Weakly supervised temporal action localization with actionness-guided false positive suppression.

Weakly supervised temporal action localization aims to locate the temporal boundaries of action instances in untrimmed videos using video-level labels and assign them the corresponding action category. Generally, it is solved by a pipeline called "localization-by-classification", which finds the action instances by classifying video snippets. However, since this approach optimizes the video-level classification objective, the generated activation sequences often suffer interference from class-related scenes, resulting in a large number of false positives in the prediction results. Many existing works treat background as an independent category, forcing models to learn to distinguish background snippets. However, under weakly supervised conditions, the background information is fuzzy and uncertain, making this method extremely difficult. To alleviate the impact of false positives, we propose a new actionness-guided false positive suppression framework. Our method seeks to suppress false positive backgrounds without introducing the background category. Firstly, we propose a self-training actionness branch to learn class-agnostic actionness, which can minimize the interference of class-related scene information by ignoring the video labels. Secondly, we propose a false positive suppression module to mine false positive snippets and suppress them. Finally, we introduce the foreground enhancement module, which guides the model to learn the foreground with the help of the attention mechanism as well as class-agnostic actionness. We conduct extensive experiments on three benchmarks (THUMOS14, ActivityNet1.2, and ActivityNet1.3). The results demonstrate the effectiveness of our method in suppressing false positives and it achieves the state-of-the-art performance. Code: https://github.com/lizhilin-ustc/AFPS.

Construction and Preclinical Evaluation of 124I/125I-Labeled Antibody Targeting T Cell Immunoglobulin and Mucin Domain-3.

T cell immunoglobulin and mucin domain-3 (TIM3; HAVCR2) is a transmembrane protein that exerts negative regulatory control over T cell responses. Studies have demonstrated an upregulation of TIM3 expression in tumor-infiltrating lymphocytes (TILs) in cancer patients. In this investigation, a series of monoclonal antibodies targeting TIM3 were produced by hybridoma technology. Among them, C23 exhibited favorable biological properties. To enable specific binding, we developed a 124I/125I-C23 radio-tracer via N-bromosuccinimide (NBS)-mediated labeling of the monoclonal antibody C23. Binding affinity and specificity were assessed using the 293T-TIM3 cell line, which overexpresses TIM3, and the parent 293T cells. Furthermore, biodistribution and in vivo imaging of 124I/125I-C23 were examined in HEK293TIM3 xenograft models and allograft models of 4T1 (mouse breast cancer cells) and CT26 (mouse colon cancer cells). Micro-PET/CT imaging was conducted at intervals of 4, 24, 48, 72, and/or 96 h post intravenous administration of 3.7-7.4 MBq 124I-C23 in the respective model mice. Additionally, immunohistochemistry (IHC) staining of TIM3 expression in dissected tumor organs was performed, along with an assessment of the corresponding expression of Programmed Death 1 (PD1), CD3, and CD8 in the tumors. The C23 monoclonal antibody (mAb) specifically binds to TIM3 protein with a dissociation constant of 23.28 nM. The 124I-C23 and 125I-C23 radio-tracer were successfully prepared with a labeling yield of 83.59 ± 0.35% and 92.35 ± 0.20%, respectively, and over 95.00% radiochemical purity. Stability results indicated that the radiochemical purity of 124I/125I-C23 in phosphate-buffered saline (PBS) and 5% human serum albumin (HSA) was still >80% after 96 h. 125I-C23 uptake in 293T-TIM3 cells was 2.80 ± 0.12%, which was significantly higher than that in 293T cells (1.08 ± 0.08%), and 125I-C23 uptake by 293T-TIM3 cells was significantly blocked at 60 and 120 min in the blocking groups. Pharmacokinetics analysis in vivo revealed an elimination time of 14.62 h and a distribution time of 0.4672 h for 125I-C23. Micro-PET/CT imaging showed that the 124I-C23 probe uptake in the 293T-TIM3 model significantly differed from that of the negative control group and blocking group. In the humanized mouse model, the 124I-C23 probe had obvious specific uptake in the 4T1 and CT26 models and maximum uptake at 24 h in tumor tissues (SUVmax (the maximum standardized uptake value) in 4T1 and CT26 humanized TIM3 murine tumor models: 0.59 ± 0.01 and 0.76 ± 0.02, respectively). Immunohistochemistry of tumor tissues from these mouse models showed comparable TIM3 expression. CD3 and CD8 cells and PD-1 expression were also observed in TIM3-expressing tumor tissues. The TIM3-targeting antibody C23 showed good affinity and specificity. The 124I/125I-C23 probe has obvious targeting specificity for TIM3 in vitro and in vivo. Our results suggest that 124I/125I-C23 is a promising tracer for TIM3 imaging and may have great potential in monitoring immune checkpoint drug efficacy.

