RESUMEN
Recently, large pretrained vision foundation models based on masked image modeling (MIM) have attracted unprecedented attention and achieved remarkable performance across various tasks. However, the study of MIM for ultrasound imaging remains relatively unexplored, and most importantly, current MIM approaches fail to account for the gap between natural images and ultrasound, as well as the intrinsic imaging characteristics of the ultrasound modality, such as the high noise-to-signal ratio. In this paper, motivated by the unique high noise-to-signal ratio property in ultrasound, we propose a deblurring MIM approach specialized to ultrasound, which incorporates a deblurring task into the pretraining proxy task. The incorporation of deblurring facilitates the pretraining to better recover the subtle details within ultrasound images that are vital for subsequent downstream analysis. Furthermore, we employ a multi-scale hierarchical encoder to extract both local and global contextual cues for improved performance, especially on pixel-wise tasks such as segmentation. We conduct extensive experiments involving 280,000 ultrasound images for the pretraining and evaluate the downstream transfer performance of the pretrained model on various disease diagnoses (nodule, Hashimoto's thyroiditis) and task types (classification, segmentation). The experimental results demonstrate the efficacy of the proposed deblurring MIM, achieving state-of-the-art performance across a wide range of downstream tasks and datasets. Overall, our work highlights the potential of deblurring MIM for ultrasound image analysis, presenting an ultrasound-specific vision foundation model.
RESUMEN
BACKGROUND: The current preoperative malignancy risk evaluation for thyroid nodules involves stepwise diagnostic modalities including ultrasonography, thyroid function serology and fine-needle aspiration (FNA) cytopathology, respectively. We aimed to substantiate the stepwise contributions of each diagnostic step and additionally investigate the diagnostic significance of quantitative chromogenic imprinted gene in-situ hybridization (QCIGISH)-an adjunctive molecular test based on epigenetic imprinting alterations. METHODS: A total of 114 cytopathologically-diagnosed and histopathologically-confirmed thyroid nodules with complete ultrasonographic and serological examination records were evaluated using QCIGISH in the study. Logistic regression models for thyroid malignancy prediction were developed with the stepwise addition of each diagnostic modality and the contribution of each step evaluated in terms of discrimination performance and goodness-of-fit. RESULTS: From the baseline model using ultrasonography [area under the receiver operating characteristics curve (AUROC): 0.79; 95% confidence interval (CI): 0.71-0.86], significant improvements in thyroid malignancy discrimination were observed with the stepwise addition of thyroid function serology (AUROC: 0.82; 95% CI: 0.74-0.90; P=0.23) and FNA cytopathology (AUROC: 0.88; 95% CI: 0.81-0.94; P=0.02), respectively. The inclusion of QCIGISH as an adjunctive molecular test further advanced the preceding model's diagnostic performance (AUROC: 0.95; 95% CI: 0.91-1.00, P=0.007). CONCLUSIONS: Our study demonstrated the significant stepwise diagnostic contributions of standard clinical assessments in the malignancy risk stratification of thyroid nodules. However, the addition of molecular imprinting detection further enabled a more accurate and definitive preoperative evaluation especially for morphologically indeterminate thyroid nodules and cases with potentially discordant results among standard modalities.
Asunto(s)
Impresión Genómica , Humanos , Femenino , Masculino , Persona de Mediana Edad , Adulto , Neoplasias de la Tiroides/genética , Biopsia con Aguja Fina/métodos , Nódulo Tiroideo/genética , Anciano , Glándula Tiroides/patologíaRESUMEN
BACKGROUND: Early neurological deterioration, a common complication in patients with intracerebral hemorrhage, is associated with poor outcomes. Despite the fact that the prevalence and predictors of early neurological impairment are widely addressed, few studies have consolidated these findings. This study aimed to systematically investigate the prevalence and predictors of early neurological deterioration. METHODS: The PubMed, Embase, Cochrane Library, CIHNAL, and Web of Science databases were systematically searched for relevant studies from the inception to December 2023. The data were extracted using a predefined worksheet. Quality assessment was conducted using the Newcastle-Ottawa Scale. Two reviewers independently performed the study selection, data extraction, and quality appraisal. The pooled effect size and 95% confidence intervals were calculated using the STATA 17.0 software package. RESULTS: In total, 32 studies and 5,014 patients were included in this meta-analysis. The prevalence of early neurological deterioration was 23% (95% CI 21-26%, p < 0.01). The initial NIHSS score (OR = 1.24, 95% CI 1.17, 1.30, p < 0.01), hematoma volume (OR = 1.07, 95% CI 1.06, 1.09, p < 0.01), intraventricular hemorrhage (OR = 3.50, 95% CI 1.64, 7.47, p < 0.01), intraventricular extension (OR = 3.95, 95% CI 1.96, 7.99, p < 0.01), hematoma expansion (OR = 9.77, 95% CI 4.43, 17.40, p < 0.01), and computed tomographic angiography spot sign (OR = 5.77, 95% CI 1.53, 20.23, p = 0.01) were predictors of early neurological deterioration. The funnel plot and Egger's test revealed significant publication bias (p < 0.001). CONCLUSIONS: This meta-analysis revealed a pooled prevalence of early neurological deterioration of 23% in patients with intracerebral hemorrhage. The initial NIHSS score, hematoma volume, intraventricular hemorrhage, intraventricular expansion, hematoma expansion, and spot sign enhanced the probability of early neurological deterioration. These findings provide healthcare providers with an evidence-based basis for detecting and managing early neurological deterioration in patients with intracerebral hemorrhage.
