Heterotrimeric Gα-subunit regulates flower and fruit development in CLAVATA signaling pathway in cucumber.

Flowers and fruits are the reproductive organs in plants and play essential roles in natural beauty and the human diet. CLAVATA (CLV) signaling has been well characterized as regulating floral organ development by modulating shoot apical meristem (SAM) size; however, the signaling molecules downstream of the CLV pathway remain largely unknown in crops. Here, we found that functional disruption of CsCLV3 peptide and its receptor CsCLV1 both resulted in flowers with extra organs and stumpy fruits in cucumber. A heterotrimeric G protein α-subunit (CsGPA1) was shown to interact with CsCLV1. Csgpa1 mutant plants derived from gene editing displayed significantly increased floral organ numbers and shorter and wider fruits, a phenotype resembling that of Csclv mutants in cucumber. Moreover, the SAM size was enlarged and the longitudinal cell size of fruit was decreased in Csgpa1 mutants. The expression of the classical stem cell regulator WUSCHEL (WUS) was elevated in the SAM, while the expression of the fruit length stimulator CRABS CLAW (CRC) was reduced in the fruit of Csgpa1 mutants. Therefore, the Gα-subunit CsGPA1 protein interacts with CsCLV1 to inhibit floral organ numbers but promote fruit elongation, via repressing CsWUS expression and activating CsCRC transcription in cucumber. Our findings identified a new player in the CLV signaling pathway during flower and fruit development in dicots, increasing the number of target genes for precise manipulation of fruit shape during crop breeding.

Electronic instability in pressured black phosphorus under strong magnetic field.

In this paper, we have systematically studied the electronic instability of pressured black phosphorous (BP) under strong magnetic field. We first present an effective model Hamiltonian for pressured BP near theLifshitzpoint. Then we show that when the magnetic field exceeds a critical value, the nodal-line semimetal (NLSM) state of BP with a small band overlap re-enters the semiconductive phase by re-opening a small gap. This results in a narrow-bandgap semiconductor with a partially flat valence band edge. Moreover, we demonstrate that above this critical magnetic field, two possible instabilities, i.e. charge density wave phase and excitonic insulator (EI) phase, are predicted as the ground state for high and low doping concentrations, respectively. By comparing our results with the experiment (Sunet al2018Sci. Bull.631539), we suggest that the field-induced instability observed experimentally corresponds to an EI. Furthermore, we propose that the semimetallic BP under pressure with small band overlaps may provide a good platform to study the magneto-exciton insulators. Our findings bring the first insight into the electronic instability of topological NLSM in the quantum limit.

Highly Stable Cesium Molybdenum Chloride Perovskite Nanocrystals for Photothermal Semihydrogenation Applications.

Metal halide perovskite materials with excellent carrier transport properties have been regarded as a new class of catalysts with great application potential. However, their development is hampered by their instability in polar solvents and high temperatures. Herein, we report a solution-processed Cs2MoCl6 perovskite nanocrystals (NCs) capped with the Mo6+, showing high thermostability in polar solvents. Furthermore, the Pd single atoms (PdSA) can be anchored on the surface of Cs2MoCl6 NCs through the unique coordination structure of Pd-Cl sites, which exhibit excellent semihydrogenation of different alkyne derivatives with high selectivity at full conversion at room temperature. Moreover, the activity could be improved greatly under Xe lamp irradiation. Detailed experimental characterization and DFT calculations indicate the improved activity under light illumination is due to the synergistic effect of photo-to-heat conversion and photoinduced electron transfer from Cs2MoCl6 to PdSA, which facilitates the activation of the C≡C group. This work not only provides a new catalyst for high selective semihydrogenation of alkyne derivatives but also opens a new avenue for metal halides as photothermal catalysts.

Research progress on the protective effect of hormones and hormone drugs in myocardial ischemia-reperfusion injury.

