Single Label Research Articles

Whole Spine Segmentation Using Object Detection and Semantic Segmentation.

Virtual and augmented reality have enjoyed increased attention in spine surgery. Preoperative planning, pedicle screw placement, and surgical training are among the most studied use cases. Identifying osseous structures is a key aspect of navigating a 3-dimensional virtual reconstruction. To automate the otherwise time-consuming process of labeling vertebrae on each slice individually, we propose a fully automated pipeline that automates segmentation on computed tomography (CT) and which can form the basis for further virtual or augmented reality application and radiomic analysis. Based on a large public dataset of annotated vertebral CT scans, we first trained a YOLOv8m (You-Only-Look-Once algorithm, Version 8 and size medium) to detect each vertebra individually. On the then cropped images, a 2D-U-Net was developed and externally validated on 2 different public datasets. Two hundred fourteen CT scans (cervical, thoracic, or lumbar spine) were used for model training, and 40 scans were used for external validation. Vertebra recognition achieved a mAP50 (mean average precision with Jaccard threshold of 0.5) of over 0.84, and the segmentation algorithm attained a mean Dice score of 0.75 ± 0.14 at internal, 0.77 ± 0.12 and 0.82 ± 0.14 at external validation, respectively. We propose a 2-stage approach consisting of single vertebra labeling by an object detection algorithm followed by semantic segmentation. In our externally validated pilot study, we demonstrate robust performance for our object detection network in identifying individual vertebrae, as well as for our segmentation model in precisely delineating the bony structures.

Compound Fault Diagnosis of Planetary Gearbox Based on Improved LTSS-BoW Model and Capsule Network

The identification of compound fault components of a planetary gearbox is especially important for keeping the mechanical equipment working safely. However, the recognition performance of existing deep learning-based methods is limited by insufficient compound fault samples and single label classification principles. To solve the issue, a capsule neural network with an improved feature extractor, named LTSS-BoW-CapsNet, is proposed for the intelligent recognition of compound fault components. Firstly, a feature extractor is constructed to extract fault feature vectors from raw signals, which is based on local temporal self-similarity coupled with bag-of-words models (LTSS-BoW). Then, a multi-label classifier based on a capsule network (CapsNet) is designed, in which the dynamic routing algorithm and average threshold are adopted. The effectiveness of the proposed LTSS-BoW-CapsNet method is validated by processing three compound fault diagnosis tasks. The experimental results demonstrate that our method can via decoupling effectively identify the multi-fault components of different compound fault patterns. The testing accuracy is more than 97%, which is better than the other four traditional classification models.

Criminal Associations: Insights into the Concept, Types, and Challenges for Legislative Improvements to Combat Them

This article suggests improvements to the criminal law measures against organized crime. The terminology and conceptual framework of criminal groups were analyzed using various methods, such as dialectical, comparative-legal, system-structural, and formal-legal.The study exposed serious flaws in the current legislative approach to regulating criminal liability for arranging or leading criminal groups and participating in their activities. The main problem is that the same objective indicators of a criminal act are often referred to by different terms, resulting in misinterpretations of their etymological meanings in the Criminal Code of the Russian Federation. In order to simplify the terminology, it was proposed to adopt the term “criminal associations” as a single label for all criminal groups. The opinions on removing criminal associations from the institution of complicity in crime were critically reviewed.

Learning Offset Probability Distribution for Accurate Object Detection

Object detection combines object classification and object localization problems. Current object detection methods heavily depend on regression networks to locate objects, which are optimized with various regression loss functions to predict offsets between candidate boxes and objects. However, these regression losses are difficult to assign the appropriate penalties for samples with large offset errors, resulting in suboptimal regression networks and inaccurate object offsets. In this article, we consider object location as offset bin classification problem, and propose a distance-aware offset bin classification network optimized with multiple binary cross entropy losses to learn various offset probability distribution, including single label distribution and distance-aware label distribution. On one hand, it provides gradient contributions for different samples based on the bounded probability instead of previous incalculable offset error. On the other hand, it explores the distance correlations between discrete offset bins to facilitate network learning. Specifically, we discretize the continuous offset into a number of bins, and predict the probability of each offset bin, in which the probability should be higher for the offset bin closer to the target offsets, and vice versa. Furthermore, we propose an expectation-based offset prediction and a hierarchical focusing method to improve the precision of prediction. We conduct extensive experiments to evaluate the effectiveness of our method. In addition, our method can be conveniently and flexibly inserted into existing object detection methods, which consistently achieves a large gain based on popular anchor-based and anchor-free methods on the PASCAL VOC, MS-COCO, KITTI, and CrowdHuman datasets. Code will be released at: https://github.com/QiuHeqian/DBC .

