Segmentation Quality Research Articles

Advanced Myocardial MRI Tissue Characterization Combining Contrast Agent‐Free T1‐Rho Mapping With Fully Automated Analysis

BackgroundMyocardial T1‐rho (T1ρ) mapping is a promising method for identifying and quantifying myocardial injuries without contrast agents, but its clinical use is hindered by the lack of dedicated analysis tools.PurposeTo explore the feasibility of clinically integrated artificial intelligence‐driven analysis for efficient and automated myocardial T1ρ mapping.Study TypeRetrospective.PopulationFive hundred seventy‐three patients divided into a training (N = 500) and a test set (N = 73) including ischemic and nonischemic cases.Field Strength/SequenceSingle‐shot bSSFP T1ρ mapping sequence at 1.5 T.AssessmentThe automated process included: left ventricular (LV) wall segmentation, right ventricular insertion point detection and creation of a 16‐segment model for segmental T1ρ value analysis. Two radiologists (20 and 7 years of MRI experience) provided ground truth annotations. Interobserver variability and segmentation quality were assessed using the Dice coefficient with manual segmentation as reference standard. Global and segmental T1ρ values were compared. Processing times were measured.Statistical TestsIntraclass correlation coefficients (ICCs) and Bland–Altman analysis (bias ±2SD); Paired Student's t‐tests and one‐way ANOVA. A P value <0.05 was considered significant.ResultsThe automated approach significantly reduced processing time (3 seconds vs. 1 minute 51 seconds ± 22 seconds). In the test set, automated LV wall segmentation closely matched manual results (Dice 81.9% ± 9.0) and closely aligned with interobserver segmentation (Dice 82.2% ± 6.5). Excellent ICCs were achieved on a patient basis (0.94 [95% CI: 0.91 to 0.96]) with bias of −0.93 cm2 ± 6.60. There was no significant difference in global T1ρ values between manual (54.9 msec ± 4.6; 95% CI: 53.8 to 56.0 msec, range: 46.6–70.9 msec) and automated processing (55.4 msec ± 5.1; 95% CI: 54.2 to 56.6 msec; range: 46.4–75.1 msec; P = 0.099). The pipeline demonstrated a high level of agreement with manual‐derived T1ρ values at the patient level (ICC = 0.85; bias +0.52 msec ± 5.18). No significant differences in myocardial T1ρ values were found between methods across the 16 segments (P = 0.75).Data ConclusionAutomated myocardial T1ρ mapping shows promise for the rapid and noninvasive assessment of heart disease.Evidence Level3Technical EfficacyStage 1

A semantic segmentation-based algorithm for fast flotation bubble size distribution measurement

The visual characteristics of froths on flotation cell surfaces are crucial for the optimal control of the flotation process. However, previous research has primarily focused on optimizing markers to prevent the over-segmentation of froth images, often overlooking computational efficiency. This work proposes a deep learning-based froth segmentation algorithm that prioritizes both segmentation speed and quality. We propose a dataset construction strategy based on pseudo-labeling, which utilizes traditional methods to extract marker maps from froth images to build the dataset for deep learning. Subsequently, a semantic segmentation model is trained to derive marker maps from raw froth images. These marker maps are then employed in a marker-based watershed algorithm, enabling end-to-end segmentation of froth images. The proposed method is validated using froth images from a gold and nickel flotation plant, demonstrating the ability to accurately and efficiently identify different bubbles in real froth images without requiring complex pre-processing or post-processing steps. Industrial experiments show a tenfold increase in segmentation efficiency with virtually no loss of accuracy.