Ischemic heart disease (IHD) is a condition where the heart muscle does not receive enough blood flow, leading to cardiac dysfunction. Restoring blood flow to the coronary artery is an effective clinical therapy for myocardial ischemia. This strategy helps lower the size of the myocardial infarction and improves the prognosis of patients. Nevertheless, if the disrupted blood flow to the heart muscle is restored within a specific timeframe, it leads to more severe harm to the previously deprived heart tissue. This condition is referred to as myocardial ischemia/reperfusion injury (MIRI). Until now, there is a dearth of efficacious strategies to prevent and manage MIRI. Hormones are specialized substances that are produced directly into the circulation by endocrine organs or tissues in humans and animals, and they have particular effects on the body. Hormonal medications utilize human or animal hormones as their active components, encompassing sex hormones, adrenaline medications, thyroid hormone medications, and others. While several studies have examined the preventive properties of different endocrine hormones, such as estrogen and hormone analogs, on myocardial injury caused by ischemia-reperfusion, there are other hormone analogs whose mechanisms of action remain unexplained and whose safety cannot be assured. The current study is on hormones and hormone medications, elucidating the mechanism of hormone pharmaceuticals and emphasizing the cardioprotective effects of different endocrine hormones. It aims to provide guidance for the therapeutic use of drugs and offer direction for the examination of MIRI in clinical therapy.

Artificial intelligence-based classification of breast lesion from contrast enhanced mammography: a multicenter study.

PURPOSE: The authors aimed to establish an artificial intelligence (AI)-based method for preoperative diagnosis of breast lesions from contrast enhanced mammography (CEM) and to explore its biological mechanism. MATERIALS AND METHODS: This retrospective study includes 1430 eligible patients who underwent CEM examination from June 2017 to July 2022 and were divided into a construction set (n=1101), an internal test set (n=196), and a pooled external test set (n=133). The AI model adopted RefineNet as a backbone network, and an attention sub-network, named convolutional block attention module (CBAM), was built upon the backbone for adaptive feature refinement. An XGBoost classifier was used to integrate the refined deep learning features with clinical characteristics to differentiate benign and malignant breast lesions. The authors further retrained the AI model to distinguish in situ and invasive carcinoma among breast cancer candidates. RNA-sequencing data from 12 patients were used to explore the underlying biological basis of the AI prediction. RESULTS: The AI model achieved an area under the curve of 0.932 in diagnosing benign and malignant breast lesions in the pooled external test set, better than the best-performing deep learning model, radiomics model, and radiologists. Moreover, the AI model has also achieved satisfactory results (an area under the curve from 0.788 to 0.824) for the diagnosis of in situ and invasive carcinoma in the test sets. Further, the biological basis exploration revealed that the high-risk group was associated with the pathways such as extracellular matrix organization. CONCLUSIONS: The AI model based on CEM and clinical characteristics had good predictive performance in the diagnosis of breast lesions.

The AGAMOUS-LIKE 16-GENERAL REGULATORY FACTOR 1 module regulates axillary bud outgrowth via catabolism of abscisic acid in cucumber.

Lateral branches are important components of shoot architecture and directly affect crop yield and production cost. Although sporadic studies have implicated abscisic acid (ABA) biosynthesis in axillary bud outgrowth, the function of ABA catabolism and its upstream regulators in shoot branching remain elusive. Here, we showed that the MADS-box transcription factor AGAMOUS-LIKE 16 (CsAGL16) is a positive regulator of axillary bud outgrowth in cucumber (Cucumis sativus). Functional disruption of CsAGL16 led to reduced bud outgrowth, whereas overexpression of CsAGL16 resulted in enhanced branching. CsAGL16 directly binds to the promoter of the ABA 8'-hydroxylase gene CsCYP707A4 and promotes its expression. Loss of CsCYP707A4 function inhibited axillary bud outgrowth and increased ABA levels. Elevated expression of CsCYP707A4 or treatment with an ABA biosynthesis inhibitor largely rescued the Csagl16 mutant phenotype. Moreover, cucumber General Regulatory Factor 1 (CsGRF1) interacts with CsAGL16 and antagonizes CsAGL16-mediated CsCYP707A4 activation. Disruption of CsGRF1 resulted in elongated branches and decreased ABA levels in the axillary buds. The Csagl16 Csgrf1 double mutant exhibited a branching phenotype resembling that of the Csagl16 single mutant. Therefore, our data suggest that the CsAGL16-CsGRF1 module regulates axillary bud outgrowth via CsCYP707A4-mediated ABA catabolism in cucumber. Our findings provide a strategy to manipulate ABA levels in axillary buds during crop breeding to produce desirable branching phenotypes.