Exploration of radionuclide labeling of a novel scFv-Fc fusion protein targeting CLDN18.2 for tumor diagnosis and treatment.

PURPOSE: Claudin 18.2 (CLDN18.2), due to its highly selective expression in tumor cells, has made breakthrough progress in clinical research and is expected to be integrated into routine tumor diagnosis and treatment. METHODS: In this research, we obtained an scFv-Fc fusion protein (SF106) targeting CLDN18.2 through hybridoma technology. The scFv-Fc fusion protein was labeled with radioactive isotopes (124I and 177Lu) to generate the radio-probes. The targeting and specificity of the radio-probes were tested in cellular models, and its diagnostic and therapeutic potential was further evaluated in tumor-bearing models. RESULTS: The molecular probes [124I]I-SF106 and [177Lu]Lu-DOTA-SF106 possess high radiochemical purity (RCP, 98.18 ± 0.93 % and 97.05 ± 1.1 %) and exhibit good stability in phosphate buffer saline and 5 % human serum albumin (92.44 ± 4.68 % and 91.03 ± 2.42 % at 120 h). [124I]I-SF106 uptake in cells expressing CLDN18.2 was well targeted and specific, and the dissociation constant was 17.74 nM [124I]I-SF106 micro-PET imaging showed that the maximum standardized uptake value (SUVmax) was significantly higher than CLDN18.2-negative tumors (1.83 ± 0.02 vs. 1.23 ± 0.04, p < 0.001). The maximum uptake was attained in tumors expressing CLDN18.2 at 48 h after injection. [124I]I-SF106 and [177Lu]Lu-DOTA-SF106 dosimetric study showed that the effective dose in humans complies with the medical safety standards required for their clinical application. The results of treatment experiments showed that 3 MBq of [177Lu]Lu-DOTA-SF106 in CLDN18.2-expressing tumor-bearing mice could significantly inhibit tumor growth. CONCLUSION: These results indicate that radionuclide-labeled scFv-Fc molecular probes ([124I]I-SF106 and [177Lu]Lu-DOTA-SF106) provide a new possibility for the diagnosis and treatment of CLDN18.2-positive cancer patients in clinical practice.

Screening, Construction, and Preliminary Evaluation of CLDN18.2-Specific Peptides for Noninvasive Molecular Imaging.

Recent global clinical trials have shown that CLDN18.2 is an ideal target for the treatment of gastric cancer and that patients with high CLDN18.2 expression can benefit from targeted therapy. Therefore, accurate and comprehensive detection of CLDN18.2 expression is important for patient screening and guidance in anti-CLDN18.2 therapy. Phage display technology was used to screen CLDN18.2-specific peptides from 100 billion libraries. 293TCLDN18.1 cells were used to exclude nonspecific binding and CLDN18.1 binding sequences, while 293TCLDN18.2 cells were used to screen CLDN18.2-specific binding peptides. The monoclonal clones obtained from phage screening were sequenced, and peptides were synthesized based on the sequencing results. Binding specificity and affinity were assessed with a fluorescein isothiocyanate (FITC)-conjugated peptide. A 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA)-conjugated peptide was also synthesized for 68Ga radiolabeling. The in vitro and in vivo stability, partition coefficients, in vivo molecular imaging, and biodistribution were also characterized. Overall, 54 monoclonal clones were selected after phage display screening. Subsequently, based on the cell ELISA results, CLDN18.2 preference monoclonal clones were selected for deoxyribonucleic acid (DNA) sequencing, and four 7-peptide sequences were obtained after sequence comparison; among them, a peptide named T37 was further validated in vitro and in vivo. The T37 peptide specifically recognized CLDN18.2 but not CLDN18.1 and bound strongly to CLDN18.2-positive cell membranes. The 68Ga-DOTA-T37 probe exhibits good in vitro properties and high stability as a hydrophilic probe; it has high biological safety, and positron emission tomography/computed tomography (PET/CT) studies have shown that it can specifically target CLDN18.2 protein and CLDN18.2-positive tumors in mice. 68Ga-DOTA-T37 demonstrated the superiority and feasibility of using a CLDN18.2-specific probe in PCT/CT imaging, which deserves further development and exploitation.

Molecular PET/CT mapping of rhACE2 distribution and quantification in organs to aid in SARS-CoV-2 targeted therapy.