SEGMENTATION OF PARCEL BOUNDARY INDICATIONS IN VERY HIGH-RESOLUTION ORTHOPHOTO MOSAICS FOR CONTROL POINT IDENTIFICATION

Several Land Parcel Geometry issues in Indonesia's Land Registration Process, such as parcel overlapping, gaps between parcels, and incorrect parcel shapes and sizes, are currently being addressed through a block adjustment approach. One crucial aspect of the block adjustment process is determining control points that tie the parcel geometry to the land coordinate system. Detailed Observations and measurements of parcel points in the field and aerial photographs established these control or tie points. Rectifying land parcels requires many control points, requiring substantial time and effort. The automation phase is critical to expedite the control point identification process. This research uses artificial intelligence techniques to identify control points in very high-resolution orthophoto mosaics. The method employed for control point identification involves the Segment Anything Model (SAM) algorithm to segment parcel boundary indications accurately. Enhance the quality of segmentation results conducted by fine-tuning, followed by centerline extraction and refinement of the extracted data. Based on the segmentation, a SAM model capable of accurately segmenting building objects is attained, After the centerline extraction process and modifications to the existing geometric operations within the GIS Tool, at the edges of buildings, fences, and walls derived points. These points can serve as control point indications in the block adjustment process.

HST-MRF: Heterogeneous Swin Transformer With Multi-Receptive Field for Medical Image Segmentation.

The Transformer has been successfully used in medical image segmentation due to its excellent long-range modeling capabilities. However, patch segmentation is necessary when building a Transformer class model. This process ignores the tissue structure features within patch, resulting in the loss of shallow representation information. In this study, we propose a Heterogeneous Swin Transformer with Multi-Receptive Field (HST-MRF) model that fuses patch information from different receptive fields to solve the problem of loss of feature information caused by patch segmentation. The heterogeneous Swin Transformer (HST) is the core module, which achieves the interaction of multi-receptive field patch information through heterogeneous attention and passes it to the next stage for progressive learning, thus complementing the patch structure information. We also designed a two-stage fusion module, multimodal bilinear pooling (MBP), to assist HST in further fusing multi-receptive field information and combining low-level and high-level semantic information for accurate localization of lesion regions. In addition, we developed adaptive patch embedding (APE) and soft channel attention (SCA) modules to retain more valuable information when acquiring patch embedding and filtering channel features, respectively, thereby improving model segmentation quality. We evaluated HST-MRF on multiple datasets for polyp, skin lesion and breast ultrasound segmentation tasks. Experimental results show that our proposed method outperforms state-of-the-art models and can achieve superior performance. Furthermore, we verified the effectiveness of each module and the benefits of multi-receptive field segmentation in reducing the loss of structural information through ablation experiments and qualitative analysis.

Classification of brain tumor types through MRIs using parallel CNNs and firefly optimization

Image segmentation is a critical and challenging endeavor in the field of medicine. A magnetic resonance imaging (MRI) scan is a helpful method for locating any abnormal brain tissue these days. It is a difficult undertaking for radiologists to diagnose and classify the tumor from several pictures. This work develops an intelligent method for accurately identifying brain tumors. This research investigates the identification of brain tumor types from MRI data using convolutional neural networks and optimization strategies. Two novel approaches are presented: the first is a novel segmentation technique based on firefly optimization (FFO) that assesses segmentation quality based on many parameters, and the other is a combination of two types of convolutional neural networks to categorize tumor traits and identify the kind of tumor. These upgrades are intended to raise the general efficacy of the MRI scan technique and increase identification accuracy. Using MRI scans from BBRATS2018, the testing is carried out, and the suggested approach has shown improved performance with an average accuracy of 98.6%.

GLCM-Based Feature Extraction for Alpha Matting on Natural Images

The main objective of this research is to determine the optimal threshold value in the unknown region in the alpha-matting operation of natural images. Alpha-mating serves to draw matte from the image used in segmentation. The alpha value is very influential on the quality of segmentation which is determined by the level of threshold value accuracy. The determination of the threshold begins by breaking the grayscale image into several sub-images using Region of Interest (RoI). Each sub-image was extracted using the Gray Level Co-occurrence Matrix (GLCM) considered by the parameters of contrast, energy, and entropy at angles of 0°, 45°, 90°, and 135 °. Each feature results in extractions, which are then averaged and normalized in each sub-image. The value is determined as the local threshold value used in the alpha matting operation. Experiments were carried out on 12 natural images from the image-mating dataset to evaluate the performance of the proposed algorithm. The increase in accuracy shows up to 63% by the measurements of experiments, compared to the calculation of adaptive threshold by using the fuzzy CMs Algorithm.