Learning spiking neuronal networks with artificial neural networks: neural oscillations.

First-principles-based modelings have been extremely successful in providing crucial insights and predictions for complex biological functions and phenomena. However, they can be hard to build and expensive to simulate for complex living systems. On the other hand, modern data-driven methods thrive at modeling many types of high-dimensional and noisy data. Still, the training and interpretation of these data-driven models remain challenging. Here, we combine the two types of methods to model stochastic neuronal network oscillations. Specifically, we develop a class of artificial neural networks to provide faithful surrogates to the high-dimensional, nonlinear oscillatory dynamics produced by a spiking neuronal network model. Furthermore, when the training data set is enlarged within a range of parameter choices, the artificial neural networks become generalizable to these parameters, covering cases in distinctly different dynamical regimes. In all, our work opens a new avenue for modeling complex neuronal network dynamics with artificial neural networks.

A novel framework for three-dimensional electrical impedance tomography reconstruction of maize ear via feature reconfiguration and residual networks.

Electrical impedance tomography (EIT) provides an indirect measure of the physiological state and growth of the maize ear by reconstructing the distribution of electrical impedance. However, the two-dimensional (2D) EIT within the electrode plane finds it challenging to comprehensively represent the spatial distribution of conductivity of the intact maize ear, including the husk, kernels, and cob. Therefore, an effective method for 3D conductivity reconstruction is necessary. In practical applications, fluctuations in the contact impedance of the maize ear occur, particularly with the increase in the number of grids and computational workload during the reconstruction of 3D spatial conductivity. These fluctuations may accentuate the ill-conditioning and nonlinearity of the EIT. To address these challenges, we introduce RFNetEIT, a novel computational framework specifically tailored for the absolute imaging of the three-dimensional electrical impedance of maize ear. This strategy transforms the reconstruction of 3D electrical conductivity into a regression process. Initially, a feature map is extracted from measured boundary voltage via a data reconstruction module, thereby enhancing the correlation among different dimensions. Subsequently, a nonlinear mapping model of the 3D spatial distribution of the boundary voltage and conductivity is established, utilizing the residual network. The performance of the proposed framework is assessed through numerical simulation experiments, acrylic model experiments, and maize ear experiments. Our experimental results indicate that our method yields superior reconstruction performance in terms of root-mean-square error (RMSE), correlation coefficient (CC), structural similarity index (SSIM), and inverse problem-solving time (IPST). Furthermore, the reconstruction experiments on maize ears demonstrate that the method can effectively reconstruct the 3D conductivity distribution.

NK92 cells and peripheral blood NK cells respond oppositely upon dasatinib treatment.

Natural killer (NK) cell is a valuable tool for immunotherapy in cancer treatment, both the cultured cell line NK92 and primary NK cells are widely studied and used in research and clinical trials. Clinical observations witnessed the improvement of patients' NK cells in terms of cell counts and cytotoxic activity upon dasatinib treatment, an approved drug for chronic myeloid leukaemia and Ph+ acute lymphocytic leukaemia. Several studies supported the clinical observations, yet others argued a detrimental effect of dasatinib on NK cells. Due to the complex conditions in different studies, the definite influence of dasatinib on NK92 and primary NK cells remains to be settled. Here, we used a well-defined in vitro system to evaluate the effects of dasatinib on NK92 cells and peripheral blood (PB)-NK cells. By co-culturing NK cells with dasatinib to test the cell counts and target cell-killing activities, we surprisingly found that the chemical influenced oppositely on these two types of NK cells. While dasatinib suppressed NK92 cell proliferation and cytotoxic activity, it improved PB-NK-killing tumour cells. RNA sequencing analysis further supported this finding, uncovering several proliferating and cytotoxic pathways responding invertedly between them. Our results highlighted an intrinsic difference between NK92 and PB-NK cells and may build clues to understand how dasatinib interacts with NK cells in vivo.