Comparison of Myocardial Blood Flow Quantification Models for Double ECG Gating Arterial Spin Labeling MRI: Reproducibility Assessment.

Arterial spin labeling (ASL) allows non-invasive quantification of myocardial blood flow (MBF). Double-ECG gating (DG) ASL is more robust to heart rate variability than single-ECG gating (SG), but its reproducibility requires further investigation. Moreover, the existence of multiple quantification models hinders its application. Frequency-offset-corrected-inversion (FOCI) pulses provide sharper edge profiles than hyperbolic-secant (HS), which could benefit myocardial ASL. To assess the performance of MBF quantification models for DG compared to SG ASL, to evaluate their reproducibility and to compare the effects of HS and FOCI pulses. Prospective. Sixteen subjects (27 ± 8 years). 1.5 T/DG and SG flow-sensitive alternating inversion recovery ASL. Three models for DG MBF quantification were compared using Monte Carlo simulations and in vivo experiments. Two models used a fitting approach (one using only a single label and control image pair per fit, the other using all available image pairs), while the third model used a T1 correction approach. Slice profile simulations were conducted for HS and FOCI pulses with varying B0 and B1. Temporal signal-to-noise ratio (tSNR) was computed for different acquisition/quantification strategies and inversion pulses. The number of images that minimized MBF error was investigated in the model with highest tSNR. Intra and intersession reproducibility were assessed in 10 subjects. Within-subject coefficient of variation, analysis of variance. P-value <0.05 was considered significant. MBF was not different across acquisition/quantification strategies (P = 0.27) nor pulses (P = 0.9). DG MBF quantification models exhibited significantly higher tSNR and superior reproducibility, particularly for the fitting model using multiple images (tSNR was 3.46 ± 2.18 in vivo and 3.32 ± 1.16 in simulations, respectively; wsCV = 16%). Reducing the number of ASL pairs to 13/15 did not increase MBF error (minimum = 0.22 mL/g/min). Reproducibility of MBF was better for DG than SG acquisitions, especially when employing a fitting model. 2 TECHNICAL EFFICACY: Stage 1.

AmbiFC: Fact-Checking Ambiguous Claims with Evidence

Abstract Automated fact-checking systems verify claims against evidence to predict their veracity. In real-world scenarios, the retrieved evidence may not unambiguously support or refute the claim and yield conflicting but valid interpretations. Existing fact-checking datasets assume that the models developed with them predict a single veracity label for each claim, thus discouraging the handling of such ambiguity. To address this issue we present AmbiFC,1 a fact-checking dataset with 10k claims derived from real-world information needs. It contains fine-grained evidence annotations of 50k passages from 5k Wikipedia pages. We analyze the disagreements arising from ambiguity when comparing claims against evidence in AmbiFC, observing a strong correlation of annotator disagreement with linguistic phenomena such as underspecification and probabilistic reasoning. We develop models for predicting veracity handling this ambiguity via soft labels, and find that a pipeline that learns the label distribution for sentence-level evidence selection and veracity prediction yields the best performance. We compare models trained on different subsets of AmbiFC and show that models trained on the ambiguous instances perform better when faced with the identified linguistic phenomena.

Toward Video Anomaly Retrieval From Video Anomaly Detection: New Benchmarks and Model.

Video anomaly detection (VAD) has been paid increasing attention due to its potential applications, its current dominant tasks focus on online detecting anomalies, which can be roughly interpreted as the binary or multiple event classification. However, such a setup that builds relationships between complicated anomalous events and single labels, e.g., "vandalism", is superficial, since single labels are deficient to characterize anomalous events. In reality, users tend to search a specific video rather than a series of approximate videos. Therefore, retrieving anomalous events using detailed descriptions is practical and positive but few researches focus on this. In this context, we propose a novel task called Video Anomaly Retrieval (VAR), which aims to pragmatically retrieve relevant anomalous videos by cross-modalities, e.g., language descriptions and synchronous audios. Unlike the current video retrieval where videos are assumed to be temporally well-trimmed with short duration, VAR is devised to retrieve long untrimmed videos which may be partially relevant to the given query. To achieve this, we present two large-scale VAR benchmarks and design a model called Anomaly-Led Alignment Network (ALAN) for VAR. In ALAN, we propose an anomaly-led sampling to focus on key segments in long untrimmed videos. Then, we introduce an efficient pretext task to enhance semantic associations between video-text fine-grained representations. Besides, we leverage two complementary alignments to further match cross-modal contents. Experimental results on two benchmarks reveal the challenges of VAR task and also demonstrate the advantages of our tailored method. Captions are publicly released at https://github.com/Roc-Ng/VAR.