Modified snake optimizer based multi-level thresholding for color image segmentation of agricultural diseases

Image segmentation is crucial for early identification and diagnosis of crop diseases in agriculture. It helps identify disease areas and characteristics, laying the groundwork for disease diagnosis and treatment. However, color image thresholding segmentation method faces challenges in finding the optimal threshold as the number of thresholds increases, which affects segmentation quality. In this paper, a modified snake optimizer (MSO) is proposed to address the color image thresholding segmentation problem using Kapur’s entropy as the objective function. MSO incorporates a dynamic adaptive parameter adjustment method. The improved global position update formula introduces guidance from the optimal individual and disturbance from random individuals, effectively achieving a balance between global and local search capabilities. A dynamic parameter adjustment method is added to accelerate convergence speed during snake movement to food. Lévy flight is introduced to the Flight mode to help the algorithm escape local extremums. A balancing strategy is implemented in the Mating mode, incorporating guidance from the optimal individual and information exchange between sub-populations, enabling balanced exploration and local exploitation capabilities. The hill-climbing jump operation for the optimal individual is added to help the algorithm overcome local stagnation. The proposed MSO is evaluated on CEC 2017 test functions and color test images, and the obtained results are compared with other segmentation methods and other intelligent optimization algorithms. The results demonstrate that MSO exhibits stable and excellent performance in both complex global optimization problems and color image thresholding segmentation problems. Finally, MSO is applied to segment rice disease images, and the results reveal significantly better segmentation performance compared to other algorithms. MSO shows great potential in solving the thresholding segmentation problem for agricultural disease images, thereby assisting in the accurate identification, diagnosis, and treatment of agricultural diseases and insect pests.

DeepSEM-Net: Enhancing SEM defect analysis in semiconductor manufacturing with a dual-branch CNN-Transformer architecture

Due to the increasing complexity of modern semiconductor integrated circuit fabrication processes, various defects may occur at each process step, leading to a semiconductor integrated circuit yield loss as well as reliability problems. This study introduces DeepSEM-Net, a novel dual-branch architecture that harmoniously integrates Convolutional Neural Networks (CNN) with Transformers. This innovative approach significantly enhances defect classification and precise segmentation in Scanning Electron Microscope (SEM) defect images, crucial for semiconductor manufacturing processes. DeepSEM-Net consists of a classification branch, which synergizes CNN and Transformers for robust global feature extraction, and a segmentation branch, which incorporates a Pyramid Attention module to meticulously recalibrate and exploit the feature map, thereby enriching the segmentation quality. A semi-supervised defect analysis system based on this model markedly reduces manual inspection efforts and is adept at operating in both simultaneous and sequential classification-segmentation modes. Rigorously tested on a diverse dataset from a real 12-inch wafer fab, DeepSEM-Net not only demonstrated a commendable classification accuracy of 97.25% on a 5-class dataset and a segmentation IoU of 84.40% but also exhibited remarkable resilience against the introduction of new defect classes in an unbalanced dataset. The dual-branch strategy of DeepSEM-Net, thus, significantly elevates the efficiency and reliability of SEM defect image analysis, offering a substantial leap forward in addressing yield loss and ensuring the reliability of semiconductor products.