1-Naphthaleneacetic Acid Improved the In Vitro Cell Culturing by Inhibiting Apoptosis.

In vitro cell culturing witnessed its applications in scientific research and industrial activities. Attempts to shorten the doubling time of cultured cells have never ceased. In plants, auxin is applied to promote plant growth, the synthetic derivative 1-Naphthaleneacetic acid (NAA) is a good example. Despite the auxin's naturally occurring receptors are not present in mammalian cells, studies suggested they may affect cell culturing. Yet the effects and mechanisms are still unclear. Here, an up to 2-fold increase in the yield of in vitro cultured human cells is observed. Different types of human cell lines and primary cells are tested and found that NAA is effective in all the cells tested. The PI staining followed by FACS suggested that NAA do not affect the cell cycling. Apoptosis-specific dye staining analysis implicated that NAA rescued cell death. Further bulk RNA sequencing is done and it is identified that the lipid metabolism-engaging and anti-apoptosis gene, ANGPTL4, is enhanced in expression upon NAA treatment. Studies on ANGPTL4 knockout cells indicated that ANGPTL4 is required for NAA-mediated response. Thus, the data identified a beneficial role of NAA in human cell culturing and highlighted its potency in in vitro cell culturing.

Molecular and circuit mechanisms underlying avoidance of rapid cooling stimuli in C. elegans.

The mechanisms by which animals respond to rapid changes in temperature are largely unknown. Here, we found that polymodal ASH sensory neurons mediate rapid cooling-evoked avoidance behavior within the physiological temperature range in C. elegans. ASH employs multiple parallel circuits that consist of stimulatory circuits (AIZ, RIA, AVA) and disinhibitory circuits (AIB, RIM) to respond to rapid cooling. In the stimulatory circuit, AIZ, which is activated by ASH, releases glutamate to act on both GLR-3 and GLR-6 receptors in RIA neurons to promote reversal, and ASH also directly or indirectly stimulates AVA to promote reversal. In the disinhibitory circuit, AIB is stimulated by ASH through the GLR-1 receptor, releasing glutamate to act on AVR-14 to suppress RIM activity. RIM, an inter/motor neuron, inhibits rapid cooling-evoked reversal, and the loop activities thus equally stimulate reversal. Our findings elucidate the molecular and circuit mechanisms underlying the acute temperature stimuli-evoked avoidance behavior.

RNA binding motif 4 inhibits the replication of ebolavirus by directly targeting 3'-leader region of genomic RNA.

Ebola virus (EBOV) belongs to Filoviridae family possessing single-stranded negative-sense RNA genome, which is a serious threat to human health. Nowadays, no therapeutics have been proven to be successful in efficiently decreasing the mortality rate. RNA binding proteins (RBPs) are reported to participate in maintaining cell integrity and regulation of viral replication. However, little is known about whether and how RBPs participate in regulating the life cycle of EBOV. In our study, we found that RNA binding motif protein 4 (RBM4) inhibited the replication of EBOV in HEK293T and Huh-7 cells by suppressing viral mRNA production. Such inhibition resulted from the direct interaction between the RRM1 domain of RBM4 and the "CU" enrichment elements located in the PE1 and TSS of the 3'-leader region within the viral genome. Simultaneously, RBM4 could upregulate the expression of some cytokines involved in the host innate immune responses to synergistically exert its antiviral function. The findings therefore suggest that RBM4 might serve as a novel target of anti-EBOV strategy.

