A two-stage approach for automatic liver segmentation with Faster R-CNN and DeepLab

Challenges and solutions of deep learning-based automated liver segmentation: A systematic review

The liver is one of the vital organs in the body. Precise liver segmentation in medical images is essential for liver disease treatment. The deep learning-based liver segmentation process faces several challenges. This research aims to analyze the challenges of liver segmentation in prior studies and identify the modifications made to network models and other enhancements implemented by researchers to tackle each challenge. In total, 88 articles from Scopus and ScienceDirect databases published between January 2016 and January 2022 have been studied. The liver segmentation challenges are classified into five main categories, each containing some subcategories. For each challenge, the proposed technique to overcome the challenge is investigated. The provided report details the authors, publication years, dataset types, imaging technologies, and evaluation metrics of all references for comparison. Additionally, a summary table outlines the challenges and solutions.

Identification and validation of a prognostic model based on three TLS-Related genes in oral squamous cell carcinoma

BACKGROUND

The tertiary lymphoid structures (TLSs) have an immunomodulatory function and have a positive impact on the survival outcomes of patients with oral squamous cell carcinoma (OSCC). However, there is a lack of standard approaches for quantifying TLSs and prognostic models using TLS-related genes (TLSRGs). These limitations limit the widespread use of TLSs in clinical practice.

METHODS

A convolutional neural network was used to automatically detect and quantify TLSs in HE-stained whole slide images. By employing bioinformatics and diverse statistical methods, this research created a prognostic model using TCGA cohorts and explored the connection between this model and immune infiltration. The expression levels of three TLSRGs in clinical specimens were detected by immunohistochemistry. To facilitate the assessment of individual prognostic outcomes, we further constructed a nomogram based on the risk score and other clinical factors.

RESULTS

TLSs were found to be an independent predictor of both overall survival (OS) and disease-free survival in OSCC patients. A larger proportion of the TLS area represented a better prognosis. After analysis, we identified 69 differentially expressed TLSRGs and selected three pivotal TLSRGs to construct the risk score model. This model emerged as a standalone predictor for OS and exhibited close associations with CD4 + T cells, CD8 + T cells, and macrophages. Immunohistochemistry revealed high expression levels of CCR7 and CXCR5 in TLS + OSCC samples, while CD86 was highly expressed in TLS- OSCC samples. The nomogram demonstrates excellent predictive ability for overall survival in OSCC patients.

CONCLUSIONS

This is the first prognostic nomogram based on TLSRGs, that can effectively predict survival outcomes and contribute to individual treatment strategies for OSCC patients.

Semantic Segmentation of CT Liver Structures: A Systematic Review of Recent Trends and Bibliometric Analysis : Neural Network-based Methods for Liver Semantic Segmentation

The use of artificial intelligence (AI) in the segmentation of liver structures in medical images has become a popular research focus in the past half-decade. The performance of AI tools in screening for this task may vary widely and has been tested in the literature in various datasets. However, no scientometric report has provided a systematic overview of this scientific area. This article presents a systematic and bibliometric review of recent advances in neuronal network modeling approaches, mainly of deep learning, to outline the multiple research directions of the field in terms of algorithmic features. Therefore, a detailed systematic review of the most relevant publications addressing fully automatic semantic segmenting liver structures in Computed Tomography (CT) images in terms of algorithm modeling objective, performance benchmark, and model complexity is provided. The review suggests that fully automatic hybrid 2D and 3D networks are the top performers in the semantic segmentation of the liver. In the case of liver tumor and vasculature segmentation, fully automatic generative approaches perform best. However, the reported performance benchmark indicates that there is still much to be improved in segmenting such small structures in high-resolution abdominal CT scans.

Customized m-RCNN and hybrid deep classifier for liver cancer segmentation and classification

Diagnosing liver disease presents a significant medical challenge in impoverished countries, with over 30 billion individuals succumbing to it each year. Existing models for detecting liver abnormalities suffer from lower accuracy and higher constraint metrics. As a result, there is a pressing need for improved, efficient, and effective liver disease detection methods. To address the limitations of current models, this method introduces a deep liver segmentation and classification system based on a Customized Mask-Region Convolutional Neural Network (cm-RCNN). The process begins with preprocessing the input liver image, which includes Adaptive Histogram Equalization (AHE). AHE helps dehaze the input image, remove color distortion, and apply linear transformations to obtain the preprocessed image. Next, a precise region of interest is segmented from the preprocessed image using a novel deep strategy called cm-RCNN. To enhance segmentation accuracy, the architecture incorporates the ReLU activation function and the modified sigmoid activation function. Subsequently, a variety of features are extracted from the segmented image, including ResNet features, shape features (area, perimeter, approximation, and convex hull), and enhanced median binary pattern. These extracted features are then used to train a hybrid classification model, which incorporates classifiers like SqueezeNet and DeepMaxout models. The final classification outcome is determined by averaging the scores obtained from both classifiers.