Clinical Characteristics and Segmental Qualities of Young Patients with Heart Failure in a Tertiary Care Centre

Background: The prevalence and hospitalizations of patients with heart failure aged <55 years have increased. The study aims to evaluate the clinical traits and segmental qualities of young patients experiencing heart failure at a tertiary care center. Methods: This cross-sectional study design enrolled 83 individuals who had recently developed heart failure at a young age. The study encompassed individuals aged 18 to 55 years, from whom clinical indicators and potential risk elements were obtained followed by 2D transthoracic echocardiography performed as part of the routine assessment. Results: The study comprised of 83. The participants with a mean age of 51.5 ± 4.0 years. A significant proportion of the individuals were noted to have diabetes (65%), while 74.7% were identified with hypertension. A statistically significant association was noted between patients who had a prior history of Ischemic heart disease (IHD) and their probability of developing Heart failure (HF) (p-value = 0.02). The primary and the most prevalent cause of heart failure among these study participants was dilated cardiomyopathy. Patients with reduced Ejection fraction (EF) had a notably higher occurrence (68%) compared to mid-range EF (14%) and preserved EF (23%). Conclusion: The study comprised of involved individuals those belonging to the age group between 45-55 years, exhibited higher rates of hypertension and diabetes. Heart Failure with reduced Ejection Fraction (HFrEF) was dominant, wherein dilated cardiomyopathy invariably being the most common cause of heart failure in younger patients. Notably, a significant portion of these individuals had a past history of IHD, which contributed a substantial role in the development of HF

Application of simultaneous uncertainty quantification and segmentation for oropharyngeal cancer use-case with Bayesian deep learning

BackgroundRadiotherapy is a core treatment modality for oropharyngeal cancer (OPC), where the primary gross tumor volume (GTVp) is manually segmented with high interobserver variability. This calls for reliable and trustworthy automated tools in clinician workflow. Therefore, accurate uncertainty quantification and its downstream utilization is critical.MethodsHere we propose uncertainty-aware deep learning for OPC GTVp segmentation, and illustrate the utility of uncertainty in multiple applications. We examine two Bayesian deep learning (BDL) models and eight uncertainty measures, and utilize a large multi-institute dataset of 292 PET/CT scans to systematically analyze our approach.ResultsWe show that our uncertainty-based approach accurately predicts the quality of the deep learning segmentation in 86.6% of cases, identifies low performance cases for semi-automated correction, and visualizes regions of the scans where the segmentations likely fail.ConclusionsOur BDL-based analysis provides a first-step towards more widespread implementation of uncertainty quantification in OPC GTVp segmentation.

Binning of biological image based on unsupervised semantic segmentation and no-reference image quality assessment

Binning of biological image based on unsupervised semantic segmentation and no-reference image quality assessment

Mandibular bone segmentation from CT scans: Quantitative and qualitative comparison among software

ObjectivesNowadays, a wide variety of software for 3D reconstruction from CT scans is available; they differ for costs, capabilities, a priori knowledge, and, it is not trivial to identify the most suitable one for specific purposes. The article is aimed to provide some more information, having set up various metrics for the evaluation of different software’s performance. MethodsMetrics include software usability, segmentation quality, geometric accuracy, mesh properties and Dice Similarity Coefficient (DSC). Five different software have been considered (Mimics, D2P, Blue Sky Plan, Relu, and 3D Slicer) and tested on four cases; the mandibular bone was used as a benchmark. ResultsRelu software, being based on AI, was able to solve some very intricate geometry and proved to have a very good usability. On the other side, the time required for segmentation was significantly higher than other software (reaching over twice the time required by Mimics). Geometric distances between nodes position calculated by different software usually kept below 2.5 mm, reaching 3.1 mm in some very critical area; 75th percentile q75 is generally less than 0.5 mm, with a maximum of 1.11 mm. Dealing with consistency among software, the maximum DSC value was observed between Mimics and Slicer, D2P and Mimics, and D2P and Slicer, reaching 0.96. SignificanceThis work has demonstrated how mandible segmentation performance among software was generally very good. Nonetheless, differences in geometric accuracy, usability, costs and times required can be significant so that information here provided can be useful to perform an informed choice.