Salt-rejecting 3D cone flowing evaporator based on bilayer photothermal paper for high-performance solar seawater desalination.

Solar energy-driven water evaporation technology is a promising, low-cost and sustainable approach to alleviate the global clean water shortage, but usually suffers from low water evaporation rate and severe salt deposition on the water evaporation surface. In this work, a hydrophilic bilayer photothermal paper-based three-dimensional (3D) cone flowing evaporator was designed and prepared for stable high-performance seawater desalination with excellent salt-rejecting ability. The as-prepared bilayer photothermal paper consisted of MXene (Ti3C2Tx) and HAA (ultralong hydroxyapatite nanowires, poly(acrylic acid), and poly(acrylic acid-2-hydroxyethyl ester)). The accordion-like multilayered MXene acted as the efficient solar light absorber, and ultralong hydroxyapatite (HAP) nanowires served as the thermally insulating and supporting skeleton with a porous networked structure. A siphon effect-driven unidirectional fluid transportation unit in the 3D cone flowing evaporator could guide the concentrated saline flowing away from the evaporating surface to prevent salt deposition on the evaporation surface, avoiding severe deterioration of the performance in solar water evaporation. Furthermore, combining high solar light absorption and high photothermal conversion efficiencies, low water evaporation enthalpy (1838 ±â€¯11 J g-1), and additional energy taken from the ambient environment, the as-prepared cone flowing evaporator exhibited a high water evaporation rate of 3.22 ±â€¯0.20 kg m-2 h-1 for real seawater under one sun illumination (1 kW m-2), which was significantly higher than many values reported in the literature. This study provides an effective approach for designing high-performance solar energy-driven water evaporators for sustainable seawater desalination and wastewater purification.

Multitask Deep Learning-Based Whole-Process System for Automatic Diagnosis of Breast Lesions and Axillary Lymph Node Metastasis Discrimination from Dynamic Contrast-Enhanced-MRI: A Multicenter Study.

BACKGROUND: Accurate diagnosis of breast lesions and discrimination of axillary lymph node (ALN) metastases largely depend on radiologist experience. PURPOSE: To develop a deep learning-based whole-process system (DLWPS) for segmentation and diagnosis of breast lesions and discrimination of ALN metastasis. STUDY TYPE: Retrospective. POPULATION: 1760 breast patients, who were divided into training and validation sets (1110 patients), internal (476 patients), and external (174 patients) test sets. FIELD STRENGTH/SEQUENCE: 3.0T/dynamic contrast-enhanced (DCE)-MRI sequence. ASSESSMENT: DLWPS was developed using segmentation and classification models. The DLWPS-based segmentation model was developed by the U-Net framework, which combined the attention module and the edge feature extraction module. The average score of the output scores of three networks was used as the result of the DLWPS-based classification model. Moreover, the radiologists' diagnosis without and with the DLWPS-assistance was explored. To reveal the underlying biological basis of DLWPS, genetic analysis was performed based on RNA-sequencing data. STATISTICAL TESTS: Dice similarity coefficient (DI), area under receiver operating characteristic curve (AUC), accuracy, sensitivity, specificity, and kappa value. RESULTS: The segmentation model reached a DI of 0.828 and 0.813 in the internal and external test sets, respectively. Within the breast lesions diagnosis, the DLWPS achieved AUCs of 0.973 in internal test set and 0.936 in external test set. For ALN metastasis discrimination, the DLWPS achieved AUCs of 0.927 in internal test set and 0.917 in external test set. The agreement of radiologists improved with the DLWPS-assistance from 0.547 to 0.794, and from 0.848 to 0.892 in breast lesions diagnosis and ALN metastasis discrimination, respectively. Additionally, 10 breast cancers with ALN metastasis were associated with pathways of aerobic electron transport chain and cytoplasmic translation. DATA CONCLUSION: The performance of DLWPS indicates that it can promote radiologists in the judgment of breast lesions and ALN metastasis and nonmetastasis. LEVEL OF EVIDENCE: 4 TECHNICAL EFFICACY STAGE: 3.