Suppressive effect of Yokukansan on glutamate released from canine keratinocytes.

Canine atopic dermatitis (CAD) is caused by skin barrier dysfunction due to allergen exposure. Excessive glutamate release in the skin is associated with delayed skin barrier function recovery and epidermal thickening and lichenification. Treatment with Yokukansan (YKS), a traditional Japanese medicine, reduces dermatitis severity and scratching behavior in NC/Nga mice by decreasing epidermal glutamate levels. However, the association between canine keratinocytes and glutamate and the mechanism by which YKS inhibits glutamate release from keratinocytes remains unknown. We aimed to investigate glutamate release from canine progenitor epidermal keratinocytes (CPEKs) and the inhibitory effect of YKS on this release. We also explored the underlying mechanism of YKS to enable its application in CAD treatment. Glutamate produced from CPEKs in the medium at 24 hours was measured. The measurement conditions varied in terms of cell density and YKS concentration. CPEKs were treated with a glutamate receptor antagonist (MK-801), a glutamate transporter antagonist (THA), and a glutamate dehydrogenase inhibitor (epigallocatechin gallate; EGCG), and the inhibitory effect of YKS, YKS + THA, MK-801, and EGCG on this release was determined. MK-801 and glutamate dehydrogenase inhibitor were tested alone, and THA was tested in combination with YKS. Finally, glutamine incorporated into CPEKs at 24 hours was measured using radioisotope labeling. CPEKs released glutamate in a cell density-dependent manner, inhibited by YKS in a concentration-dependent manner. Moreover, YKS reduced the intracellular uptake of radioisotope-labeled glutamine in a concentration-dependent manner. No involvement of glutamate receptor antagonism or activation of glutamate transporters was found, as suggested by previous studies. In addition, EGCG could inhibit glutamate release from CPEKs. Our findings indicated that glutamate release from CPEKs could be effectively inhibited by YKS, suggesting the utility of YKS in maintaining skin barrier function during CAD. In addition, CPEKs are appropriate for analyzing the mechanism of YKS. However, we found that the mechanism of action of YKS differs from that reported in previous studies, suggesting that it may have had a similar effect to EGCG in this study. Further research is warranted to understand the exact mechanism and clinical efficacy in treating CAD.

Open-world barely-supervised learning via augmented pseudo labels

&lt;p&gt;Open-world semi-supervised learning (OWSSL) has received significant attention since it addresses the issue of unlabeled data containing classes not present in the labeled data. Unfortunately, existing OWSSL methods still rely on a large amount of labeled data from seen classes, overlooking the reality that a substantial amount of labels is difficult to obtain in real scenarios. In this paper, we explored a new setting called open-world barely-supervised learning (OWBSL), where only a single label was provided for each seen class, greatly reducing labeling costs. To tackle the OWBSL task, we proposed a novel framework that leveraged augmented pseudo-labels generated for the unlabeled data. Specifically, we first generated initial pseudo-labels for the unlabeled data using visual-language models. Subsequently, to ensure that the pseudo-labels remained reliable while being updated during model training, we enhanced them using predictions from weak data augmentation. This way, we obtained the augmented pseudo-labels. Additionally, to fully exploit the information from unlabeled data, we incorporated consistency regularization based on strong and weak augmentations into our framework. Our experimental results on multiple benchmark datasets demonstrated the effectiveness of our method.&lt;/p&gt;

Supervision by Denoising.

