a complete 3D volume. Dice Coefficient (DC), Hausdorff Distance (HD), and Modified Hausdorf Distance (MHD), were used to establish volume similarity between Growth Cut toolbox output and manual segmentation on metastatic lesions. Additionally, lesion volume, diffusion (ADC), and perfusion (rCBV) were analyzed to determine if segmentation discrepancies resulted in differences in imaging parameter estimates. There was strong similarity of the 3D volumes according to the DC (µ = 0.77), HD (µ = 3.89), and MHD (µ = 0.85). Imaging parameters between segmentation methods, agreed well for both diffusion (ADC R2 = 0.94), and perfusion (rCBV R2 = 0.986). Using 3DSlicer did not substantially decrease tracing time; however, it produced similar results by a novice to those of experts. This study supports the use of the Growth Cut toolbox when expert manual segmentation does not exist. This finding is particularly useful for large scale studies that would otherwise require significant clinician involvement for ROI generation.

BackgroundMagnetic resonance imaging (MRI) quantification of intramuscular fat accumulation is a responsive biomarker in neuromuscular diseases. Despite emergence of automated methods, manual muscle segmentation remains an essential foundation. We aimed to develop a training programme for new observers to demonstrate competence in lower limb muscle segmentation and establish reliability benchmarks for future human observers and machine learning segmentation packages.MethodsThe learning phase of the training programme comprised a training manual, direct instruction, and eight lower limb MRI scans with reference standard large and small regions of interest (ROIs). The assessment phase used test–retest scans from two patients and two healthy controls. Interscan and interobserver reliability metrics were calculated to identify underperforming outliers and to determine competency benchmarks.ResultsThree experienced observers undertook the assessment phase, whilst eight new observers completed the full training programme. Two of the new observers were identified as underperforming outliers, relating to variation in size or consistency of segmentations; six had interscan and interobserver reliability equivalent to those of experienced observers. The calculated benchmark for the Sørensen-Dice similarity coefficient between observers was greater than 0.87 and 0.92 for individual thigh and calf muscles, respectively. Interscan and interobserver reliability were significantly higher for large than small ROIs (all p < 0.001).ConclusionsWe developed, implemented, and analysed the first formal training programme for manual lower limb muscle segmentation. Large ROI showed superior reliability to small ROI for fat fraction assessment.Relevance statementObservers competent in lower limb muscle segmentation are critical to application of quantitative muscle MRI biomarkers in neuromuscular diseases. This study has established competency benchmarks for future human observers or automated segmentation methods.Key points• Observers competent in muscle segmentation are critical for quantitative muscle MRI biomarkers.• A training programme for muscle segmentation was undertaken by eight new observers.• We established competency benchmarks for future human observers or automated segmentation methods.Graphical

Three-dimensional (3D) mesh reconstruction of the cardiac anatomy from medical images is useful for shape and motion measurements and biophysics simulations. However, 3D medical images are often acquired as 2D slices that are sparsely sampled (e.g., large slice spacing) and noisy, and 3D mesh reconstruction on such data is a challenging task. Traditional voxel-based approaches utilize non-differentiable pre- and post-processing that compromises fidelity to images, while mesh-level deep learning approaches require large 3D mesh annotations that are difficult to obtain. Differentiable cross-domain supervision from 2D images to 3D meshes is therefore crucial for enabling end-to-end optimization in medical imaging. While there have been attempts to approximate the voxelization and slicing of meshes that are being optimized, there has not yet been a method for directly using 2D slices to supervise 3D mesh reconstruction in a differentiable manner. Here, we propose a novel explicit differentiable voxelization and slicing (DVS) algorithm allowing gradient backpropagation to a 3D mesh from its slices, which facilitates refined mesh optimization directly supervised by the losses defined on 2D images. Further, we propose an innovative framework for extracting patient-specific left ventricle (LV) meshes from medical images by coupling DVS with a graph harmonic deformation (GHD) mesh morphing descriptor of cardiac shape that naturally preserves mesh quality and smoothness during optimization. The proposed framework achieves state-of-the-art performance in cardiac mesh reconstruction tasks from densely sampled (CT) as well as sparsely sampled (MRI stack with few slices) images, outperforming alternatives, including Marching Cubes, statistical shape models, algorithms with vertex-based mesh morphing algorithms and alternative methods for image-supervision of mesh reconstruction. Experimental results demonstrate that our method achieves an overall Dice score of 90% during a sparse fitting on multi-datasets. The proposed method can further quantify clinically useful parameters such as ejection fraction and global myocardial strains, closely matching the ground truth and outperforming the traditional voxel-based approach in sparse images.