UNet based on dynamic convolution decomposition and triplet attention

The robustness and generalization of medical image segmentation models are being challenged by the differences between different disease types, different image types, and different cases.Deep learning based semantic segmentation methods have been providing state-of-the-art performance in the last few years. One deep learning technique, U-Net, has become the most popular architecture in the medical imaging segmentation. Despite outstanding overall performance in segmenting medical images, it still has the problems of limited feature expression ability and inaccurate segmentation. To this end, we propose a DTA-UNet based on Dynamic Convolution Decomposition (DCD) and Triple Attention (TA). Firstly, the model with Attention U-Net as the baseline network uses DCD to replace all the conventional convolution in the encoding-decoding process to enhance its feature extraction capability. Secondly, we combine TA with Attention Gate (AG) to be used for skip connection in order to highlight lesion regions by removing redundant information in both spatial and channel dimensions. The proposed model are tested on the two public datasets and actual clinical dataset such as the public COVID-SemiSeg dataset, the ISIC 2018 dataset, and the cooperative hospital stroke segmentation dataset. Ablation experiments on the clinical stroke segmentation dataset show the effectiveness of DCD and TA with only a 0.7628 M increase in the number of parameters compared to the baseline model. The proposed DTA-UNet is further evaluated on the three datasets of different types of images to verify its universality. Extensive experimental results show superior performance on different segmentation metrics compared to eight state-of-art methods.The GitHub URL of our code is https://github.com/shuaihou1234/DTA-UNet .

Residual Deformable Split Channel and Spatial U-Net for Automated Liver and Liver Tumour Segmentation

Accurate segmentation of the liver and liver tumour (LT) is challenging due to its hazy boundaries and large shape variability. Although using U-Net for liver and LT segmentation achieves better results than manual segmentation, it loses spatial and channel features during segmentation, leading to inaccurate liver and LT segmentation. A residual deformable split depth-wise separable U-Net (RDSDSU-Net) is proposed to increase the accuracy of liver and LT segmentation. The residual deformable convolution layer (DCL) with deformable pooling (DP) is used in the encoder as an attention mechanism to adaptively extract liver and LT shape and position characteristics. Afterward, a convolutional spatial and channel features split graph network (CSCFSG-Net) is introduced in the middle processing layer to improve the expression capability of the liver and LT features by capturing spatial and channel features separately and to extract global contextual liver and LT information from spatial and channel features. Sub-pixel convolutions (SPC) are used in the decoder section to prevent the segmentation results from having a chequerboard artefact effect. Also, the residual deformable encoder features are combined with the decoder through summation to avoid increasing the number of feature maps (FM). Finally, the efficiency of the RDSDSU-Net is evaluated on the 3DIRCADb and LiTS datasets. The DICE score of the proposed RDSDSU-Net achieved 98.21% for liver segmentation and 93.25% for LT segmentation on 3DIRCADb. The experimental outcomes illustrate that the proposed RDSDSU-Net model achieved better segmentation results than the existing techniques.

A Boundary-Enhanced Liver Segmentation Network for Multi-Phase CT Images with Unsupervised Domain Adaptation