Coronavirus disease 2019 (COVID-19), which is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, poses a significant threat to public health. Angiotensin-converting enzyme 2 (ACE2) is a key receptor for SARS-CoV-2 infection. Recombinant human ACE2 (RhACE2), as a soluble supplement for human ACE2, can competitively block SARS-CoV-2 infection. In this study, a mouse organ in situ rhACE2 high aggregation model was constructed for the first time, and in vivo real-time positron emission tomography (PET) imaging of rhACE2 in the mouse model was performed using an ACE2-specific agent 68 Ga-HZ20. This radiotracer exhibits reliable radiochemical properties in vitro and maintains a high affinity for rhACE2 in vivo. In terms of probe uptake, 68 Ga-HZ20 showed a good target-to-nontarget ratio and was rapidly cleared from the circulatory system and excreted by the kidneys and urinary system. PET imaging with this radiotracer can noninvasively and accurately monitor the content and distribution of rhACE2 in the body, which clarifies that rhACE2 can aggregate in multiple organs, suggesting the preventive and therapeutic potential of rhACE2 for SARS-CoV-2 and COVID-19.

Preclinical evaluation and pilot clinical study of [68Ga]Ga-THP-APN09, a novel PD-L1 targeted nanobody radiotracer for rapid one-step radiolabeling and PET imaging.

PURPOSE: Programmed cell death protein-1/ligand-1 (PD-1/L1) blockade has been a breakthrough in the treatment of patients with non-small cell lung cancer (NSCLC), but there is still a lack of effective methods to screen patients. Here we report a novel 68 Ga-labeled nanobody [68 Ga]Ga-THP-APN09 for PET imaging of PD-L1 status in mouse models and a first-in-human study in NSCLC patients. METHODS: [68 Ga]Ga-THP-APN09 was prepared by site-specific radiolabeling, with no further purification. Cell uptake assays were completed in the human lung adenocarcinoma cell line A549, NSCLC cell line H1975 and human PD-L1 gene-transfected A549 cells (A549PD-L1). The imaging to image PD-L1 status and biodistribution were investigated in tumor-bearing mice of these three tumor cell types. The first-in-human clinical translational trial was registered as NCT05156515. The safety, radiation dosimetry, biodistribution, and correlations of tracer uptake with immunohistochemical staining and major pathologic response (MPR) were evaluated in NSCLC patients who underwent adjuvant immunotherapy combined with chemotherapy. RESULTS: Radiosynthesis of [68 Ga]Ga-THP-APN09 was achieved at room temperature and a pH of 6.0-6.5 in 10 min with a high radiochemical yield (> 99%) and 13.9-27.8 GBq/µmol molar activity. The results of the cell uptake study reflected variable levels of surface PD-L1 expression observed by flow cytometry in the order A549PD-L1 > H1975 > A549. In small-animal PET/CT imaging, H1975 and A549PD-L1 tumors were clearly visualized in an 8.3:1 and 2.2:1 ratios over PD-L1-negative A549 tumors. Ex vivo biodistribution studies showed that tumor uptake was consistent with the PET results, with the highest A549PD-L1 being taken up the most (8.20 ± 0.87%ID/g), followed by H1975 (3.69 ± 0.50%ID/g) and A549 (0.90 ± 0.16%ID/g). Nine resectable NSCLC patients were enrolled in the clinical study. Uptake of [68 Ga]Ga-THP-APN09 was mainly observed in the kidneys and spleen, followed by low uptake in bone marrow. The radiation dose is within a reliable range. Tumor uptake was positively correlated with PD-L1 expression TPS (rs = 0.8763, P = 0.019). Tumor uptake of [68 Ga]Ga-THP-APN09 (SUVmax) in MPR patients was higher than that in non-MPR patients (median SUVmax 2.73 vs. 2.10, P = 0.036, determined with Mann-Whitney U-test). CONCLUSION: [68 Ga]Ga-THP-APN09 has the potential to be transformed into a kit-based radiotracer for rapid, simple, one-step, room temperature radiolabeling. The tracer can detect PD-L1 expression levels in tumors, and it may make it possibility to predict the response of PD-1 immunotherapy combined with chemotherapy. Confirmation in a large number of cases is needed. TRIAL REGISTRATION: Clinical Trial (NCT05156515). Registered 12 December 2021.

Development of a CLDN18.2-targeting immuno-PET probe for non-invasive imaging in gastrointestinal tumors.