Uncertainty estimation using a 3D probabilistic U-Net for segmentation with small radiotherapy clinical trial datasets

Background and objectivesBio-medical image segmentation models typically attempt to predict one segmentation that resembles a ground-truth structure as closely as possible. However, as medical images are not perfect representations of anatomy, obtaining this ground truth is not possible. A surrogate commonly used is to have multiple expert observers define the same structure for a dataset. When multiple observers define the same structure on the same image there can be significant differences depending on the structure, image quality/modality and the region being defined. It is often desirable to estimate this type of aleatoric uncertainty in a segmentation model to help understand the region in which the true structure is likely to be positioned. Furthermore, obtaining these datasets is resource intensive so training such models using limited data may be required. With a small dataset size, differing patient anatomy is likely not well represented causing epistemic uncertainty which should also be estimated so it can be determined for which cases the model is effective or not. MethodsWe use a 3D probabilistic U-Net to train a model from which several segmentations can be sampled to estimate the range of uncertainty seen between multiple observers. To ensure that regions where observers disagree most are emphasised in model training, we expand the Generalised Evidence Lower Bound (ELBO) with a Constrained Optimisation (GECO) loss function with an additional contour loss term to give attention to this region. Ensemble and Monte-Carlo dropout (MCDO) uncertainty quantification methods are used during inference to estimate model confidence on an unseen case. We apply our methodology to two radiotherapy clinical trial datasets, a gastric cancer trial (TOPGEAR, TROG 08.08) and a post-prostatectomy prostate cancer trial (RAVES, TROG 08.03). Each dataset contains only 10 cases each for model development to segment the clinical target volume (CTV) which was defined by multiple observers on each case. An additional 50 cases are available as a hold-out dataset for each trial which had only one observer define the CTV structure on each case. Up to 50 samples were generated using the probabilistic model for each case in the hold-out dataset. To assess performance, each manually defined structure was matched to the closest matching sampled segmentation based on commonly used metrics. ResultsThe TOPGEAR CTV model achieved a Dice Similarity Coefficient (DSC) and Surface DSC (sDSC) of 0.7 and 0.43 respectively with the RAVES model achieving 0.75 and 0.71 respectively. Segmentation quality across cases in the hold-out datasets was variable however both the ensemble and MCDO uncertainty estimation approaches were able to accurately estimate model confidence with a p-value < 0.001 for both TOPGEAR and RAVES when comparing the DSC using the Pearson correlation coefficient. ConclusionsWe demonstrated that training auto-segmentation models which can estimate aleatoric and epistemic uncertainty using limited datasets is possible. Having the model estimate prediction confidence is important to understand for which unseen cases a model is likely to be useful.

A novel CNN-based image segmentation pipeline for individualized feline spinal cord stimulation modeling

Objective. Spinal cord stimulation (SCS) is a well-established treatment for managing certain chronic pain conditions. More recently, it has also garnered attention as a means of modulating neural activity to restore lost autonomic or sensory-motor function. Personalized modeling and treatment planning are critical aspects of safe and effective SCS (Rowald and Amft 2022 Front. Neurorobotics 16 983072, Wagner et al 2018 Nature 563 65–71). However, the generation of spine models at the required level of detail and accuracy requires time and labor intensive manual image segmentation by human experts. This study aims to develop a maximally automated segmentation routine capable of producing high-quality anatomical models, even with limited data, to facilitate safe and effective personalized SCS treatment planning. Approach. We developed an automated image segmentation and model generation pipeline based on a novel convolutional neural network (CNN) architecture trained on feline spinal cord magnetic resonance imaging data. The pipeline includes steps for image preprocessing, data augmentation, transfer learning, and cleanup. To assess the relative importance of each step in the pipeline and our choice of CNN architecture, we systematically dropped steps or substituted architectures, quantifying the downstream effects in terms of tissue segmentation quality (Jaccard index and Hausdorff distance) and predicted nerve recruitment (estimated axonal depolarization). Main results. The leave-one-out analysis demonstrated that each pipeline step contributed a small but measurable increment to mean segmentation quality. Surprisingly, minor differences in segmentation accuracy translated to significant deviations (ranging between 4% and 13% for each pipeline step) in predicted nerve recruitment, highlighting the importance of careful workflow design. Additionally, transfer learning techniques enhanced segmentation metric consistency and allowed generalization to a completely different spine region with minimal additional training data. Significance. To our knowledge, this work is the first to assess the downstream impacts of segmentation quality differences on neurostimulation predictions. It highlights the role of each step in the pipeline and paves the way towards fully automated, personalized SCS treatment planning in clinical settings.