Learning-based image reconstruction models, such as those based on the U-Net, require a large set of labeled images if good generalization is to be guaranteed. In some imaging domains, however, labeled data with pixel- or voxel-level label accuracy are scarce due to the cost of acquiring them. This problem is exacerbated further in domains like medical imaging, where there is no single ground truth label, resulting in large amounts of repeat variability in the labels. Therefore, training reconstruction networks to generalize better by learning from both labeled and unlabeled examples (called semi-supervised learning) is problem of practical and theoretical interest. However, traditional semi-supervised learning methods for image reconstruction often necessitate handcrafting a differentiable regularizer specific to some given imaging problem, which can be extremely time-consuming. In this work, we propose "supervision by denoising" (SUD), a framework to supervise reconstruction models using their own denoised output as labels. SUD unifies stochastic averaging and spatial denoising techniques under a spatio-temporal denoising framework and alternates denoising and model weight update steps in an optimization framework for semi-supervision. As example applications, we apply SUD to two problems from biomedical imaging-anatomical brain reconstruction (3D) and cortical parcellation (2D)-to demonstrate a significant improvement in reconstruction over supervised-only and ensembling baselines. Our code available at https://github.com/seannz/sud.

Multi-Label Chest X-Ray Image Classification with Single Positive Labels.

Deep learning approaches for multi-label Chest X-ray (CXR) images classification usually require large-scale datasets. However, acquiring such datasets with full annotations is costly, time-consuming, and prone to noisy labels. Therefore, we introduce a weakly supervised learning problem called Single Positive Multi-label Learning (SPML) into CXR images classification (abbreviated as SPML-CXR), in which only one positive label is annotated per image. A simple solution to SPML-CXR problem is to assume that all the unannotated pathological labels are negative, however, it might introduce false negative labels and decrease the model performance. To this end, we present a Multi-level Pseudo-label Consistency (MPC) framework for SPML-CXR. First, inspired by the pseudo-labeling and consistency regularization in semi-supervised learning, we construct a weak-to-strong consistency framework, where the model prediction on weakly-augmented image is treated as the pseudo label for supervising the model prediction on a strongly-augmented version of the same image, and define an Image-level Perturbation-based Consistency (IPC) regularization to recover the potential mislabeled positive labels. Besides, we incorporate Random Elastic Deformation (RED) as an additional strong augmentation to enhance the perturbation. Second, aiming to expand the perturbation space, we design a perturbation stream to the consistency framework at the feature-level and introduce a Feature-level Perturbation-based Consistency (FPC) regularization as a supplement. Third, we design a Transformer-based encoder module to explore the sample relationship within each mini-batch by a Batch-level Transformer-based Correlation (BTC) regularization. Extensive experiments on the CheXpert and MIMIC-CXR datasets have shown the effectiveness of our MPC framework for solving the SPML-CXR problem.

Heterogeneity in the projections and excitability of tyraminergic/octopaminergic neurons that innervate the Drosophila reproductive tract.

Aminergic nuclei in mammals are generally composed of relatively small numbers of cells with broad projection patterns. Despite the gross similarity of many individual neurons, recent transcriptomic, anatomic and behavioral studies suggest previously unsuspected diversity. Smaller clusters of aminergic neurons in the model organism Drosophila melanogaster provide an opportunity to explore the ramifications of neuronal diversity at the level of individual cells. A group of approximately 10 tyraminergic/octopaminergic neurons innervates the female reproductive tract in flies and has been proposed to regulate multiple activities required for fertility. The projection patterns of individual neurons within the cluster are not known and it remains unclear whether they are functionally heterogenous. Using a single cell labeling technique, we show that each region of the reproductive tract is innervated by a distinct subset of tyraminergic/octopaminergic cells. Optogenetic activation of one subset stimulates oviduct contractions, indicating that the cluster as a whole is not required for this activity, and underscoring the potential for functional diversity across individual cells. Using whole cell patch clamp, we show that two adjacent and morphologically similar cells are tonically inhibited, but each responds differently to injection of current or activation of the inhibitory GluCl receptor. GluCl appears to be expressed at relatively low levels in tyraminergic/octopaminergic neurons within the cluster, suggesting that it may regulate their excitability via indirect pathways. Together, our data indicate that specific tyraminergic/octopaminergic cells within a relatively homogenous cluster have heterogenous properties and provide a platform for further studies to determine the function of each cell.

The Sustainable Cocoa Debate: Key controversies

Abstract The sustainability of agriculture has been a topic of public debate over the last twenty years. The cocoa sector has been at the centre of this debate because of the large demand for chocolate in the Global North, which was one of the first sectors to begin to adopt sustainability standards. In this article, we focus on the actions taken by Northern and Southern actors to try to achieve cocoa sustainability over the past five years. Based on interviews and observations in Côte d’Ivoire, Ghana and European countries, complemented by document analysis, we map the controversies in the sector. Three issues emerge from our analysis: fair remuneration for producers, child labour and deforestation. We note differences of views between the actors of the North and the South, especially around the question of the fair remuneration of small producers. Emerging concerns are the single ingredient labels and the European regulations on zero imported deforestation and human rights due diligence.