Claudin18.2 (CLDN18.2) is a tight junction protein that is overexpressed in a variety of solid tumors such as gastrointestinal cancer and oesophageal cancer. It has been identified as a promising target and a potential biomarker to diagnose tumor, evaluate efficacy, and determine patient prognosis. TST001 is a recombinant humanized CLDN18.2 antibody that selectively binds to the extracellular loop of human Claudin18.2. In this study, we constructed a solid target radionuclide zirconium-89 (89Zr) labled-TST001 to detect the expression of in the human stomach cancer BGC823CLDN18.2 cell lines. The [89Zr]Zr-desferrioxamine (DFO)-TST001 showed high radiochemical purity (RCP, >99%) and specific activity (24.15 ± 1.34 GBq/µmol), and was stable in 5% human serum albumin, and phosphate buffer saline (>85% RCP at 96 h). The EC50 values of TST001 and DFO-TST001 were as high as 0.413 ± 0.055 and 0.361 ± 0.058 nM (P > 0.05), respectively. The radiotracer had a significantly higher average standard uptake values in CLDN18.2-positive tumors than in CLDN18.2-negative tumors (1.11 ± 0.02 vs. 0.49 ± 0.03, P = 0.0016) 2 days post injection (p.i.). BGC823CLDN18.2 mice models showed high tumor/muscle ratios 96 h p.i. with [89Zr]Zr-DFO-TST001 was much higher than those of the other imaging groups. Immunohistochemistry results showed that BGC823CLDN18.2 tumors were highly positive (+++) for CLDN18.2, while those in the BGC823 group did not express CLDN18.2 (-). The results of ex vivo biodistribution studies showed that there was a higher distribution in the BGC823CLDN18.2 tumor bearing mice (2.05 ± 0.16 %ID/g) than BGC823 mice (0.69 ± 0.02 %ID/g) and blocking group (0.72 ± 0.02 %ID/g). A dosimetry estimation study showed that the effective dose of [89Zr]Zr-DFO-TST001 was 0.0705 mSv/MBq, which is within the range of acceptable doses for nuclear medicine research. Taken together, these results suggest that Good Manufacturing Practices produced by this immuno-positron emission tomography probe can detect CLDN18.2-overexpressing tumors.

A High-Quality Dopant-Free Electron-Selective Passivating Contact Made from Ultra-Low Concentration Water Solution.

Crystalline silicon solar cells produced by doping processes have intrinsic shortages of high Auger recombination and/or severe parasitic optical absorption. Dopant-free carrier-selective contacts (DF-CSCs) are alternative routines for the next generation of highly efficient solar cells. However, it is difficult to achieve both good passivating and low contact resistivity for most DF-CSCs. In this paper, a high-quality dopant-free electron-selective passivating contact made from ultra-low concentration water solution is reported. Both low recombination current (J0) ~10 fA/cm2 and low contact resistivity (ρc) ~31 mΩ·cm2 are demonstrated with this novel contact on intrinsic amorphous silicon thin film passivated n-Si. The electron selectivity is attributed to relieving of the interfacial Fermi level pinning because of dielectric properties (decaying of the metal-induced gap states (MIGS)). The full-area implementation of the novel passivating contact shows 20.4% efficiency on a prototype solar cell without an advanced lithography process. Our findings offer a very simple, cost-effective, and efficient solution for future semiconductor devices, including photovoltaics and thin-film transistors.

Learning Lightweight Dynamic Kernels With Attention Inside via Local-Global Context Fusion.

Traditional convolutional neural networks (CNNs) share their kernels among all positions of the input, which may constrain the representation ability in feature extraction. Dynamic convolution proposes to generate different kernels for different inputs to improve the model capacity. However, the total parameters of the dynamic network can be significantly huge. In this article, we propose a lightweight dynamic convolution method to strengthen traditional CNNs with an affordable increase of total parameters and multiply-adds. Instead of generating the whole kernels directly or combining several static kernels, we choose to "look inside", learning the attention within convolutional kernels. An extra network is used to adjust the weights of kernels for every feature aggregation operation. By combining local and global contexts, the proposed approach can capture the variance among different samples, the variance in different positions of the feature maps, and the variance in different positions inside sliding windows. With a minor increase in the number of model parameters, remarkable improvements in image classification on CIFAR and ImageNet with multiple backbones have been obtained. Experiments on object detection also verify the effectiveness of the proposed method.

Construction of an Iodine-Labeled CS1001 Antibody for Targeting PD-L1 Detection and Comparison with Low-Molecular-Peptide Micro-PET Imaging.