Improving assessment of lesions in longitudinal CT scans: a bi-institutional reader study on an AI-assisted registration and volumetric segmentation workflow

PurposeAI-assisted techniques for lesion registration and segmentation have the potential to make CT-based tumor follow-up assessment faster and less reader-dependent. However, empirical evidence on the advantages of AI-assisted volumetric segmentation for lymph node and soft tissue metastases in follow-up CT scans is lacking. The aim of this study was to assess the efficiency, quality, and inter-reader variability of an AI-assisted workflow for volumetric segmentation of lymph node and soft tissue metastases in follow-up CT scans. Three hypotheses were tested: (H1) Assessment time for follow-up lesion segmentation is reduced using an AI-assisted workflow. (H2) The quality of the AI-assisted segmentation is non-inferior to the quality of fully manual segmentation. (H3) The inter-reader variability of the resulting segmentations is reduced with AI assistance.Materials and methodsThe study retrospectively analyzed 126 lymph nodes and 135 soft tissue metastases from 55 patients with stage IV melanoma. Three radiologists from two institutions performed both AI-assisted and manual segmentation, and the results were statistically analyzed and compared to a manual segmentation reference standard.ResultsAI-assisted segmentation reduced user interaction time significantly by 33% (222 s vs. 336 s), achieved similar Dice scores (0.80–0.84 vs. 0.81–0.82) and decreased inter-reader variability (median Dice 0.85–1.0 vs. 0.80–0.82; ICC 0.84 vs. 0.80), compared to manual segmentation.ConclusionThe findings of this study support the use of AI-assisted registration and volumetric segmentation for lymph node and soft tissue metastases in follow-up CT scans. The AI-assisted workflow achieved significant time savings, similar segmentation quality, and reduced inter-reader variability compared to manual segmentation.

Inter-scanner super-resolution of 3D cine MRI using a transfer-learning network for MRgRT

Objective. Deep-learning networks for super-resolution (SR) reconstruction enhance the spatial-resolution of 3D magnetic resonance imaging (MRI) for MR-guided radiotherapy (MRgRT). However, variations between MRI scanners and patients impact the quality of SR for real-time 3D low-resolution (LR) cine MRI. In this study, we present a personalized super-resolution (psSR) network that incorporates transfer-learning to overcome the challenges in inter-scanner SR of 3D cine MRI. Approach: Development of the proposed psSR network comprises two-stages: (1) a cohort-specific SR (csSR) network using clinical patient datasets, and (2) a psSR network using transfer-learning to target datasets. The csSR network was developed by training on breath-hold and respiratory-gated high-resolution (HR) 3D MRIs and their k-space down-sampled LR MRIs from 53 thoracoabdominal patients scanned at 1.5 T. The psSR network was developed through transfer-learning to retrain the csSR network using a single breath-hold HR MRI and a corresponding 3D cine MRI from 5 healthy volunteers scanned at 0.55 T. Image quality was evaluated using the peak-signal-noise-ratio (PSNR) and the structure-similarity-index-measure (SSIM). The clinical feasibility was assessed by liver contouring on the psSR MRI using an auto-segmentation network and quantified using the dice-similarity-coefficient (DSC). Results. Mean PSNR and SSIM values of psSR MRIs were increased by 57.2% (13.8–21.7) and 94.7% (0.38–0.74) compared to cine MRIs, with the reference 0.55 T breath-hold HR MRI. In the contour evaluation, DSC was increased by 15% (0.79–0.91). Average time consumed for transfer-learning was 90 s, psSR was 4.51 ms per volume, and auto-segmentation was 210 ms, respectively. Significance. The proposed psSR reconstruction substantially increased image and segmentation quality of cine MRI in an average of 215 ms across the scanners and patients with less than 2 min of prerequisite transfer-learning. This approach would be effective in overcoming cohort- and scanner-dependency of deep-learning for MRgRT.