Asunto(s)
Hemorragia Cerebral , Humanos , Hemorragia Cerebral/diagnóstico por imagen , Hemorragia Cerebral/complicaciones , Enfermedades del Sistema Nervioso/etiología , Enfermedades del Sistema Nervioso/epidemiología , Deterioro Clínico , Progresión de la EnfermedadRESUMEN
Background: Ultrasound-guided fine needle aspiration thyroglobulin (FNA-Tg) is recommended for the diagnosis of lymph node metastasis (LNM) in differentiated thyroid cancer (DTC), but its optimal cutoff value remains controversial, and the effect of potential influencing factors on FNA-Tg levels is unclear. Method: In this study, a retrospective analysis was conducted on 281 patients diagnosed with DTC, encompassing 333 lymph nodes. We analyze the optimal cutoff value and diagnostic efficacy of FNA-Tg, while also evaluating the potential influence of various factors on FNA-Tg. Results: For FNA-Tg, the optimal cutoff value was 16.1 ng/mL (area under the curve (AUC)= 0.942). The optimal cutoff value for FNA-Tg/sTg was 1.42 (AUC = 0.933). The AUC for FNA combined with FNA-Tg yielded the highest value compared to other combined diagnostic methods (AUC = 0.955). It has been found that serum thyroglobulin (sTg) is positively correlated with FNA-Tg (Rs = 0.318), while serum thyroglobulin antibodies (sTgAb) is negatively correlated with FNA-Tg (Rs = -0.147). In cases where the TNM stage indicated N1b, the presence of large or high volume lymph node metastasis(HVLNM), lymph node lateralization/suspicion (L/S) ratio ≤ 2, ultrasound findings indicating lymph node liquefaction, calcification, and increased blood flow, patients with coexisting Hashimoto's thyroiditis (HT), a tumor size ≥10 mm, and postoperative pathology confirming invasion of the thyroid capsule, higher levels of FNA-Tg were observed. However, the subgroup classification of DTC and the presence or absence of thyroid tissue did not demonstrate any significant impact on the levels of FNA-Tg. Conclusion: The findings of this study indicate that the utilization of FNA in conjunction with FNA-Tg is a crucial approach for detecting LNM in DTC. TNM stage indicated N1b, the presence of HVLNM, the presence of HT, lymph node L/S ratio, liquefaction, calcification, tumor diameter, sTg and sTgAb are factors that can impact FNA-Tg levels.In the context of clinical application, it is imperative to individualize the use of FNA-Tg.