Automated segmentation of skeletal muscle from computed tomography (CT) images is essential for large-scale quantitative body composition analysis. However, manual segmentation is time-consuming and impractical for routine or high-throughput use. This study presents a systematic comparison of two-dimensional (2D) and three-dimensional (3D) deep learning architectures for segmenting skeletal muscle at the anatomically standardized level of the third lumbar vertebra (L3) in low-dose computed tomography (LDCT) scans. We implemented and evaluated the DeepLabv3+ (2D) and UNet3+ (3D) architectures on a curated dataset of 534 LDCT scans, applying preprocessing protocols, L3 slice selection, and region of interest extraction. The model performance was evaluated using a comprehensive set of evaluation metrics, including Dice similarity coefficient (DSC) and 95th percentile Hausdorff distance (HD95). DeepLabv3+ achieved the highest segmentation accuracy (DSC = 0.982 ± 0.010, HD95 = 1.04 ± 0.46 mm), while UNet3+ showed competitive performance (DSC = 0.967 ± 0.013, HD95 = 1.27 ± 0.58 mm) with 26 times fewer parameters (1.27 million vs. 33.6 million) and lower inference time. Both models exceeded or matched results reported in the recent CT-based muscle segmentation literature. This work offers practical insights into architecture selection for automated LDCT-based muscle segmentation workflows, with a focus on the L3 vertebral level, which remains the gold standard in muscle quantification protocols.

Magnetic resonance imaging (MRI)-based brain segmentation has recently been revolutionized by deep learning methods. These methods use large numbers of annotated segmentations to train algorithms that have the potential to perform brain segmentations reliably and quickly. However, training data for these algorithms are frequently obtained from automated brain segmentation systems, which may contain inaccurate neuroanatomy. Thus, the neuroimaging community would benefit from an open source database of high quality, neuroanatomically curated and manually edited MRI brain images, as well as the publicly available tools and detailed procedures for generating these curated data. Manual segmentation approaches are regarded as the gold standard for brain segmentation and parcellation. These approaches underpin the construction of neuroanatomically accurate human brain atlases. In addition, neuroanatomically precise definitions of MRI-based regions of interest (ROIs) derived from manual brain segmentation are essential for accuracy in structural connectivity studies and in surgical planning for procedures such as deep brain stimulation. However, manual segmentation procedures are time and labor intensive, and not practical in studies utilizing very large datasets, large cohorts, or multimodal imaging. Automated segmentation methods were developed to overcome these issues, and provide high data throughput, increased reliability, and multimodal imaging capability. These methods utilize manually labeled brain atlases to automatically parcellate the brain into different ROIs, but do not have the anatomical accuracy of skilled manual segmentation approaches. In the present study, we developed a custom software module for manual editing of brain structures in the freely available 3D Slicer software platform that employs principles and tools based on pioneering work from the Center for Morphometric Analysis (CMA) at Massachusetts General Hospital. We used these novel 3D Slicer segmentation tools and techniques in conjunction with well-established neuroanatomical definitions of subcortical brain structures to manually segment 50 high resolution T1w MRI brains from the Human Connectome Project (HCP) Young Adult database. The structural definitions used herein are associated with specific neuroanatomical ontologies to systematically interrelate histological and MRI-based morphometric definitions. The resulting brain datasets are publicly available and will provide the basis for a larger database of anatomically curated brains as an open science resource.