Multi-phase computed tomography (CT) images have gained significant popularity in the diagnosis of hepatic disease. There are several challenges in the liver segmentation of multi-phase CT images. (1) Annotation: due to the distinct contrast enhancements observed in different phases (i.e., each phase is considered a different domain), annotating all phase images in multi-phase CT images for liver or tumor segmentation is a task that consumes substantial time and labor resources. (2) Poor contrast: some phase images may have poor contrast, making it difficult to distinguish the liver boundary. In this paper, we propose a boundary-enhanced liver segmentation network for multi-phase CT images with unsupervised domain adaptation. The first contribution is that we propose DD-UDA, a dual discriminator-based unsupervised domain adaptation, for liver segmentation on multi-phase images without multi-phase annotations, effectively tackling the annotation problem. To improve accuracy by reducing distribution differences between the source and target domains, we perform domain adaptation at two levels by employing two discriminators, one at the feature level and the other at the output level. The second contribution is that we introduce an additional boundary-enhanced decoder to the encoder-decoder backbone segmentation network to effectively recognize the boundary region, thereby addressing the problem of poor contrast. In our study, we employ the public LiTS dataset as the source domain and our private MPCT-FLLs dataset as the target domain. The experimental findings validate the efficacy of our proposed methods, producing substantially improved results when tested on each phase of the multi-phase CT image even without the multi-phase annotations. As evaluated on the MPCT-FLLs dataset, the existing baseline (UDA) method achieved IoU scores of 0.785, 0.796, and 0.772 for the PV, ART, and NC phases, respectively, while our proposed approach exhibited superior performance, surpassing both the baseline and other state-of-the-art methods. Notably, our method achieved remarkable IoU scores of 0.823, 0.811, and 0.800 for the PV, ART, and NC phases, respectively, emphasizing its effectiveness in achieving accurate image segmentation.

A Coarse-to-Fine Fusion Network for Small Liver Tumor Detection and Segmentation: A Real-World Study

Liver tumor semantic segmentation is a crucial task in medical image analysis that requires multiple MRI modalities. This paper proposes a novel coarse-to-fine fusion segmentation approach to detect and segment small liver tumors of various sizes. To enhance the segmentation accuracy of small liver tumors, the method incorporates a detection module and a CSR (convolution-SE-residual) module, which includes a convolution block, an SE (squeeze and excitation) module, and a residual module for fine segmentation. The proposed method demonstrates superior performance compared to conventional single-stage end-to-end networks. A private liver MRI dataset comprising 218 patients with a total of 3605 tumors, including 3273 tumors smaller than 3.0 cm, were collected for the proposed method. There are five types of liver tumors identified in this dataset: hepatocellular carcinoma (HCC); metastases of the liver; cholangiocarcinoma (ICC); hepatic cyst; and liver hemangioma. The results indicate that the proposed method outperforms the single segmentation networks 3D UNet and nnU-Net as well as the fusion networks of 3D UNet and nnU-Net with nnDetection. The proposed architecture was evaluated on a test set of 44 images, with an average Dice similarity coefficient (DSC) and recall of 86.9% and 86.7%, respectively, which is a 1% improvement compared to the comparison method. More importantly, compared to existing methods, our proposed approach demonstrates state-of-the-art performance in segmenting small objects with sizes smaller than 10 mm, achieving a Dice score of 85.3% and a malignancy detection rate of 87.5%.

YOLO-Based Deep Learning Model for Pressure Ulcer Detection and Classification

Pressure ulcers are significant healthcare concerns affecting millions of people worldwide, particularly those with limited mobility. Early detection and classification of pressure ulcers are crucial in preventing their progression and reducing associated morbidity and mortality. In this work, we present a novel approach that uses YOLOv5, an advanced and robust object detection model, to detect and classify pressure ulcers into four stages and non-pressure ulcers. We also utilize data augmentation techniques to expand our dataset and strengthen the resilience of our model. Our approach shows promising results, achieving an overall mean average precision of 76.9% and class-specific mAP50 values ranging from 66% to 99.5%. Compared to previous studies that primarily utilize CNN-based algorithms, our approach provides a more efficient and accurate solution for the detection and classification of pressure ulcers. The successful implementation of our approach has the potential to improve the early detection and treatment of pressure ulcers, resulting in better patient outcomes and reduced healthcare costs.

Automatic segmentation of the tumor in nonsmall-cell lung cancer by combining coarse and fine segmentation

OBJECTIVES

Radiotherapy plays an important role in the treatment of nonsmall-cell lung cancer (NSCLC). Accurate delineation of tumor is the key to successful radiotherapy. Compared with the commonly used manual delineation ways, which are time-consuming and laborious, the automatic segmentation methods based on deep learning can greatly improve the treatment efficiency.