Asunto(s)
Carcinoma Papilar , Enfermedad de Hashimoto , Neoplasias de la Tiroides , Humanos , Tiroglobulina , Cáncer Papilar Tiroideo/diagnóstico , Biopsia con Aguja Fina/métodos , Metástasis Linfática , Estudios Retrospectivos , Carcinoma Papilar/diagnóstico por imagen , Carcinoma Papilar/patología , Neoplasias de la Tiroides/patología , Ultrasonografía IntervencionalRESUMEN
Objective: To evaluate the inhibitory effect of ginsenoside Rg3 combined with 5-fluorouracil (5-FU) on tumor angiogenesis and tumor growth in colon cancer in mice. Methods: CT26 mouse model of colon cancer was established and the mice were randomly assigned to the control group, the ginsenoside Rg3 group, the 5-FU group, and the Rg3 combined with 5-FU group. The 5-FU group was injected intraperitoneally at the dose of 20 mg/kg, 0.2 mL/animal, and once a day for 10 days. Treatment for the Rg3 group was given at the dose of 20 mg/kg, 0.2 mL/animal, and once a day for 21 days via gastric gavage. The dose and the mode of treatment for the Rg3+5-FU combination group were the same as those for the 5-FU and the Rg3 group. The control group was intraperitoneally injected with 0.2 mL/d of normal saline for 10 days. The expression of vascular endothelial growth factor (VEGF) and CD31 and the microvascular density (MVD) of the tumor tissues were examined by immunohistochemistry. The blood flow signals and tumor necrosis were examined by color Doppler flow imaging (CDFI). The quality of life, survival rate, tumor volume, tumor mass, and tumor inhibition rate of the mice were monitored. Results: After 21 days of treatment, the tumor volume and the tumor mass of all treatment groups were significantly decreased compared with those the control group, with the combination treatment group exhibiting the most significant decrease. The tumor inhibition rates of the Rg3 group, the 5-FU group, and the combination group were 29.96%, 68.78%, and 73.42%, respectively. Rg3 treatment alone had inhibitory effect on tumor growth to a certain degree, while 5-FU treatment alone or 5-FU combined with Rg3 had a stronger inhibitory effect on tumor growth. The tumor inhibition rate of the combination group was higher than that of the 5-FU group, but the difference was not statistically significant (P>0.05). Color Doppler ultrasound showed that there were multiple localized and large tumor necrotic areas that were obvious and observable in the Rg3 group and the combination group, and that there were only small tumor necrotic areas in the 5-FU group and the control group. The tumor necrosis rate of the combination group was (55.63±3.12)%, which was significantly higher than those of the other groups (P<0.05). CDFI examination of the blood flow inside of the tumor of the mice showed that the blood flow signals in the combination group were mostly grade 0-â , and that the blood flow signals in the control group were the most abundant, being mostly grade â ¡-â ¢. The abundance of the blood flow signals in the Rg3 and 5-FU groups were between those of the control group and the combination group. Compared with those of the control group, the expression levels of MVD and VEGF in the tumor tissues of the Rg3 group, the 5-FU group, and the combination group were significantly decreased, with the combination group showing the most significant decrease (P<0.05). HE staining results indicated that there was significant tumor necrosis in mice in the control group and that there were more blood vessels. In contrast, in the tumor of the Rg3 group and the 5-FU group, there were fewer blood vessels and necrotic gaps appeared within the tumors. In the combination group, the tumor tissues had the fewest blood vessels and rope-like necrosis was observed. The mice started dying on the 18th day after treatment started, and all the mice in the control group died on the 42nd day. By this time, there were 3, 5, and 7 mice still alive in the Rg3 group, the 5-FU group, and the combination group, respectively, presenting a survival rate of 30%, 50%, and 70%, respectively. All mice in all the groups died on day 60 after treatment started. Conclusion: Ginsenoside Rg3 combined with 5-FU can significantly inhibit tumor angiogenesis and tumor growth of colon cancer in mice and improve the survival and quality of life of tumor-bearing mice.
Asunto(s)
Neoplasias del Colon , Ginsenósidos , Ratones , Animales , Fluorouracilo/uso terapéutico , Factor A de Crecimiento Endotelial Vascular/metabolismo , Angiogénesis , Calidad de Vida , Ginsenósidos/farmacología , Ginsenósidos/uso terapéutico , Neovascularización Patológica/tratamiento farmacológico , Necrosis/tratamiento farmacológico , Línea Celular TumoralRESUMEN
Purpose: To assess and compare the effectiveness of ultrasound-guided core needle biopsy (CNB) in comparison to repeat fine-needle aspiration(rFNA) for thyroid nodules that yield inconclusive results following the initial fine-needle aspiration (FNA). Methods: A cohort of 471 patients who received an inconclusive cytological diagnosis following the initial FNA were included in this study. These patients subsequently underwent either CNB (n=242) or rFNA (n=229). The inconclusive FNA results encompassed categories I, III, and IV of The Bethesda System for Reporting Thyroid Cytopathology(TBSRTC), as well as the ultrasound images indicating malignancy despite FNA results falling under TBSRTC category II. This study assessed the sampling satisfaction rate, diagnostic efficacy, and complications associated with CNB compared to rFNA. Additionally, the impact of repeat puncture time and nodule size on diagnostic efficacy was analyzed. Results: Following repeat punctures, the satisfaction rate of the CNB sampling was found to be significantly higher than that of rFNA (83.9% vs 66.8%). The diagnostic rate in the CNB group was significantly greater compared to that of the rFNA group (70.7% vs 35.8%). In patients with nodule maximum diameters ranging from 5 mm to 20 mm, the diagnostic accuracy was significantly higher in the CNB group compared to that in the rFNA group. In patients with intervals less than 90 days, between 90 days and one year, the diagnostic rate in the CNB group was found to be higher compared to that in the rFNA group. In CNB, not immediately adjacent to the capsule was a risk factor for nodular puncture bleeding (37.0% vs 22.7%.). Conclusion: CNB demonstrated higher rates of satisfaction and diagnosis compared to the rFNA. The diagnostic effectiveness of CNB was not influenced by the time interval or the size of the thyroid nodule. Therefore, in cases where the initial FNA diagnosis of thyroid nodules is inconclusive, CNB should be considered as a viable option for re-puncture.