Metrical Ambiguity

In most instances the meter of a Russian poem becomes clear virtually from the start, after a single line or perhaps just a few lines. However, there are also poems for which a simple metrical classification remains problematic even upon consideration of the entire work. In some cases, an abundance of internal rhyme leads to the appearance of a “shadow meter” that creates an alternative way to describe the meter over at least a portion of the poem. In others, it turns out to be possible to interpret an entire poem as belonging to any of two or more meters, often because the work does not precisely match the norms for any one type while bearing reasonably close resemblances to more than one metrical category. In this paper I examine several instances of metrical ambiguity in Russian verse and conclude that for such poems it is best not to employ a single metrical label but to offer a more detailed characterization that does justice to the work’s complexity.

Collaborative Learning for Annotation-Efficient Volumetric MR Image Segmentation.

Deep learning has presented great potential in accurate MR image segmentation when enough labeled data are provided for network optimization. However, manually annotating three-dimensional (3D) MR images is tedious and time-consuming, requiring experts with rich domain knowledge and experience. To build a deep learning method exploring sparse annotations, namely only a single two-dimensional slice label for each 3D training MR image. Retrospective. Three-dimensional MR images of 150 subjects from two publicly available datasets were included. Among them, 50 (1377 image slices) are for prostate segmentation. The other 100 (8800 image slices) are for left atrium segmentation. Five-fold cross-validation experiments were carried out utilizing the first dataset. For the second dataset, 80 subjects were used for training and 20 were used for testing. 1.5 T and 3.0 T; axial T2-weighted and late gadolinium-enhanced, 3D respiratory navigated, inversion recovery prepared gradient echo pulse sequence. A collaborative learning method by integrating the strengths of semi-supervised and self-supervised learning schemes was developed. The method was trained using labeled central slices and unlabeled noncentral slices. Segmentation performance on testing set was reported quantitatively and qualitatively. Quantitative evaluation metrics including boundary intersection-over-union (B-IoU), Dice similarity coefficient, average symmetric surface distance, and relative absolute volume difference were calculated. Paired t test was performed, and P < 0.05 was considered statistically significant. Compared to fully supervised training with only the labeled central slice, mean teacher, uncertainty-aware mean teacher, deep co-training, interpolation consistency training (ICT), and ambiguity-consensus mean teacher, the proposed method achieved a substantial improvement in segmentation accuracy, increasing the mean B-IoU significantly by more than 10.0% for prostate segmentation (proposed method B-IoU: 70.3% ± 7.6% vs. ICT B-IoU: 60.3% ± 11.2%) and by more than 6.0% for left atrium segmentation (proposed method B-IoU: 66.1% ± 6.8% vs. ICT B-IoU: 60.1% ± 7.1%). A collaborative learning method trained using sparse annotations can segment prostate and left atrium with high accuracy. 0 TECHNICAL EFFICACY: Stage 1.

CheXNet and feature pyramid network: a fusion deep learning architecture for multilabel chest X-Ray clinical diagnoses classification.

The existing multilabel X-Ray image learning tasks generally contain much information on pathology co-occurrence and interdependency, which is very important for clinical diagnosis. However, the challenging part of this subject is to accurately diagnose multiple diseases that occurred in a single X-Ray image since multiple levels of features are generated in the images, and create different features as in single label detection. Various works were developed to address this challenge with proposed deep learning architectures to improve classification performance and enrich diagnosis results with multi-probability disease detection. The objective is to create an accurate result and a faster inference system to support a quick diagnosis in the medical system. To contribute to this state-of-the-art, we designed a fusion architecture, CheXNet and Feature Pyramid Network (FPN), to classify and discriminate multiple thoracic diseases from chest X-Rays. This concept enables the model to extract while creating a pyramid of feature maps with different spatial resolutions that capture low-level and high-level semantic information to encounter multiple features. The model's effectiveness is evaluated using the NIH ChestXray14 dataset, with the Area Under Curve (AUC) and accuracy metrics used to compare the results against other cutting-edge approaches. The overall results demonstrate that our method outperforms other approaches and has become promising for multilabel disease classification in chest X-Rays, with potential applications in clinical practice. The result demonstrated that we achieved an average AUC of 0.846 and an accuracy of 0.914. Further, our proposed architecture diagnoses images in 0.013s, faster than the latest approaches.