The exocyst subunit CsExo70B promotes both fruit length and disease resistance via regulating receptor kinase abundance at plasma membrane in cucumber.

Plant defence against pathogens generally occurs at the expense of growth and yield. Uncoupling the inverse relationship between growth and defence is of great importance for crop breeding, while the underlying genes and regulatory mechanisms remain largely elusive. The exocytosis complex was shown to play an important role in the trafficking of receptor kinases (RKs) to the plasma membrane (PM). Here, we found a Cucumis sativus exocytosis subunit Exo70B (CsExo70B) regulates the abundance of both development and defence RKs at the PM to promote fruit elongation and disease resistance in cucumber. Knockout of CsExo70B resulted in shorter fruit and susceptibility to pathogens. Mechanistically, CsExo70B associates with the developmental RK CsERECTA, which promotes fruit longitudinal growth in cucumber, and contributes to its accumulation at the PM. On the other side, CsExo70B confers to the spectrum resistance to pathogens in cucumber via a similar regulatory module of defence RKs. Moreover, CsExo70B overexpression lines showed an increased fruit yield as well as disease resistance. Collectively, our work reveals a regulatory mechanism that CsExo70B promotes both fruit elongation and disease resistance by maintaining appropriate RK levels at the PM and thus provides a possible strategy for superior cucumber breeding with high yield and robust pathogen resistance.

Engineering Copper-Containing Nanoparticles-Loaded Silicene Nanosheets with Triple Enzyme Mimicry Activities and Photothermal Effect for Promoting Bacteria-Infected Wound Healing.

Skin wounds accompanied by bacterial infections threaten human health, and conventional antibiotic treatments are ineffective for drug-resistant bacterial infections and chronically infected wounds. The development of non-antibiotic-dependent therapeutics is highly desired but remains a challenging issue. Recently, 2D silicene nanosheets with considerable biocompatibility, biodegradability, and photothermal-conversion performance have received increasing attention in biomedical fields. Herein, copper-containing nanoparticles-loaded silicene (Cu2.8O@silicene-BSA) nanosheets with triple enzyme mimicry catalytic (peroxidase, catalase, and oxidase-like) activities and photothermal function are rationally designed and fabricated for efficient bacterial elimination, angiogenesis promotion, and accelerated wound healing. Cu2.8O@silicene-BSA nanosheets display excellent antibacterial activity through synergistic effects of reactive oxygen species generated from multiple catalytic reactions, intrinsic bactericidal activity of released Cu2+ ions, and photothermal effects, achieving high antibacterial efficiencies on methicillin-resistant Staphylococcus aureus (MRSA) of 99.1 ± 0.7% in vitro and 97.2 ± 1.6% in vivo. In addition, Cu2.8O@silicene-BSA nanosheets exhibit high biocompatibility for promoting human umbilical vein endothelial cell (HUVEC) proliferation and angiogenic differentiation. In vivo experiments reveal that Cu2.8O@silicene-BSA nanosheets with synergistic photothermal/chemodynamic therapeutics effectively accelerate MRSA-infected wound healing by eliminating bacteria, alleviating inflammation, boosting collagen deposition, and promoting angiogenesis. This research presents a promising strategy to engineer photothermal-assisted nanozyme catalysis for bacteria-invaded wound healing.

A Delta-Omicron Bivalent Subunit Vaccine Elicited Antibody Responses in Mice against Both Ancestral and Variant Strains of SARS-CoV-2.