Fundus image segmentation based on random collision whale optimization algorithm

Medical image segmentation is an important technical tool, OTSU algorithm is a common method in threshold segmentation, but with the increase of the number of threshold segmentation, the selection of its threshold is a big problem, and the segmentation effect is difficult to be guaranteed. In order to solve this problem, this paper proposes a random collision whale optimization algorithm to optimize OTSU for reliable image segmentation. The algorithm is called RCWOA for short. Firstly, the Halton sequence is used to uniformly initialize the population to make the population position distribution uniform, and then the dimensional Opposition-based learning of small-hole imaging is introduced to update the whale position and find out the missing feasible solution. Finally, the random collision theory is used to update the position of the optimal individual to improve the quality of the solution, At the same time, it also improves the search ability of the algorithm. In 12 test functions, RCWOA was compared with 6 other algorithms, demonstrating the feasibility and novelty of RCWOA. In 8 experiments of fundus image segmentation, RCWOA was compared with 9 other algorithms. The results showed that RCWOA had a Friedman test composite ranking of 1.3516, ranking at the forefront, and exhibited significantly improved segmentation quality.

#439 Nephron number determination in whole slide mice kidney specimens using a deep learning assisted disector/fractionator approach

Abstract Background and Aims Nephron number highly varies between different species and within species, ranging in humans from 200 000 to 2 000 000 nephrons per kidney. Low nephron numbers can promote early onset of end-stage kidney disease. Nephron assessment is highly needed for better diagnosis and more targeted treatment. However, its assessment in vivo is not established in clinical practice yet. Pioneer studies suggest nephron number estimation in vivo through a combination of glomerular densities in kidney biopsies together with medical imaging. In contrast, post-mortem nephron assessment either using optical clearing or by exhaustive kidney sectioning is well studied but remains laborious. Here, we propose a new semi-automated nephron counting approach using serial PAS-stained whole slide kidney sections, deep learning segmentation and slide registration. Method Mice were sacrificed, kidneys harvested, measured, halved in the coronary plane and then opposingly embedded in paraffin. Subsequently, the entire kidney was sectioned with 10 µm thickness per slide. Tissue was stained with PAS and slides were digitalized using Aperio GT 450 DX scanner (Leica, Wetzlar, Germany) (Fig. a). Deep learning segmentation of glomeruli was performed in the first and every 10th (reference) section and subsequent (look-up) section using Tensorflow. Obtained segmentation results were manually quality controlled and adjusted, if necessary, in QuPath (Fig. b). Final segmentations of the look-up slides were registered to the reference slides in ImageJ using manual landmarks with the ImageJ BigWarp plugin (Fig. c). Glomeruli being present either in the look-up slide, or in the reference slide were counted as disector particles (Q-). Kidney areas on every examined slide were measured. Total number of glomeruli and total kidney area were determined as previously described (Fig. d). Results A kidney from a 12-week-old mouse suffering chronic kidney disease (mouse line Alb-creERT2 (ki/ki); Glut9 (fl/fl)) was used. Kidney weight was 162 mg, length 11 mm, width 6 mm, depth 4 mm and kidney volume 260 mm³. In total the kidney was sectioned in 95 slices. 20 reference sections and its respective look-up sections were stained and digitalized. In total 5981 glomeruli were segmented in 40 kidney slices using the deep learning algorithm. Glomerular segmentation quality was verified by two independent observers (mean dice score 0.79). Inaccuracies, if any, where manually corrected. For image registration in average 18.1 landmarks per slide were created within the ImageJ BigWarp plugin. Registration took in average 5 minutes per slide. Glomeruli without match (disector particles, Q-) were automatically counted. Number of disector particles (Q-) ranged from 32 to 244 per slide. Total nephron number was 11.085 and kidney area 210 mm³. The differences to the manual measured volume (260 mm³) might be explained by discard of small kidney fractions for the used sectioning technique. Conclusion We report, to our knowledge, the first deep learning assisted disector/fractionator approach to determine nephron number in whole slide kidney specimens. This method requires no physical disector setup, is scalable, cheap, time efficient, and unbiased. The method was so far validated in a mouse model of chronic kidney disease. The method will further be validated on different animal models. Additionally, occurrence of atubular glomeruli will be considered in nephron number determination. Furthermore, the integration of GFR measurements might allow the calculation of mean single nephron GFR and might together with automated glomerular area measurements unravel pathophysiologic processes in kidney ageing and disease.