Programmed death-1 (PD-1) and programmed death ligand-1 (PD-L1), the research focus in immune checkpoint regulation, play an important role in tumor immunotherapy. Inhibitors of this pathway are also the focus of tumor immunotherapy research. The PD-1/PD-L1 pathway can be blocked by selective binding to PD-L1. Clinical trials have been conducted in a variety of patients with advanced solid tumors. CS1001 is a high-affinity humanized full-length anti-PD-L1 monoclonal antibody with great clinical significance. We constructed a PD-L1-targeted radioactive molecular probe, 124/125I-labeled full-length antibody CS1001, and evaluated its binding specificity and targeting ability to PD-L1 in tumor cells and tumor models. Additionally, a comparison study with 68Ga-WL12, a PD-L1 targeting peptide, was conducted. The binding potency of 125I-CS1001 to human PD-L1 was evaluated by enzyme-linked immunosorbent assay (ELISA), and the Kd value was 52.1 ± 19.3 nM. The cellular uptake of 125I-CS1001 was examined in Chinese hamster ovary cells (CHO) and CHO expressing human PD-L1 (CHO-hPD-L1). At 2 h, the uptake values of 125I-CS1001 in CHO-hPD-L1 without blocking and in the presence of 0.1 mg non-radiolabeled CS1001 were 3.60 ± 0.08 and 0.09 ± 0.005 (%AD/2 × 105 cells, p < 0.001). Micro-PET imaging was performed between 8 to 192 h after injection of 124I-CS1001 into normal KM mice and CHO-hPD-L1 and HeLa tumor models. The standard uptake value (SUV) of relevant organs in PET images was calculated by drawing regions of interest (ROI). SUVmean of CHO-hPD-L1 tumors was significantly higher than that of HeLa tumors at 48 h (1.98 ± 0.04 vs 0.73 ± 0.14, p = 0.005). The SUVmean of 124I-CS1001 in CHO-hPD-L1 tumors at 48 h was higher than that of 68Ga-WL12 in CHO-hPD-L1 tumors at 0.5 h (1.98 ± 0.04 vs 1.09 ± 0.1 SUVmean, p = 0.007). In conclusion, this work provides a new method for monitoring and evaluating the in vivo expression of PD-L1 in tumors.

Construction of a 124I-Labeled Specific Antibody for the Noninvasive Detection of Mesothelin-Overexpressing Tumors.

Mesothelin (MSLN) is a molecular biomarker of many types of solid tumors, such as mesothelioma, pancreatic cancer, and colon cancer. Owing to the significant difference in expression between cancer cells and normal cells, mesothelin has been widely used as a key target in cancer immunotherapy. In this study, we used iodine isotope (nat/124/125I)-labeled mesothelin antibodies to noninvasively detect MSLN expression in mice with LS174T colon cancer. The 124I-labeled MSLN antibody showed a high radiochemical purity (RCP, >99%) and specific activity (20.8-67.8 GBq/µmol) after purification and was stable in 5% HSA and PBS (>95% RCP at 8 days). Western blot analysis indicated that the LS174T cells showed a higher MSLN protein level than the HepG2 cells. The half maximal effective concentration (EC50) values of the MSLN antibody and natI-anti-MSLN were 34.77 ± 3.72 ng/mL and 32.60 ± 2.52 ng/mL (P = 0.63), respectively. The dissociation constant of 124I-anti-MSLN binding to MSLN protein was 16.0 nM. The radiotracer showed a significantly higher uptake in LS174T cells than in HepG2 tumor cells (1.56 ± 0.09 vs 0.81 ± 0.03, P = 0.0016) 2 days postinjection. The LS174T mouse models showed extremely low organ uptake and high tumor uptake 96 h after the injection of 124I-anti-MSLN, and the T/M values were much higher than those of the other imaging groups (10.56 ± 1.20 for 124I-anti-MSLN in LS174T mice vs 3.27 ± 0.20 for 124I-anti-MSLN in HepG2 mice vs 3.53 ± 0.2 for 124I-IgG in LS174T mice). The immunochemical histology results showed that LS174T tumors were strongly positive (+++) for MSLN, while those in the HepG2 group showed slight expression (+). The dosimetry estimation study showed that the effective dose of 124I-anti-MSLN was 0.185 mSv/MBq, which is within the range of acceptable doses for further nuclear medicine translational research. Taken together, these results suggest that this radiotracer has the potential for detecting mesothelin-overexpressing tumors.

Dual Functional Dopant-Free Contacts with Titanium Protecting Layer: Boosting Stability while Balancing Electron Transport and Recombination Losses.

Combining electron- and hole-selective materials in one crystalline silicon (Si) solar cell, thereby avoiding any dopants, is not considered for application to photovoltaic industry until only comparable efficiency and stable performance are achievable. Here, it is demonstrated how a conventionally unstable electron-selective contact (ESC) is optimized with huge boost in stability as well as improved electron transport. With the introduction of a Ti thin film between a-Si:H(i)/LiF and Al electrode, high-level passivation (Seff  = 4.6 cm s-1 ) from a-Si:H(i) and preferential band alignment (ρC  = 7.9 mΩ cm2 ) from low work function stack of LiF/Ti/Al are both stably retained in the newly constructed n-Si/a-Si:H(i)/LiF/Ti/Al ESC. A detailed interfacial elements analysis reveals that the efficiently blocked inward diffusion of Al from electrode by the Ti protecting layer balances transport and recombination losses in general. This excellent electron-selective properties in combination with large process tolerance that enable remarkable device performance, particularly high efficiencies of 22.12% and 23.61%, respectively, are successfully approached by heterojunction solar cells with dopant-free ESC and dopant-free contacts for both polarities.