Sparse annotation learning for dense volumetric MR image segmentation with uncertainty estimation

Objective. Training neural networks for pixel-wise or voxel-wise image segmentation is a challenging task that requires a considerable amount of training samples with highly accurate and densely delineated ground truth maps. This challenge becomes especially prominent in the medical imaging domain, where obtaining reliable annotations for training samples is a difficult, time-consuming, and expert-dependent process. Therefore, developing models that can perform well under the conditions of limited annotated training data is desirable. Approach. In this study, we propose an innovative framework called the extremely sparse annotation neural network (ESA-Net) that learns with only the single central slice label for 3D volumetric segmentation which explores both intra-slice pixel dependencies and inter-slice image correlations with uncertainty estimation. Specifically, ESA-Net consists of four specially designed distinct components: (1) an intra-slice pixel dependency-guided pseudo-label generation module that exploits uncertainty in network predictions while generating pseudo-labels for unlabeled slices with temporal ensembling; (2) an inter-slice image correlation-constrained pseudo-label propagation module which propagates labels from the labeled central slice to unlabeled slices by self-supervised registration with rotation ensembling; (3) a pseudo-label fusion module that fuses the two sets of generated pseudo-labels with voxel-wise uncertainty guidance; and (4) a final segmentation network optimization module to make final predictions with scoring-based label quantification. Main results. Extensive experimental validations have been performed on two popular yet challenging magnetic resonance image segmentation tasks and compared to five state-of-the-art methods. Significance. Results demonstrate that our proposed ESA-Net can consistently achieve better segmentation performances even under the extremely sparse annotation setting, highlighting its effectiveness in exploiting information from unlabeled data.

A comparative study of multi and single post labeling delay pseudocontinuous arterial spin labeling in patients with carotid artery stenosis

PurposeArterial Spin Labeling (ASL) allows for the non-invasive visualization of brain perfusion to detect abnormalities. In unilateral carotid artery stenosis, one hemisphere is less supplied with blood which results in a lower cerebral blood flow (CBF) compared to the healthy side. ASL can be performed time-resolved using multiple post labeling delay (PLD) times after labeling or static with a single delay, the latter allowing for a faster and more robust acquisition while bearing the risk of a falsely set delay resulting in unusable images. The purpose of this study is to compare the performance of multi-PLD and single-PLD ASL in patients with unilateral carotid artery stenosis both as means of diagnosis and therapeutic follow-up examination. MethodsASL perfusion data of 17 patients with known unilateral carotid artery stenosis was used to compare the diagnostic performance of the multi-PLD and single-PLD approach. Comparisons were made based on the CBF values and the added benefit of arrival time maps showing slower blood flow in multi-PLD ASL which might be overlooked in the individual delay images both before and after therapy. ResultsBoth the multi-PLD and the single-PLD data could identify the side of the stenosis with hemispheric differences in each approach (p < 0.001) and depict the normalization of CBF after therapy (p > 0.05). There were no differences between the individual methods (p > 0.05). ConclusionIn this work, we could show that multi-PLD ASL in patients with unilateral carotid artery stenosis is beneficial as it provides both CBF and arrival time maps, however when only a single-PLD acquisition is available, this appears sufficient in a clinical setting to investigate the presence of a unilateral stenosis.

Explore Contextual Information for 3D Scene Graph Generation.

3D scene graph generation (SGG) has been of high interest in computer vision. Although the accuracy of 3D SGG on coarse classification and single relation label has been gradually improved, the performance of existing works is still far from being perfect for fine-grained and multi-label situations. In this paper, we propose a framework fully exploring contextual information for the 3D SGG task, which attempts to satisfy the requirements of fine-grained entity class, multiple relation labels, and high accuracy simultaneously. Our proposed approach is composed of a Graph Feature Extraction module and a Graph Contextual Reasoning module, achieving appropriate information-redundancy feature extraction, structured organization, and hierarchical inferring. Our approach achieves superior or competitive performance over previous methods on the 3DSSG dataset, especially on the relationship prediction sub-task.