Study on the Fracture of a Shield Segment in a Fully Excavated Hard Rock Section under the Influence of Construction Loads

In this paper, the initiation of the fracture of a segment caused by the pressure of the jack and other factors during shield construction is discussed. Based on the Rots model in the finite element software Diana 10.4, the fracture width is solved. Combined with in situ measurements, the mechanisms of concrete fracturing of a segment under external loads, such as the jack thrust deflection angle and uneven jack thrust caused by the changes in the segment due to the upward buoyancy and shield attitude, are studied; additionally, the occurrence conditions and engineering control measures for segment fracture are summarized. The results show that when the attitudes of the shield and segment are identical, the total thrust of the shield is recommended not to exceed 21,000 kN, and is strictly limited to 24,000 kN. When the attitude inclination angle between the shield machine and the segment is less than 1°, the impact on the segment quality is small. When the inclination angle reaches 2°, the total thrust of the shield is recommended not to exceed 16,000 kN, and is strictly limited to 18,000 kN. When the inclination reaches 3°, a fracture is easily produced. When the total thrust is 19,000 kN, it is recommended that the loading increase or decrease in the left and right four grippers should not exceed 20%, and they are prohibited to exceed 30%. The fracture width increases exponentially with the increase in misalignment between adjacent segment rings. These research results provide a theoretical basis for jack pressure control during shield construction.

Unsupervised Image Segmentation Parameters Evaluation for Urban Land Use/Land Cover Applications

Image segmentation plays an important role in object-based classification. An optimal image segmentation should result in objects being internally homogeneous and, at the same time, distinct from one another. Strategies that assess the quality of image segmentation through intra- and inter-segment homogeneity metrics cannot always predict possible under- and over-segmentations of the image. Although the segmentation scale parameter determines the size of the image segments, it cannot synchronously guarantee that the produced image segments are internally homogeneous and spatially distinct from their neighbors. The majority of image segmentation assessment methods largely rely on a spatial autocorrelation measure that makes the global objective function fluctuate irregularly, resulting in the image variance increasing drastically toward the end of the segmentation. This paper relied on a series of image segmentations to test a more stable image variance measure based on the standard deviation model as well as a more robust hybrid spatial autocorrelation measure based on the current Moran’s index and the spatial autocorrelation coefficient models. The results show that there is a positive and inversely proportional correlation between the inter-segment heterogeneity and the intra-segment homogeneity since the global heterogeneity measure increases with a decrease in the image variance measure. It was also found that medium-scale parameters produced better quality image segments when used with small color weights, while large-scale parameters produced good quality segments when used with large color factor weights. Moreover, with optimal segmentation parameters, the image autocorrelation measure stabilizes and follows a near horizontal fluctuation while the image variance drops to values very close to zero, preventing the heterogeneity function from fluctuating irregularly towards the end of the image segmentation process.