124I Radiolabeled Basiliximab for CD25-Targeted Immuno-PET Imaging of Activated T Cells.

Activated T cells played critical roles in immunotherapy and adoptive T cell therapy, and a non-invasive imaging strategy can provide us useful information concerning the transportation, accumulation, and homing of T cells in vivo. In this paper, by utilizing the long half-life radionuclide iodine-124 (124I) and CD25 specific monoclonal antibody Basiliximab, we have fabricated a novel probe, namely, 124I-Basiliximab, which was highly promising in the immuno-PET imaging of T cells. In vitro, 124I-Basiliximab had superior affinity to CD25 protein (Kd = 5.31 nM) and exhibited much higher accumulation in CD25 high-expression lymphoma cell line Karpas299 than that in CD25-negative cell line Daudi. In vivo, 124I-Basiliximab was excreted slowly from the body of mice, rendering it a relatively high effective dose (0.393 mSv/MBq) when applied in the immuno-PET imaging. In Karpas299 tumor xenograft, 124I-Basiliximab probe was observed to accumulate in the tumor quickly after tracer administration, with the optimal image acquired at 24 h post-injection. More importantly, PHA-activated hPBMC had much higher uptake of 124I-Basiliximab, indicating the potential utility of 124I-Basiliximab to discriminate activated hPBMC from its non-activated status. In summary, 124I-Basiliximab was fabricated for the first time, which can be applied in CD25-targeted immuno-PET imaging of activated T cells in vivo.

Noninvasive Mapping of Angiotensin Converting Enzyme-2 in Pigeons Using Micro Positron Emission Tomography.

The ACE2 receptor, as the potential entrance site of SARS-CoV-2-affected cells, plays a crucial role in spreading infection. The DX600 peptide is a competitive inhibitor of ACE2. We previously constructed the 68Ga-labeled DOTA-DX600 (also known as 68Ga-HZ20) peptide and confirmed its ACE2 binding ability both in vitro and in vivo. In this research, we aimed to investigate the noninvasive mapping of ACE2 expression in fowl using 68Ga-HZ20 micro-PET. We chose pigeons as an animal model and first studied the administration method of 68Ga-HZ20 by direct site injection or intravenous injection. Then, the dynamic micro-PET scan of 68Ga-HZ20 was conducted at 0-40 min. Additionally, 18F-FDG was used for comparison. Finally, the pigeons were sacrificed, and the main organs were collected for further immunoPET and IHC staining. Micro PET/CT imaging results showed that 68Ga-HZ20 uptake was distributed from the heart at the preliminary injection to the kidneys, liver, stomach, and lungs over time, where the highest uptake was observed in the kidneys (SUVmax = 6.95, 20 min) and lung (SUVmax = 1.11, 20 min). Immunohistochemical experiments were carried out on its main organs. Compared to the SUVmax data, the IHC results showed that ACE2 was highly expressed in both kidneys and intestines, and the optimal imaging time was determined to be 20 min after injection through correlation analysis. These results indicated that 68Ga-HZ20 is a potential target molecule for SARS-CoV-2 in fowl, which is worthy of promotion and further study.

Development and Clinical Validation of Semi-Supervised Generative Adversarial Networks for Detection of Retinal Disorders in Optical Coherence Tomography Images Using Small Dataset.

PURPOSE: To develop and test semi-supervised generative adversarial networks (GANs) that detect retinal disorders on optical coherence tomography (OCT) images using a small-labeled dataset. METHODS: From a public database, we randomly chose a small supervised dataset with 400 OCT images (100 choroidal neovascularization, 100 diabetic macular edema, 100 drusen, and 100 normal) and assigned all other OCT images to unsupervised dataset (107,912 images without labeling). We adopted a semi-supervised GAN and a supervised deep learning (DL) model for automatically detecting retinal disorders from OCT images. The performance of the 2 models was compared in 3 testing datasets with different OCT devices. The evaluation metrics included accuracy, sensitivity, specificity, and the area under the receiver operating characteristic curves. RESULTS: The local validation dataset included 1000 images with 250 from each category. The independent clinical dataset included 366 OCT images using Cirrus OCT Shanghai Shibei Hospital and 511 OCT images using RTVue OCT from Xinhua Hospital respectively. The semi-supervised GANs classifier achieved better accuracy than supervised DL model (0.91 vs 0.86 for local cell validation dataset, 0.91 vs 0.86 in the Shanghai Shibei Hospital testing dataset, and 0.93 vs 0.92 in Xinhua Hospital testing dataset). For detecting urgent referrals (choroidal neo-vascularization and diabetic macular edema) from nonurgent referrals (drusen and normal) on OCT images, the semi-supervised GANs classifier also achieved better area under the receiver operating characteristic curves than supervised DL model (0.99 vs 0.97, 0.97 vs 0.96, and 0.99 vs 0.99, respectively). CONCLUSIONS: A semi-supervised GAN can achieve better performance than that of a supervised DL model when the labeled dataset is limited. The current study offers utility to various research and clinical studies using DL with relatively small datasets. Semi-supervised GANs can detect retinal disorders from OCT images using relatively small dataset.