Asunto(s)
Nódulo Tiroideo , Humanos , Nódulo Tiroideo/diagnóstico , Nódulo Tiroideo/patología , Biopsia con Aguja Fina/métodos , Biopsia con Aguja Gruesa/métodos , Estudios RetrospectivosRESUMEN
Background and Objective: Deep learning (DL) has contributed substantially to the evolution of image analysis by unlocking increased data and computational power. These DL algorithms have further facilitated the growing trend of implementing precision medicine, particularly in areas of diagnosis and therapy. Thyroid imaging, as a routine means to screening for thyroid diseases on large-scale populations, is a massive data source for the development of DL models. Thyroid disease is a global health problem and involves structural and functional changes. The objective of this study was to evaluate the general rules and future directions of DL networks in thyroid medical image analysis through a review of original articles published between 2018 and 2023. Methods: We searched for English-language articles published between April 2018 and September 2023 in the databases of PubMed, Web of Science, and Google Scholar. The keywords used in the search included artificial intelligence or DL, thyroid diseases, and thyroid nodule or thyroid carcinoma. Key Content and Findings: The computer vision tasks of DL in thyroid imaging included classification, segmentation, and detection. The current applications of DL in clinical workflow were found to mainly include management of thyroid nodules/carcinoma, risk evaluation of thyroid cancer metastasis, and discrimination of functional thyroid diseases. Conclusions: DL is expected to enhance the quality of thyroid images and provide greater precision in the assessment of thyroid images. Specifically, DL can increase the diagnostic accuracy of thyroid diseases and better inform clinical decision-making.
RESUMEN
Background: The role of quantitative contrast-enhanced ultrasound (CEUS) in the evaluation of thyroid nodules with Hashimoto's thyroiditis (HT) has received little attention. Methods: This was a retrospective cohort study. We consecutively enrolled 242 patients (49 males, 193 females, average age 52 years) with a combined total of 248 thyroid nodules coexisting with HT who underwent biopsy/resection-proven pathology from December 2016 to June 2021. All patients underwent preoperative ultrasound (US) and CEUS examinations performed by 2 radiologists independently. Quantitative analysis of CEUS using time-intensity curves (TIC) was measured by an expert radiologist from the thyroid intra-nodule and the surrounding parenchyma and their ratios. Receiver operating characteristic (ROC) analysis was performed to evaluate their diagnostic performance. Results: The patients were divided into the nodular HT (NHT) group (n=42), the papillary thyroid carcinoma (PTC) group (n=154), and the primary thyroid lymphoma (PTL) group (n=52) according to their pathological results. TIC parameters revealed that PTC and PTL showed faster time to peak (TTP) (P=0.044, P=0.049), lower peak intensity (PI) (both P<0.001), and smaller areas under the curve (both P<0.001) than those of NHT. The intra nodule of PTL showed an obviously slower perfusion (ratio =0.90, P<0.001) and lower PI (ratio =0.84, P<0.001) compared with the thyroid parenchyma. TIC improved performance in distinguishing PTL from NHT [area under the curve (AUC): 0.947, 95% confidence interval (CI): 0.903-0.991], but inferior performance in differentiating PTC and NHT (AUC: 0.838, 95% CI: 0.759-0.917). Conclusions: CEUS quantitative analysis could be valuable in differentiating thyroid malignancies in patients with HT.