Results from a Phase (Ph) 1 Clinical Study of the All-Oral Regimen of CC-486 and Venetoclax for Relapsed and Refractory Acute Myeloid Leukemia

Background: Venetoclax (Ven) with azacitidine (Aza) is the standard of care for newly diagnosed AML patients who are unfit for intensive induction chemotherapy. It is well known that this treatment also has activity in the relapsed and refractory (R/R) setting. However, this regimen requires 7 consecutive daily subcutaneous or intravenous doses of Aza each month, which continues indefinitely. The relative inconvenience of this, and its negative impact on quality of life, may lead to its early discontinuation, which may decrease the efficacy or durability of this regimen. A more convenient all-oral regimen may decrease visits to an infusion center, improve quality of life, and potentially increase the efficacy of the regimen due to improved compliance. CC-486 is the oral formulation of Aza that is currently approved for post-induction chemotherapy maintenance in AML. Methods: This is a single center open label, Ph I study investigating CC-486 and Ven in R/R AML patients. In the dose escalation phase, subjects received CC-486 at one of two cohorts (200 mg PO days 1-14 and 300 mg PO days 1-14). Ven was given at the 400 mg/day PO regimen, for 28 days, after an initial intra-patient dose escalation per the standard of care. Using a 3+3 study design for these two cohorts, we aimed to determine the maximum tolerated dose (MTD) of CC-486 in combination with Ven. An expansion cohort is planned for 10 additional patients treated at the MTD. To assess whether CC-486 exerts similar effects in the leukemia stem cell population (LSCs) compared to conventional IV Aza, we retrospectively analyzed patient samples from cohort 1 with a patient sample treated with Ven/IV Aza outside of the study. We performed mass spectrometry-based metabolomics analyses in LSCs isolated from patient samples as well as single-cell RNA sequencing. Results: A total of 9 patients were accrued for this Ph I study (Figure 1A). The average age at the time of enrolment was 66, and median prior treatment regimens was 1 (range 1-3). Five patients had received Ven/Aza in prior treatments. There were no dose-limiting toxicity (DLT) events and the MTD was determined to be 300mg. The most common hematologic toxicities were Gr 3-4 neutropenia (n = 8, 89%), Gr 3-4 anemia (n = 7, 78%), and Gr 3-4 thrombocytopenia (n = 5, 56%). The most common non-hematologic toxicities were Gr 1-2 nausea (n = 5, 56%), fatigue (n = 2, 22%), and diarrhea (n = 2, 22%). Febrile neutropenia was also seen in two subjects (Gr 1-2 n = 1, 11%, Gr 3-4 n= 1, 11%). Best responses were complete remission (CR) (n = 2, 22%) partial remission (PR) (n = 1, 11%), and stable disease (SD) (n= 3, 33%). Three patients proceeded to an allogeneic stem cell transplant. Overall survival was 288 days, 95% CI: (39 days, 424 days). To determine whether CC-486/Ven exerts similar effects on the LSC population compared to conventional IV Aza/Ven, LSCs were isolated from the bone marrow of a responder and non-responder, and these were compared to LSCs from a patient who underwent treatment and responded to conventional IV Aza/Ven. Given subjects who achieved a CR in the study cleared their blasts by day 4 rendering no leukemia cells for analyses, samples used were from Subject 1 (stable disease with unchanged blasts), and Subject 2 (stable disease with significant blast reduction). Metabolomics assays performed in LSCs at day 4 and day 28 suggested mitochondrial dysfunction, decreased electron transport chain complex II activity and decreased glutathione levels in Subject 2 and conventional IV Aza/Ven but not on Subject 1. These findings are consistent with Pollyea et al. 2018 suggesting CC-486/Ven targets LSCs by perturbing complex II activity and OXPHOS in LSCs. Further, scRNA seq performed in Subject 1 and 2 showed gene expression patterns similar to those seen in patients who respond to Aza/Ven vs not, including a significant shift of BCL-2 to MCL-1 expression in the cell population harboring LSCs (Figure 1B). Conclusion: CC-486 and Ven is an all-oral regimen currently being investigated for the treatment of R/R AML. Our Ph I dose escalation demonstrated a favorable safety profile for this regimen with a MTD of 300mg CC-486. Responses were seen in both dose cohorts including 2 CRs and 1 PR, and patients have been successfully bridged to hematopoietic stem cell transplant. Analyses performed in patient samples from this study suggested a similar effect in LSCs in CC-486/Ven compared to conventional IV Aza/Ven.