Semi-supervised generative adversarial networks for closed-angle detection on anterior segment optical coherence tomography images: an empirical study with a small training dataset.

BACKGROUND: Semi-supervised learning algorithms can leverage an unlabeled dataset when labeling is limited or expensive to obtain. In the current study, we developed and evaluated a semi-supervised generative adversarial networks (GANs) model that detects closed-angle on anterior segment optical coherence tomography (AS-OCT) images using a small labeled dataset. METHODS: In this cross-sectional study, a semi-supervised GANs model was developed for automatic closed-angle detection training on a small labeled and large unsupervised training dataset collected from the Joint Shantou International Eye Center of Shantou University and the Chinese University of Hong Kong (JSIEC). The closed-angle was defined as iris-trabecular contact beyond the scleral spur in AS-OCT images. We further developed two supervised deep learning (DL) models training on the same supervised dataset and the whole dataset separately. The semi-supervised GANs model and supervised DL models' performance were compared on two independent testing datasets from JSIEC (515 images) and the Department of Ophthalmology (84 images), National University Health System, respectively. The diagnostic performance was assessed by evaluation matrices, including the accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC). RESULTS: For closed-angle detection using clinician grading of AS-OCT imaging as the reference standard, the semi-supervised GANs model showed comparable performance, with AUCs of 0.97 (95% CI, 0.96-0.99) and 0.98 (95% CI, 0.94-1.00), compared with the supervised DL model (using the whole dataset) [AUCs of 0.97 (95% CI, 0.96-0.99), and 0.97 (95% CI, 0.94-1.00)]. When training on the same small supervised dataset, the semi-supervised GANs achieved performance at least as well as, if not better than, the supervised DL model [AUCs of 0.90 (95% CI: 0.84-0.96), and 0.92 (95% CI, 0.86-0.97)]. CONCLUSIONS: The semi-supervised GANs method achieves diagnostic performance at least as good as a supervised DL model when trained on small labeled datasets. Further development of semi-supervised learning methods could be useful within clinical and research settings. TRIAL REGISTRATION NUMBER: ChiCTR2000037892.

GINS1 Induced Sorafenib Resistance by Promoting Cancer Stem Properties in Human Hepatocellular Cancer Cells.

Hepatocellular carcinoma (HCC) is characterized by a high rate of incidence and recurrence, and resistance to chemotherapy may aggravate the poor prognosis of HCC patients. Sorafenib resistance is a conundrum to the treatment of advanced/recurrent HCC. Therefore, studies on the molecular pathogenesis of HCC and the resistance to sorafenib are of great interest. Here, we report that GINS1 was highly expressed in HCC tumors, associated with tumor grades, and predicted poor patient survival using Gene Expression Omnibus (GEO) databases exploration. Cell cycle, cell proliferation assay and in vivo xenograft mouse model indicated that knocking down GINS1 induced in G1/S phase cell cycle arrest and decreased tumor cells proliferation in vitro and in vivo. Spheroid formation assay results showed that GINS1 promoted the stem cell activity of HCC tumor cells. Furthermore, GEO database (GSE17112) analysis showed that HRAS oncogenic gene set was enriched in GINS1 high-expressed cancer cells, and quantitative real-time PCR, and Western blot results proved that GINS1 enhanced HCC progression through regulating HRAS signaling pathway. Moreover, knocking down endogenous GINS1 with shGINS1 increased the sensitivity of HCC cells to sorafenib, and restoring HRAS or stem associated pathway partly recovered the sorafenib resistance. Overall, the collective findings highlight GINS1 functions in hepatocarcinogenesis and sorafenib resistance, and indicate its potential use of GINS1 in drug-resistant HCC.

Assessment of Generative Adversarial Networks for Synthetic Anterior Segment Optical Coherence Tomography Images in Closed-Angle Detection.