RESUMEN
Purpose: Thyroid nodules classified as indeterminate in previous fine-needle aspiration cytology often necessitate additional evaluation to determine their histology, while shear wave elastography (SWE) offers an alternative option in this regard. The objective of this study was to assess the diagnostic effectiveness of SWE in evaluating indeterminate nodules. Methods: The PubMed, EMBASE, and Web of Science databases were searched from 1st January 1970 to 1st March 2023. The studies were reviewed and the data was extracted by two separate reviewers. A Bayesian bivariate model was utilized to quantitatively synthesize the diagnostic accuracy and yield of the studies in R. Results: A total of seven studies, involving indeterminate thyroid nodules undergoing SWE were included, and the overall malignancy rate was 34.1% (307/900). The summarized estimates of sensitivity and specificity were 0.792 (95% credible interval [CI], 0.727-0.850) and 0.845 (95% CI, 0.797-0.887), respectively. The summarized estimate for the diagnostic odds ratio (DOR) was 17.8 (95% CI, 14.0-22.6). Summarized receiver operating characteristic (SROC) plots indicated a trade-off between sensitivity and specificity, and the estimate of AUC was 0.866 (95% CI, 0.834-0.895). The summary estimates for positive and negative likelihood ratios were 4.67 (95% CI, 3.98-5.85) and 0.26 (95% CI, 0.23-0.28), respectively. Conclusions: The overall accuracy of SWE remains satisfactory in indeterminate thyroid nodules. However, it should be noted that the available data are still extremely limited, and more studies or guidelines are required to provide further insights.
RESUMEN
In recent years, the incidence of thyroid cancer has been increasing. Thyroid nodule detection is critical for both the detection and treatment of thyroid cancer. Convolutional neural networks (CNNs) have achieved good results in thyroid ultrasound image analysis tasks. However, due to the limited valid receptive field of convolutional layers, CNNs fail to capture long-range contextual dependencies, which are important for identifying thyroid nodules in ultrasound images. Transformer networks are effective in capturing long-range contextual information. Inspired by this, we propose a novel thyroid nodule detection method that combines the Swin Transformer backbone and Faster R-CNN. Specifically, an ultrasound image is first projected into a 1D sequence of embeddings, which are then fed into a hierarchical Swin Transformer. The Swin Transformer backbone extracts features at five different scales by utilizing shifted windows for the computation of self-attention. Subsequently, a feature pyramid network (FPN) is used to fuse the features from different scales. Finally, a detection head is used to predict bounding boxes and the corresponding confidence scores. Data collected from 2,680 patients were used to conduct the experiments, and the results showed that this method achieved the best mAP score of 44.8%, outperforming CNN-based baselines. In addition, we gained better sensitivity (90.5%) than the competitors. This indicates that context modeling in this model is effective for thyroid nodule detection.
Asunto(s)
Neoplasias de la Tiroides , Nódulo Tiroideo , Humanos , Nódulo Tiroideo/diagnóstico por imagen , Neoplasias de la Tiroides/diagnóstico por imagen , Ultrasonografía , Suministros de Energía Eléctrica , Procesamiento de Imagen Asistido por ComputadorRESUMEN
Pluripotent stem cells (PSCs) are a promising source of stem cells for regenerative therapies. Stem cell function depends on telomere maintenance mechanisms that provide them with the proliferative capacity and genome stability necessary to multiply and regenerate tissues. We show here that established human embryonic stem cells (hESCs) have stable telomere length that is dependent on telomerase but not on alternative mechanisms based on homologous recombination pathways. Here, we show that human-induced pluripotent stem cells (hiPSCs) reprogrammed from somatic cells show progressive telomere lengthening until reaching a length similar to ESCs. hiPSCs also acquire telomeric chromatin marks of ESCs including decreased abundance of tri-methylated histone H3K9 and H4K20 and HP1 heterochromatic marks, as well as of the shelterin component TRF2. These chromatin features are accompanied with increased abundance of telomere transcripts or TERRAs. We also found that telomeres of both hESCs and hiPSCs are well protected from DNA damage during telomere elongation and once full telomere length is achieved, and exhibit stable genomes. Collectively, this study highlights that hiPSCs acquire ESC features during reprogramming and reveals the telomere biology in human pluripotent stem cells (hPSCs).