Objective To investigate the application of CT image and cone beam computed tomography (CBCT) image registration based on 3D Slicer software in image-guided radiotherapy for uterine cervical neoplasms. Methods Based on 3D Slicer software and Slicer RT toolkit, 10 positioning CT images and 50 CBCT images of 10 patients with uterine cervical neoplasms in Henan Provincial People's Hospital between January 2018 and October 2018 had rigid registration and b-spline deformation registration respectively. The dice similarity coefficient (DSC) and Hausdorff distance (HD) of the bladder, rectum, femoral head, spinal cord, and body of CT-CBCT images were compared by using paired t-test before and after the registration. Results Pre-registration, rigid registration and after b-spline deformation registration of CT images and CBCT images, the DSC in the bladder (0.459±0.177, 0.528±0.184, 0.542±0.187, respectively), the rectum (0.564±0.141, 0.632±0.091, 0.684±0.097, respectively), the femoral head (0.695±0.088, 0.833±0.030, 0.865±0.027, respectively), the spinal cord (0.587±0.119, 0.746±0.085, 0.834±0.032, respectively) and the body surface (0.922±0.013, 0.948±0.011, 0.959±0.009, respectively) showed an increased trend; HD in the bladder (12.8±7.2, 12.2±7.1, 11.7±7.3, respectively), the rectum (5.0±1.8, 4.4±1.2, 3.4±1.2, respectively), the femoral head (3.6±1.2, 1.8±0.5, 1.5±0.5, respectively), the spinal cord (4.0±1.0, 2.7±1.3, 1.8±0.5, respectively) and the body surface (6.3±2.1, 5.2±2.0, 4.3±2.0, respectively) showed a decreased trend. The differences of pairwise comparison in the same parts were statistically significant (all P < 0.05). Conclusions Both rigid registration and b-spline deformation registration of CT-CBCT images based on 3D Slicer software can improve the radiotherapy accuracy of uterine cervical neoplasms, and b-spline deformation registration has more significant advantages. Key words: Uterine cervical neoplasms; Radiotherapy; 3D Slicer software; Cone-beam computed tomography; Image registration; Image-guided radiation therapy

Simple SummaryUterine cervical cancer (CC) is a leading cause of cancer-related deaths in women worldwide. Pelvic magnetic resonance imaging (MRI) allows the assessment of local tumor extent and guides the choice of primary treatment. MRI tumor segmentation enables whole-volume radiomic tumor profiling, which is potentially useful for prognostication and individualization of therapy in CC. Manual tumor segmentation is, however, labor intensive and thus not part of routine clinical workflow. In the current work, we trained a deep learning (DL) algorithm to automatically segment the primary tumor in CC patients. Although the achieved segmentation performance of the trained DL algorithm is slightly lower than that for human experts, it is still relatively good. This study suggests that automated MRI primary tumor segmentations by DL algorithms without any human interaction is possible in patients with CC.Uterine cervical cancer (CC) is the most common gynecologic malignancy worldwide. Whole-volume radiomic profiling from pelvic MRI may yield prognostic markers for tailoring treatment in CC. However, radiomic profiling relies on manual tumor segmentation which is unfeasible in the clinic. We present a fully automatic method for the 3D segmentation of primary CC lesions using state-of-the-art deep learning (DL) techniques. In 131 CC patients, the primary tumor was manually segmented on T2-weighted MRI by two radiologists (R1, R2). Patients were separated into a train/validation (n = 105) and a test- (n = 26) cohort. The segmentation performance of the DL algorithm compared with R1/R2 was assessed with Dice coefficients (DSCs) and Hausdorff distances (HDs) in the test cohort. The trained DL network retrieved whole-volume tumor segmentations yielding median DSCs of 0.60 and 0.58 for DL compared with R1 (DL-R1) and R2 (DL-R2), respectively, whereas DSC for R1-R2 was 0.78. Agreement for primary tumor volumes was excellent between raters (R1-R2: intraclass correlation coefficient (ICC) = 0.93), but lower for the DL algorithm and the raters (DL-R1: ICC = 0.43; DL-R2: ICC = 0.44). The developed DL algorithm enables the automated estimation of tumor size and primary CC tumor segmentation. However, segmentation agreement between raters is better than that between DL algorithm and raters.