METHODS

In this paper, we introduce an automatic segmentation method by combining coarse and fine segmentations for NSCLC. Coarse segmentation network is the first level, identifing the rough region of the tumor. In this network, according to the tissue structure distribution of the thoracic cavity where tumor is located, we designed a competition method between tumors and organs at risk (OARs), which can increase the proportion of the identified tumor covering the ground truth and reduce false identification. Fine segmentation network is the second level, carrying out precise segmentation on the results of the coarse level. These two networks are independent of each other during training. When they are used, morphological processing of small scale corrosion and large scale expansion is used for the coarse segmentation results, and the outcomes are sent to the fine segmentation part as input, so as to achieve the complementary advantages of the two networks.

RESULTS

In the experiment, CT images of 200 patients with NSCLC are used to train the network, and CT images of 60 patients are used to test. Finally, our method produced the Dice similarity coefficient of 0.78 ± 0.10.

CONCLUSIONS

The experimental results show that the proposed method can accurately segment the tumor with NSCLC, and can also provide support for clinical diagnosis and treatment.

Comparing the Clinical Viability of Automated Fundus Image Segmentation Methods

Recent methods for automatic blood vessel segmentation from fundus images have been commonly implemented as convolutional neural networks. While these networks report high values for objective metrics, the clinical viability of recovered segmentation masks remains unexplored. In this paper, we perform a pilot study to assess the clinical viability of automatically generated segmentation masks in the diagnosis of diseases affecting retinal vascularization. Five ophthalmologists with clinical experience were asked to participate in the study. The results demonstrate low classification accuracy, inferring that generated segmentation masks cannot be used as a standalone resource in general clinical practice. The results also hint at possible clinical infeasibility in experimental design. In the follow-up experiment, we evaluate the clinical quality of masks by having ophthalmologists rank generation methods. The ranking is established with high intra-observer consistency, indicating better subjective performance for a subset of tested networks. The study also demonstrates that objective metrics are not correlated with subjective metrics in retinal segmentation tasks for the methods involved, suggesting that objective metrics commonly used in scientific papers to measure the method's performance are not plausible criteria for choosing clinically robust solutions.

Endoscopy Artefact Detection by Deep Transfer Learning of Baseline Models

To visualise the tumours inside the body on a screen, a long and thin tube is inserted with a light source and a camera at the tip to obtain video frames inside organs in endoscopy. However, multiple artefacts exist in these video frames that cause difficulty during the diagnosis of cancers. In this research, deep learning was applied to detect eight kinds of artefacts: specularity, bubbles, saturation, contrast, blood, instrument, blur, and imaging artefacts. Based on transfer learning with pre-trained parameters and fine-tuning, two state-of-the-art methods were applied for detection: faster region-based convolutional neural networks (Faster R-CNN) and EfficientDet. Experiments were implemented on the grand challenge dataset, Endoscopy Artefact Detection and Segmentation (EAD2020). To validate our approach in this study, we used phase I of 2,200 frames and phase II of 331 frames in the original training dataset with ground-truth annotations as training and testing dataset, respectively. Among the tested methods, EfficientDet-D2 achieves a score of 0.2008 (mAPd[Formula: see text]0.6+mIoUd[Formula: see text]0.4) on the dataset that is better than three other baselines: Faster-RCNN, YOLOv3, and RetinaNet, and competitive to the best non-baseline result scored 0.25123 on the leaderboard although our testing was on phase II of 331 frames instead of the original 200 testing frames. Without extra improvement techniques beyond basic neural networks such as test-time augmentation, we showed that a simple baseline could achieve state-of-the-art performance in detecting artefacts in endoscopy. In conclusion, we proposed the combination of EfficientDet-D2 with suitable data augmentation and pre-trained parameters during fine-tuning training to detect the artefacts in endoscopy.

Image Segmentation via Multiscale Perceptual Grouping

The human eyes observe an image through perceptual units surrounded by symmetrical or asymmetrical object contours at a proper scale, which enables them to quickly extract the foreground of the image. Inspired by this characteristic, a model combined with multiscale perceptual grouping and unit-based segmentation is proposed in this paper. In the multiscale perceptual grouping part, a novel total variation regularization is proposed to smooth the image into different scales, which removes the inhomogeneity and preserves the edges. To simulate perceptual units surrounded by contours, the watershed method is utilized to cluster pixels into groups. The scale of smoothness is determined by the number of perceptual units. In the segmentation part, perceptual units are regarded as the basic element instead of discrete pixels in the graph cut. The appearance models of the foreground and background are constructed by combining the perceptual units. According to the relationship between perceptual units and the appearance model, the foreground can be segmented through a minimum-cut/maximum-flow algorithm. The experiment conducted on the CMU-Cornell iCoseg database shows that the proposed model has a promising performance.