We show that established human embryonic stem cells (hESCs) have a maximum and stable telomere length that is dependent on telomerase but not on the alternative homologous recombination pathway or ALT. Human-induced pluripotent stem cells (hiPSCs) reprogrammed from somatic cells show progressive telomere lengthening until reaching a length similar maximum telomere length than ESCs, suggesting that telomere length is regulated by epigenetic mechanisms in human cells. In this regard, hiPSCs acquire telomeric chromatin marks characteristic of an "open chromatin" including increased abundance of telomere transcripts or TERRAs. Telomeres of both hESCs and hiPSCs are well protected during telomere elongation and exhibit stable genomes. Collectively, this study highlights that hiPSCs acquire ESC features during reprogramming and reveals the telomere biology in human pluripotent stem cells (hiPSCs).
Asunto(s)
Células Madre Pluripotentes , Telomerasa , Humanos , Células Madre Pluripotentes/metabolismo , Homeostasis del Telómero , Histonas/metabolismo , Telómero/metabolismo , Telomerasa/genética , Telomerasa/metabolismo , Cromatina/metabolismoRESUMEN
It remains a major challenge to simultaneously achieve bone regeneration and prevent infection in the complex microenvironment of repairing bone defects. Here, we developed a novel ECM-mimicking scaffold by coaxial electrospinning to be endowed with multibiological functions. Lysophosphatidic acid (LPA) and zinc oxide (ZnO) nanoparticles were loaded into the poly-lactic-co-glycolic acid/polycaprolactone (PLGA/PCL, PP) sheath layer of coaxial nanofibers, and deferoxamine (DFO) nanoparticles were loaded into its core layer. The novel scaffold PP-LPA-ZnO/DFO maintained a porous nanofibrous architecture after incorporating three active nanoparticles, showing better physicochemical properties and eximious biocompatibility. In vitro studies showed that the bio-scaffold loaded with LPA nanoparticles had excellent cell adhesion, proliferation, and differentiation for MC3T3-E1 cells and synergistic osteogenesis with the addition of ZnO and DFO nanoparticles. Further, the PP-LPA-ZnO/DFO scaffold promoted tube formation and facilitated the expression of vascular endothelial markers in HUVECs. In vitro antibacterial studies against Escherichia Coli and Staphylococcus aureus demonstrated effective antibacterial activity of the PP-LPA-ZnO/DFO scaffold. In vivo studies showed that the PP-LPA-ZnO/DFO scaffold exhibited excellent biocompatibility after subcutaneous implantation and remarkable osteogenesis at 4 weeks post-implantation in the mouse alveolar bone defects. Importantly, the PP-LPA-ZnO/DFO scaffold showed significant antibacterial activity, prominent neovascularization, and new bone formation in the rat fenestration defect model. Overall, the spatially sustained release of LPA, ZnO, and DFO nanoparticles through the coaxial scaffold synergistically enhanced biocompatibility, osteogenesis, angiogenesis, and effective antibacterial properties, which is ultimately beneficial for bone regeneration. This project provides the optimized design of bone regenerative biomaterials and a new strategy for bone regeneration, especially in the potentially infected microenvironment.
Asunto(s)
Nanopartículas , Óxido de Zinc , Animales , Antibacterianos/farmacología , Regeneración Ósea , Matriz Extracelular , Ratones , Nanopartículas/química , Osteogénesis , Ratas , Ingeniería de Tejidos , Andamios del Tejido/química , Óxido de Zinc/farmacologíaRESUMEN
BACKGROUND: The intelligent diagnosis of thyroid nodules in ultrasound image is an important research issue. Automatically locating the region of interest (ROI) of thyroid nodules and providing pre-diagnosis results can help doctors to diagnose faster and more accurate. OBJECTIVES: This study aims to propose a model, which can detect multiple nodules stably and accurately in order to avoid missed detection and misjudgment. In addition, the detection speed of the model needs to be fast for real-time diagnosis in ultrasound images. METHODS: Based on the object detection technology, we propose an accurate, robust and high-speed network with multiscale fusion strategy called Efficient-YOLO, which can realize the localization and recognition of nodules at the same time. Finally, multiple metrics are used to measure the diagnostic ability of the model. RESULTS: Experimental results conducted on 3,562 ultrasound images show that our new model greatly increases the accuracy and speed of the detection compared with the baseline model. The best mAP is 92.64%, and the fastest detection speed is 45.1 frames per second. CONCLUSIONS: This study proposed an effective method to diagnosis thyroid nodules automatically, which can meet the real-time requirements, indicating that its effectiveness and feasibility for future clinical application.