Genus zero graph segmentation: Estimation of intracranial volume

PurposeTo show feasibility of computerized techniques for ocular redness quantification in clinical studies, and to propose an automatic, objective method.MethodsSoftware for quantification of redness of the bulbar conjunctiva was developed. It provides an interface for manual and automatic sclera segmentation along with automated alignment of region of interest to enable estimation of changes in redness. The software also includes the redness scoring methods: (1) contrast-limited adaptive histogram equalization (CLAHE) in red-green-blue (RGB) color model, (2) product of saturation and hue in hue-saturation-value (HSV), and (3) average of angular sections in HSV. Our validation pipeline compares the scoring outcomes from the perspectives of segmentation reliability, segmentation precision, segmentation automation, and the choice of redness scoring methods.ResultsNinety-two photographs of eyes before and after provoked redness were evaluated. Redness in manually segmented images was significantly different within human observers (interobserver, P = 0.04) and two scoring sessions (intraobserver, P < 0.001). Automated segmentation showed the smallest variability, and can therefore be seen as a robust segmentation method. The RGB-based scoring method was less sensitive in redness assessment.ConclusionsComputation of ocular redness depends heavily on sclera segmentation. Manual segmentation appears to be subjective, resulting in systematic errors in intraobserver and interobserver settings. At the same time, automatic segmentation seems to be consistent. The scoring methods relying on HSV color space appeared to be more consistent.Translational RelevanceComputerized quantification of ocular redness holds great promise to objectify ocular redness in the standard clinical care and, in particular, in clinical trials.

PurposeA lightweight deep learning network SMA-Net was proposed to intelligently segment the skeletal muscle of the third lumbar (L3) level in patients with cervical cancer radiotherapy, and the segmentation performance of the network was evaluated.Methods and MaterialsA total of 160 eligible patients with cervical cancer admitted to the oncology department of our hospital from September 2021 to June 2024 were randomly divided into training set (N = 112), validation set (N = 16) and test set (N = 32) according to 7 : 1 : 2. The lightweight Mamba architecture is introduced into the UNet network, and the SAB and CAB attention mechanisms are introduced on the skip connection. The attention mechanism is used to suppress the irrelevant information in the image and highlight the important local features. The trained network is geometrically evaluated on the test set for segmentation performance, comparison of manual segmentation and predicted skeletal muscle area (SMA). Compare the parameters and computations of SMA-Net with existing networks.ResultsThe dice similarity coefficient of SMA-Net network for skeletal muscle segmentation was 89.16%, the sensitivity SEN was 88.21%, the positive predictive value PPV was 90.13% and the 95% Hausdorff distance was 5.30mm. Manual segmentation is basically close to the predicted SMA. Our proposed network for cervical cancer patients predicted sarcopenia with 87.5% accuracy, 92.31% precision, 80% recall, 85.72% F1-Score, and 0.871 AUC. The calculation amount of SMA-UNet network is 1.50 GFLOPS, and the parameter amount is 1.24 M. The radiologist's scores show that minor and no revision accounted for 93.75% on manual revision of skeletal muscle.ConclusionThe lightweight SMA-Net proposed in this study can accurately segment L3 skeletal muscle and quickly calculate its area, which basically meets the clinical application and is convenient for clinical deployment. It is helpful for clinicians to quickly diagnose sarcopenia in patients with cervical cancer, save medical resources, reduce the workload of physicians, and improve diagnostic efficiency.