Chain Graph Explanation of Neural Network Based on Feature-Level Class Confusion

Despite increasing interest in developing interpretable machine learning methods, most recent studies have provided explanations only for single instances, require additional datasets, and are sensitive to hyperparameters. This paper proposes a confusion graph that reveals model weaknesses by constructing a confusion dictionary. Unlike other methods, which focus on the performance variation caused by single-neuron suppression, it defines the role of each neuron in two different perspectives: ‘correction’ and ‘violation’. Furthermore, our method can identify the class relationships in similar positions at the feature level, which can suggest improvements to the model. Finally, the proposed graph construction is model-agnostic and does not require additional data or tedious hyperparameter tuning. Experimental results show that the information loss from omitting the channels guided by the proposed graph can result in huge performance degradation, from 91% to 33%, while the proposed graph only retains 1% of total neurons.

Detection of Aerobics Action Based on Convolutional Neural Network

To further improve the accuracy of aerobics action detection, a method of aerobics action detection based on improving multiscale characteristics is proposed. In this method, based on faster R-CNN and aiming at the problems existing in faster R-CNN, the feature pyramid network (FPN) is used to extract aerobics action image features. So, the low-level semantic information in the images can be extracted, and it can be converted into high-resolution deep-level semantic information. Finally, the target detector is constructed by the above-extracted anchor points so as to realize the detection of aerobics action. The results show that the loss function of the neural network is reduced to 0.2 by using the proposed method, and the accuracy of the proposed method can reach 96.5% compared with other methods, which proves the feasibility of this study.

Efficient knowledge distillation for liver CT segmentation using growing assistant network

Segmentation has been widely used in diagnosis, lesion detection, and surgery planning. Although deep learning (DL)-based segmentation methods currently outperform traditional methods, most DL-based segmentation models are computationally expensive and memory inefficient, which are not suitable for the intervention of liver surgery. To address this issue, a simple solution is to make a segmentation model very small for the fast inference time, however, there is a trade-off between the model size and performance. In this paper, we propose a DL-based real-time 3-D liver CT segmentation method, where knowledge distillation (KD) method, known as knowledge transfer from teacher to student models, is incorporated to compress the model while preserving the performance. Because it is well known that the knowledge transfer is inefficient when the disparity of teacher and student model sizes is large, we propose a growing teacher assistant network (GTAN) to gradually learn the knowledge without extra computational cost, which can efficiently transfer knowledge even with the large gap of teacher and student model sizes. In our results, dice similarity coefficient of the student model with KD improved 1.2% (85.9% to 87.1%) compared to the student model without KD, which is a similar performance of the teacher model using only 8% (100k) parameters. Furthermore, with a student model of 2% (30k) parameters, the proposed model using the GTAN improved the dice coefficient about 2% compared to the student model without KD, and the inference time is 13 ms per a 3-D image. Therefore, the proposed method has a great potential for intervention in liver surgery as well as in many real-time applications.

Effects of Multiple Filters on Liver Tumor Segmentation From CT Images

Segmentation of liver tumors from Computerized Tomography (CT) images remains a challenge due to the natural variation in tumor shape and structure as well as the noise in CT images. A key assumption is that the performance of liver tumor segmentation depends on the characteristics of multiple features extracted from multiple filters. In this paper, we design an enhanced approach based on a two-class (liver, tumor) convolutional neural network that discriminates tumor as well as liver from CT images. First, the contrast and intensity values in CT images are adjusted and high frequencies are removed using Hounsfield units (HU) filtering and standardization. Then, the liver tumor is segmented from entire images with multiple filter U-net (MFU-net). Finally, a quantitative analysis is carried out to evaluate the segmentation results using three different methods: boundary-distance-based metrics, size-based metrics, and overlap-based metrics. The proposed method is validated on CT images from the 3Dircadb and LiTS dataset. The results demonstrate that the multiple filters are useful for extracting local and global feature simultaneously, minimizing the boundary distance errors, and our approach demonstrates better performance in heterogeneous tumor regions of CT images.