Asunto(s)
Nódulo Tiroideo , Benchmarking , Humanos , Redes Neurales de la Computación , Nódulo Tiroideo/diagnóstico por imagen , Ultrasonografía/métodosRESUMEN
Background: Radiomics strategies exhibit great promise in the context of thyroid nodule diagnosis. This study aimed to compare radiomics features of different sizes of medullary thyroid carcinoma (MTC) and papillary thyroid carcinoma (PTC) tumors and to compare the efficiency of radiomics approaches as a means of differentiating between these tumor types. Methods: In total, 86 MTC and 330 PTC nodules were divided into the macronodular (>10 mm) and micronodular (⩽10 mm) categories. The radiomics features of these nodules were analyzed to identify independent prognosis factors and evaluate the efficacy of individual and combined indicators as predictors of tumor type. Results: In total, 12 radiomics features were found to differ significantly between MTC and PTC macronodules, while 6 differed significantly between MTC and PTC micronodules. Shape 2D_Sphericity, firstorder_Skewness, glrlm_RunLengthNonUniformity, glszm_GrayLevelNonUniformity, and glszm_SizeZoneNonUniformity were features that were independently associated with the differential diagnoses of MTC and PTC macronodules. Receiver operating characteristic (ROC) curve analyses of the efficacy of these 5 single indicators and a combined indicator composed thereof yielded area under the curve (AUC) values of 0.621, 0.678, 0.704, 0.762, 0.747, and 0.824, respectively, with respective sensitivities of 55.3%, 43.0%, 53.1%, 56.3%, 46.9%, and 65.6%, and respective specificity values of 65.6%, 89.1%, 81.6%, 88.8%, 95.0%, and 91.1%. The glrlm_RunEntropy and glszm_SizeZoneNonUniformity features were identified as independent factors associated with the differential diagnoses of MTC and PTC micronodules. Receiver operating characteristic curve analyses of the efficacy of these 2 single indicators and a combined indicator composed thereof yielded respective AUC values of 0.678, 0.678, and 0.771; Sensitivities of 57.0%, 72.7%, and 72.7%; and specificities of 77.3%, 64.2%, and 77.5%. Conclusions: A range of different radiomics features can enable effective differentiation between MTC and PTC nodules of different sizes. Moreover, analyses of combinations of radiomics features yielded diagnostic efficiency values higher than those associated with single radiomics features, highlighting a more reliable approach to diagnosing MTC and PTC tumors.
RESUMEN
Thyroid nodules are one of the most common nodular lesions. The incidence of thyroid cancer has increased rapidly in the past three decades and is one of the cancers with the highest incidence. As a non-invasive imaging modality, ultrasonography can identify benign and malignant thyroid nodules, and it can be used for large-scale screening. In this study, inspired by the domain knowledge of sonographers when diagnosing ultrasound images, a local and global feature disentangled network (LoGo-Net) is proposed to classify benign and malignant thyroid nodules. This model imitates the dual-pathway structure of human vision and establishes a new feature extraction method to improve the recognition performance of nodules. We use the tissue-anatomy disentangled (TAD) block to connect the dual pathways, which decouples the cues of local and global features based on the self-attention mechanism. To verify the effectiveness of the model, we constructed a large-scale dataset and conducted extensive experiments. The results show that our method achieves an accuracy of 89.33%, which has the potential to be used in the clinical practice of doctors, including early cancer screening procedures in remote or resource-poor areas.
Asunto(s)
Nódulo Tiroideo , Humanos , Nódulo Tiroideo/diagnóstico por imagen , Nódulo Tiroideo/patología , Ultrasonografía/métodosRESUMEN
PURPOSE: This study investigates the efficiency of deep learning models in the automated diagnosis of Hashimoto's thyroiditis (HT) using real-world ultrasound data from ultrasound examinations by computer-assisted diagnosis (CAD) with artificial intelligence. METHODS: We retrospectively collected ultrasound images from patients with and without HT from 2 hospitals in China between September 2008 and February 2018. Images were divided into a training set (80%) and a validation set (20%). We ensembled 9 convolutional neural networks (CNNs) as the final model (CAD-HT) for HT classification. The model's diagnostic performance was validated and compared to 2 hospital validation sets. We also compared the accuracy of CAD-HT against seniors/junior radiologists. Subgroup analysis of CAD-HT performance for different thyroid hormone levels (hyperthyroidism, hypothyroidism, and euthyroidism) was also evaluated. RESULTS: 39 280 ultrasound images from 21 118 patients were included in this study. The accuracy, sensitivity, and specificity of the HT-CAD model were 0.892, 0.890, and 0.895, respectively. HT-CAD performance between 2 hospitals was not significantly different. The HT-CAD model achieved a higher performance (P < 0.001) when compared to senior radiologists, with a nearly 9% accuracy improvement. HT-CAD had almost similar accuracy (range 0.87-0.894) for the 3 subgroups based on thyroid hormone level. CONCLUSION: The HT-CAD strategy based on CNN significantly improved the radiologists' diagnostic accuracy of HT. Our model demonstrates good performance and robustness in different hospitals and for different thyroid hormone levels.