Manual segmentation in lumbar paraspinal muscle studies is time-consuming, subject to variability, and limits comprehensive evaluation. To solve the problems, we aim to develop a deep learning algorithm for the automatic segmentation and multi-parameter quantification of eight paraspinal muscles in lumbar CT. We collected CT images covering the range from the superior endplate of the L1 vertebra to the inferior endplate of the S1 vertebra. The dataset was partitioned into training, validation, and test sets in a ratio of 7:1:2. Six convolutional neural networks were employed to automatically segment the bilateral psoas major, quadratus lumborum, erector spinae, and multifidus across all slices. Model performance was evaluated using the Dice Similarity Coefficient (DSC), Hausdorff Distance (HD), and mean Intersection over Union (mIoU). Muscle parameters including cross-sectional area, muscle volume, fat infiltration, CT density, and paraspinal muscle index were calculated. The Intraclass Correlation Coefficient [ICC(3,1)] was used to assess agreement between model's segmentations and manual ground truth. 100 lumbar CT scans (age, 55.02 ± 16.2; 62 female) were used for model development. The mean DSC across all six models ranged from 0.754 to 0.903. TransUNet demonstrated the best overall (DSC, 0.903; 95% CI: 0.902, 0.913) (HD, 9.934mm; 95% CI: 9.815, 11.902) (mIoU, 0.841; 95% CI: 0.839, 0.850). For individual muscles segmented by TransUNet, the highest DSC was 0.936 (95% CI: 0.934-0.938) for the right multifidus. The mean ICC for all parameters was 0.931. The highest ICC value was 0.999 (p < 0.001). The deep learning tool enables automatic and accurate segmentation of eight lumbar paraspinal muscles. Muscle parameter measurements derived from model's segmentations and manual ground truth exhibit high agreement. Accurate segmentation and quantification of paraspinal muscle parameters facilitates large-scale epidemiological studies of paraspinal muscles and spine-related diseases, overcoming the limitations in manual methods.

Breast cancer, the most common cancer type among women worldwide, requires early detection and accurate diagnosis for improved treatment outcomes. Segmenting fat and fibroglandular tissue (FGT) in magnetic resonance imaging (MRI) is essential for creating volumetric models, enhancing surgical workflow, and improving clinical outcomes. Manual segmentation is time-consuming and subjective, prompting the development of automated deep-learning algorithms to perform this task. However, configuring these algorithms for 3D medical images is challenging due to variations in image features and preprocessing distortions. Automated machine learning (AutoML) frameworks automate model selection, hyperparameter tuning, and architecture optimization, offering a promising solution by reducing reliance on manual intervention and expert knowledge. We compare nnU-Net and Auto3Dseg, two AutoML frameworks, in segmenting fat and FGT on T1-weighted MRI images from the Duke breast MRI dataset (100 patients). We used threefold cross-validation, employing the Dice similarity coefficient (DSC) and Hausdorff distance (HD) metrics for evaluation. The -test and Tukey honestly significant difference analysis were used to assess statistical differences across methods. nnU-Net achieved DSC scores of (fat) and (FGT), whereas Auto3DSeg achieved (fat) and (FGT). Significant differences in fat HD ( , ) originated from the full resolution and the 3D cascade U-Net. No evidence of significant differences was found in FGT HD or DSC metrics. Ensemble approaches of Auto3Dseg and nnU-Net demonstrated comparable performance in segmenting fat and FGT on breast MRI. The significant differences in fat HD underscore the importance of boundary-focused metrics in evaluating segmentation methods.

PurposeTo assess the reproducibility of a manual muscle MRI segmentation method that follows a specific set of recommendations developed in our center.Materials and methodsNine healthy volunteers underwent a muscle MRI examination that included a TSE T2 sequence of the thighs. Muscle segmentation was performed by three operators: an expert operator (OP1) with 3 years of experience and two radiology residents (OP2 and 3) who were both given basic segmentation instructions, whereas only OP2 underwent additional supervised training from OP1. Intra- and inter-operator Dice similarity coefficient (DSC) was calculated.ResultsOP1 showed the highest average intra-operator DSC values (0.885), whereas OP2 had higher average DSC (0.856) compared to OP3 (0.818). The highest inter-operator agreement was observed between Operators 1 and 2 (0.814) and the lowest between OP2 and OP3 (0.702). Confidence interval (CI) analysis showed that the most experienced operator also had the least variability in drawing the ROIs, whereas OP2 showed both higher intra-operator reproducibility compared to OP3 and higher inter-operator agreement with OP1. The muscles that showed the least reproducibility were the semimembranosus and the short head of the biceps femoris.DiscussionFollowing specific recommendations such as these ones derived from our single-center experience leads to an overall high reproducibility of manual muscle segmentation and is helpful in improving both intra-operator and inter-operator reproducibility in less experienced operators.