Continued mutation of the SARS-CoV-2 genome has led to multiple waves of COVID-19 infections, and new variants have continued to emerge and dominate. The emergence of Omicron and its subvariants has substantially increased the infectivity of SARS-CoV-2. RBD genes of the wild-type SARS-CoV-2 strain and the Delta, Omicron BA.1 and Omicron BA.2 variants were used to construct plasmids and express the proteins in glycoengineered Pichia pastoris. A stable 4 L-scale yeast fermentation and purification process was established to obtain high-purity RBD proteins with a complex glycoform N-glycosyl structure that was fucose-free. The RBD glycoproteins were combined with two adjuvants, Al(OH)3 and CpG, which mitigated the typical disadvantage of low immunogenicity associated with recombinant subunit vaccines. To improve the broad-spectrum antiviral activity of the candidate vaccine, Delta RBD proteins were mixed with BA.2 RBD proteins at a ratio of 1:1 and then combined with two adjuvants-Al(OH)3 and CpG-to prepare a bivalent vaccine. The bivalent vaccine effectively induced mice to produce pseudovirus-neutralizing antibodies against SARS-CoV-2 variants, Delta, Beta, and Omicron sublineages BA.1, BA.2, BA.5. The bivalent vaccine could neutralize the authentic wild-type SARS-CoV-2 strain, Delta, BA.1.1, BA.2.2, BA2.3, and BA.2.12.1 viruses, providing a new approach for improving population immunity and delivering broad-spectrum protection under the current epidemic conditions.

Genomes of cultivated and wild Capsicum species provide insights into pepper domestication and population differentiation.

Pepper (Capsicum spp.) is one of the earliest cultivated crops and includes five domesticated species, C. annuum var. annuum, C. chinense, C. frutescens, C. baccatum var. pendulum and C. pubescens. Here, we report a pepper graph pan-genome and a genome variation map of 500 accessions from the five domesticated Capsicum species and close wild relatives. We identify highly differentiated genomic regions among the domesticated peppers that underlie their natural variations in flowering time, characteristic flavors, and unique resistances to biotic and abiotic stresses. Domestication sweeps detected in C. annuum var. annuum and C. baccatum var. pendulum are mostly different, and the common domestication traits, including fruit size, shape and pungency, are achieved mainly through the selection of distinct genomic regions between these two cultivated species. Introgressions from C. baccatum into C. chinense and C. frutescens are detected, including those providing genetic sources for various biotic and abiotic stress tolerances.

Detection and classification of breast lesions using multiple information on contrast-enhanced mammography by a multiprocess deep-learning system: A multicenter study.

Objective: Accurate detection and classification of breast lesions in early stage is crucial to timely formulate effective treatments for patients. We aim to develop a fully automatic system to detect and classify breast lesions using multiple contrast-enhanced mammography (CEM) images. Methods: In this study, a total of 1,903 females who underwent CEM examination from three hospitals were enrolled as the training set, internal testing set, pooled external testing set and prospective testing set. Here we developed a CEM-based multiprocess detection and classification system (MDCS) to perform the task of detection and classification of breast lesions. In this system, we introduced an innovative auxiliary feature fusion (AFF) algorithm that could intelligently incorporates multiple types of information from CEM images. The average free-response receiver operating characteristic score (AFROC-Score) was presented to validate system's detection performance, and the performance of classification was evaluated by area under the receiver operating characteristic curve (AUC). Furthermore, we assessed the diagnostic value of MDCS through visual analysis of disputed cases, comparing its performance and efficiency with that of radiologists and exploring whether it could augment radiologists' performance. Results: On the pooled external and prospective testing sets, MDCS always maintained a high standalone performance, with AFROC-Scores of 0.953 and 0.963 for detection task, and AUCs for classification were 0.909 [95% confidence interval (95% CI): 0.822-0.996] and 0.912 (95% CI: 0.840-0.985), respectively. It also achieved higher sensitivity than all senior radiologists and higher specificity than all junior radiologists on pooled external and prospective testing sets. Moreover, MDCS performed superior diagnostic efficiency with an average reading time of 5 seconds, compared to the radiologists' average reading time of 3.2 min. The average performance of all radiologists was also improved to varying degrees with MDCS assistance. Conclusions: MDCS demonstrated excellent performance in the detection and classification of breast lesions, and greatly enhanced the overall performance of radiologists.