Asunto(s)
Inteligencia Artificial , Enfermedad de Hashimoto , Diagnóstico por Computador , Enfermedad de Hashimoto/diagnóstico por imagen , Humanos , Redes Neurales de la Computación , Estudios RetrospectivosRESUMEN
BACKGROUND & AIMS: Mature transfer RNAs (tRNA) charged with amino acids decode mRNA to synthesize proteins. Dysregulation of translational machineries has a fundamental impact on cancer biology. This study aims to map the tRNAome landscape in liver cancer patients and to explore potential therapeutic targets at the interface of charging amino acid with tRNA. METHODS: Resected tumour and paired tumour-free (TFL) tissues from hepatocellular carcinoma (HCC) patients (n = 69), and healthy liver tissues from organ transplant donors (n = 21), HCC cell lines, and cholangiocarcinoma (CC) patient-derived tumour organoids were used. RESULTS: The expression levels of different mature tRNAs were highly correlated and closely clustered within individual tissues, suggesting that different members of the tRNAome function cooperatively in protein translation. Interestingly, high expression of tRNA-Lys-CUU in HCC tumours was associated with more tumour recurrence (HR 1.1; P = .022) and worse patient survival (HR 1.1; P = .0037). The expression of Lysyl-tRNA Synthetase (KARS), the enzyme catalysing the charge of lysine to tRNA-Lys-CUU, was significantly upregulated in HCC tumour tissues compared to tumour-free liver tissues. In HCC cell lines, lysine deprivation, KARS knockdown or treatment with the KARS inhibitor cladosporin effectively inhibited overall cell growth, single cell-based colony formation and cell migration. This was mechanistically mediated by cell cycling arrest and induction of apoptosis. Finally, these inhibitory effects were confirmed in 3D cultured patient-derived CC organoids. CONCLUSIONS: The biological process of charging tRNA-Lys-CUU with lysine sustains liver cancer cell growth and migration, and is clinically relevant in HCC patients. This process can be therapeutically targeted and represents an unexplored territory for developing novel treatment strategies against liver cancer.
Asunto(s)
Fenómenos Biológicos , Carcinoma Hepatocelular , Neoplasias Hepáticas , Carcinoma Hepatocelular/tratamiento farmacológico , Carcinoma Hepatocelular/genética , Humanos , Neoplasias Hepáticas/tratamiento farmacológico , Neoplasias Hepáticas/genética , Lisina , Recurrencia Local de Neoplasia , Aminoacilación de ARN de TransferenciaRESUMEN
BACKGROUND & AIMS: Cancer-associated fibroblasts (CAFs) play a key role in the cancer process, but the research progress is hampered by the paucity of preclinical models that are essential for mechanistic dissection of cancer cell-CAF interactions. Here, we aimed to establish 3-dimensional (3D) organotypic co-cultures of primary liver tumor-derived organoids with CAFs, and to understand their interactions and the response to treatment. METHODS: Liver tumor organoids and CAFs were cultured from murine and human primary liver tumors. 3D co-culture models of tumor organoids with CAFs and Transwell culture systems were established in vitro. A xenograft model was used to investigate the cell-cell interactions in vivo. Gene expression analysis of CAF markers in our hepatocellular carcinoma cohort and an online liver cancer database indicated the clinical relevance of CAFs. RESULTS: To functionally investigate the interactions of liver cancer cells with CAFs, we successfully established murine and human 3D co-culture models of liver tumor organoids with CAFs. CAFs promoted tumor organoid growth in co-culture with direct cell-cell contact and in a Transwell system via paracrine signaling. Vice versa, cancer cells secrete paracrine factors regulating CAF physiology. Co-transplantation of CAFs with liver tumor organoids of mouse or human origin promoted tumor growth in xenograft models. Moreover, tumor organoids conferred resistance to clinically used anticancer drugs including sorafenib, regorafenib, and 5-fluorouracil in the presence of CAFs, or the conditioned medium of CAFs. CONCLUSIONS: We successfully established murine and human 3D co-culture models and have shown robust effects of CAFs in liver cancer nurturing and treatment resistance.