The measurement of Cobb angle based on spine X-ray images using multi-scale convolutional neural network

Adolescent idiopathic scoliosis (AIS) is a structural spinal deformity mainly in the coronal plane and is among the most frequent deformities in children, adolescents, and young adults, with an overall prevalence of 0.47-5.2%. The Cobb angle is an objective measure to determine the progression of deformity and plays a critical role in the planning of surgical treatment. However, existing studies suggested that Cobb angle measurement is susceptible to inter- and intra-observer variability, as well as a high variability in the definition of the end vertebra. In this study, we proposed an automatic method for the spine vertebrae segmentation using Deeplab V3+, a powerful tool that has shown success in the image segmentation of other anatomical regions but spine, and Cobb angle measurement. The segmentation performance was compared to existing mainstay neural networks. Compared to U-Net, Residual U-Net and Dilated U-Net, our method using Deeplab V3+ showed the best performance in the Dice Similarity Coefficient (DSC), accuracy, sensitivity and Jaccard Index. An excellent correlation in the final Cobb angle calculation was achieved between the smallest distance point (SDP) method and two experts (> 0.95), with a small error in the angle estimation compared (MAE < 3°). The proposed method could provide a potential tool for the automatic estimation of the Cobb angle to improve the efficiency and accuracy of the treatment workflow for AIS.

Deep learning system for lymph node quantification and metastatic cancer identification from whole-slide pathology images

BACKGROUND

Traditional diagnosis methods for lymph node metastases are labor-intensive and time-consuming. As a result, diagnostic systems based on deep learning (DL) algorithms have become a hot topic. However, current research lacks testing with sufficient data to verify performance. The aim of this study was to develop and test a deep learning system capable of identifying lymph node metastases.

METHODS

921 whole-slide images of lymph nodes were divided into two cohorts: training and testing. For lymph node quantification, we combined Faster RCNN and DeepLab as a cascade DL algorithm to detect regions of interest. For metastatic cancer identification, we fused Xception and DenseNet-121 models and extracted features. Prospective testing to verify the performance of the diagnostic system was performed using 327 unlabeled images. We further validated the proposed system using Positive Predictive Value (PPV) and Negative Predictive Value (NPV) criteria.

RESULTS

We developed a DL-based system capable of automated quantification and identification of metastatic lymph nodes. The accuracy of lymph node quantification was shown to be 97.13%. The PPV of the combined Xception and DenseNet-121 model was 93.53%, and the NPV was 97.99%. Our experimental results show that the differentiation level of metastatic cancer affects the recognition performance.

CONCLUSIONS

The diagnostic system we established reached a high level of efficiency and accuracy of lymph node diagnosis. This system could potentially be implemented into clinical workflow to assist pathologists in making a preliminary screening for lymph node metastases in gastric cancer patients.

Semantic Segmentation Framework for Glomeruli Detection and Classification in Kidney Histological Sections

The evaluation of kidney biopsies performed by expert pathologists is a crucial process for assessing if a kidney is eligible for transplantation. In this evaluation process, an important step consists of the quantification of global glomerulosclerosis, which is the ratio between sclerotic glomeruli and the overall number of glomeruli. Since there is a shortage of organs available for transplantation, a quick and accurate assessment of global glomerulosclerosis is essential for retaining the largest number of eligible kidneys. In the present paper, the authors introduce a Computer-Aided Diagnosis (CAD) system to assess global glomerulosclerosis. The proposed tool is based on Convolutional Neural Networks (CNNs). In particular, the authors considered approaches based on Semantic Segmentation networks, such as SegNet and DeepLab v3+. The dataset has been provided by the Department of Emergency and Organ Transplantations (DETO) of Bari University Hospital, and it is composed of 26 kidney biopsies coming from 19 donors. The dataset contains 2344 non-sclerotic glomeruli and 428 sclerotic glomeruli. The proposed model consents to achieve promising results in the task of automatically detecting and classifying glomeruli, thus easing the burden of pathologists. We get high performance both at pixel-level, achieving mean F-score higher than 0.81, and Weighted Intersection over Union (IoU) higher than 0.97 for both SegNet and Deeplab v3+ approaches, and at object detection level, achieving 0.924 as best F-score for non-sclerotic glomeruli and 0.730 as best F-score for sclerotic glomeruli.