Purpose: To develop generative adversarial networks (GANs) that synthesize realistic anterior segment optical coherence tomography (AS-OCT) images and evaluate deep learning (DL) models that are trained on real and synthetic datasets for detecting angle closure. Methods: The GAN architecture was adopted and trained on the dataset with AS-OCT images collected from the Joint Shantou International Eye Center of Shantou University and the Chinese University of Hong Kong, synthesizing open- and closed-angle AS-OCT images. A visual Turing test with two glaucoma specialists was performed to assess the image quality of real and synthetic images. DL models, trained on either real or synthetic datasets, were developed. Using the clinicians' grading of the AS-OCT images as the reference standard, we compared the diagnostic performance of open-angle vs. closed-angle detection of DL models and the AS-OCT parameter, defined as a trabecular-iris space area 750 µm anterior to the scleral spur (TISA750), in a small independent validation dataset. Results: The GAN training included 28,643 AS-OCT anterior chamber angle (ACA) images. The real and synthetic datasets for DL model training have an equal distribution of open- and closed-angle images (all with 10,000 images each). The independent validation dataset included 238 open-angle and 243 closed-angle AS-OCT ACA images. The image quality of real versus synthetic AS-OCT images was similar, as assessed by the two glaucoma specialists, except for the scleral spur visibility. For the independent validation dataset, both DL models achieved higher areas under the curve compared with TISA750. Two DL models had areas under the curve of 0.97 (95% confidence interval, 0.96-0.99) and 0.94 (95% confidence interval, 0.92-0.96). Conclusions: The GAN synthetic AS-OCT images appeared to be of good quality, according to the glaucoma specialists. The DL models, trained on all-synthetic AS-OCT images, can achieve high diagnostic performance. Translational Relevance: The GANs can generate realistic AS-OCT images, which can also be used to train DL models.

Detection of Referable Horizontal Strabismus in Children's Primary Gaze Photographs Using Deep Learning.

Purpose: This study implements and demonstrates a deep learning (DL) approach for screening referable horizontal strabismus based on primary gaze photographs using clinical assessments as a reference. The purpose of this study was to develop and evaluate deep learning algorithms that screen referable horizontal strabismus in children's primary gaze photographs. Methods: DL algorithms were developed and trained using primary gaze photographs from two tertiary hospitals of children with primary horizontal strabismus who underwent surgery as well as orthotropic children who underwent routine refractive tests. A total of 7026 images (3829 non-strabismus from 3021 orthoptics [healthy] subjects and 3197 strabismus images from 2772 subjects) were used to develop the DL algorithms. The DL model was evaluated by 5-fold cross-validation and tested on an independent validation data set of 277 images. The diagnostic performance of the DL algorithm was assessed by calculating the accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC). Results: Using 5-fold cross-validation during training, the average AUCs of the DL models were approximately 0.99. In the external validation data set, the DL algorithm achieved an AUC of 0.99 with a sensitivity of 94.0% and a specificity of 99.3%. The DL algorithm's performance (with an accuracy of 0.95) in diagnosing referable horizontal strabismus was better than that of the resident ophthalmologists (with accuracy ranging from 0.81 to 0.85). Conclusions: We developed and evaluated a DL model to automatically identify referable horizontal strabismus using primary gaze photographs. The diagnostic performance of the DL model is comparable to or better than that of ophthalmologists. Translational Relevance: DL methods that automate the detection of referable horizontal strabismus can facilitate clinical assessment and screening for children at risk of strabismus.

Meta-USR: A Unified Super-Resolution Network for Multiple Degradation Parameters.

Recent research on single image super-resolution (SISR) has achieved great success due to the development of deep convolutional neural networks. However, most existing SISR methods merely focus on super-resolution of a single fixed integer scale factor. This simplified assumption does not meet the complex conditions for real-world images which often suffer from various blur kernels or various levels of noise. More importantly, previous methods lack the ability to cope with arbitrary degradation parameters (scale factors, blur kernels, and noise levels) with a single model. A few methods can handle multiple degradation factors, e.g., noninteger scale factors, blurring, and noise, simultaneously within a single SISR model. In this work, we propose a simple yet powerful method termed meta-USR which is the first unified super-resolution network for arbitrary degradation parameters with meta-learning. In Meta-USR, a meta-restoration module (MRM) is proposed to enhance the traditional upscale module with the capability to adaptively predict the weights of the convolution filters for various combinations of degradation parameters. Thus, the MRM can not only upscale the feature maps with arbitrary scale factors but also restore the SR image with different blur kernels and noise levels. Moreover, the lightweight MRM can be placed at the end of the network, which makes it very efficient for iteratively/repeatedly searching the various degradation factors. We evaluate the proposed method through extensive experiments on several widely used benchmark data sets on SISR. The qualitative and quantitative experimental results show the superiority of our Meta-USR.