Rationale:Triple-negative breast cancer (TNBC) remains one of the most aggressive subtypes due to its poor immunogenicity and resistance to systemic therapies. Methods:Here, we report a multifunctional NIR-II light-triggered theranostic nanoplatform, termed LQPO, that integrates liposome, Ti3C2 quantum dots (QDs), perfluorohexane (PFH), and ozone to enable multimodal imaging-guided photo-gas synergistic therapy. The QDs, embedded within the liposomal bilayer, act as highly efficient NIR-II photothermal transducers, while the PFH core serves as a liquid-gas phase-change medium capable of dissolving and releasing ozone under localized heating. Results:Upon NIR-II irradiation, LQPO produced strong photothermal and photoacoustic signals for real-time photoacoustic (PA) and contrast-enhanced ultrasound (CEUS) imaging, accompanied by vaporization of the PFH core and burst release of ozone. This spatiotemporally coordinated cascade induced potent oxidative stress and hyperthermia, synergistically driving GSDME-dependent pyroptosis and immunogenic cell death (ICD). The resulting "in situ vaccination" effect remodeled the tumor immune microenvironment and primed tumors for PD-1 blockade therapy. In vivo, NIR-II-activated LQPO achieved efficient tumor accumulation, strong PA/CEUS imaging contrast, and pronounced inhibition of both primary and abscopal tumors when combined with αPD-1 therapy. No significant systemic toxicity was observed, confirming its favorable biosafety. Conclusions:Overall, this study establishes a single-laser-activated nanoplatform that unifies real-time multimodal imaging and photo-gas synergistic therapy, and converts localized treatment into a systemic antitumor immune response upon integration with checkpoint inhibition.

OBJECTIVES:This study aimed to develop and validate an ultrasound-based IHC4-associated radiomic model for predicting late recurrence in ER-positive breast cancer and to evaluate its potential to support risk stratification and guide decisions on extended endocrine therapy. METHODS:In this retrospective multicenter study, patients were divided into a training cohort, an internal validation cohort, and two external validation cohorts. Radiomic features associated with the immunohistochemical four-marker (IHC4) score were selected to construct a support vector machine (SVM)-based IHC4-associated radiomic model for generating an IHC4-associated radiomic score. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC). Late distant recurrence (DR) was defined as the primary endpoint, and progression-free survival (PFS) was defined as the secondary endpoint. The prognostic value of the score was further evaluated using Cox regression analysis. RESULTS:A total of 523 patients were included in this study. Seven IHC4-associated radiomic features were selected to construct the IHC4-associated radiomic model. The model demonstrated consistent performance across cohorts, with AUCs of 0.81, 0.84, 0.79, and 0.81 in the training, internal validation, and two external validation cohorts, respectively. The IHC4-associated radiomic score stratified late DR risk across the training, internal validation, and external validation cohorts. In the secondary exploratory PFS analysis, the score remained associated with PFS in multivariable Cox analysis (HR = 4.246, 95% CI: 1.749-10.307, P = .001). CONCLUSIONS:The IHC4-associated radiomic model may serve as a noninvasive biomarker for predicting prognosis in ER-positive breast cancer.

PURPOSE:To assess the added diagnostic value of contrast-enhanced ultrasound (CEUS) in differentiating small breast lesions initially categorized as ACR BI-RADS 4A. MATERIALS AND METHODS:A total of 1595 patients with small breast lesions (≤ 20 mm) from 37 tertiary hospitals were enrolled from August 2021 to August 2022. B-mode ultrasound, color Doppler, and CEUS were performed to evaluate the lesions. The integration of CEUS led to the reclassification of BI-RADS 4A lesions into three pathways: upgrade to 4B, downgrade to 3, or no change in category. The diagnostic performance of CEUS was evaluated, and the contrast modes which could be used to correctly downgrade or upgrade BI-RADS 4A lesions were explored. RESULTS:A total of 1340 lesions were finally included in the analysis of CEUS performance. The average age of all included participants was 43 ± 8 (range from 20 to 84). The average diameter of the lesions was 12.4 ± 3.8 mm (range from 5 mm to 20 mm). The diagnostic performances of CEUS + BI-RADS were 98.3% for sensitivity, 81.7% for specificity, 34.1% for positive predictive value, 99.8% for negative predictive value, and 83.1% for overall accuracy, respectively. The sensitivity was slightly downgraded (98.3% vs. 100%, p = 0.1555), while the specificity and accuracy were greatly elevated (p < 0.05). The receiver operating characteristic curve (ROC) value was calculated using a binary classification threshold, with BI-RADS 3 as low-risk and BI-RADS 4A as high-risk. The area under ROC was 0.896 (95% CI: 0.878, 0.911) for BI-RADS + CEUS. The decision curve analysis showed that using CEUS + BI-RADS to guide biopsy decisions provided net benefit compared to the default strategy of biopsying all lesions. CONCLUSION:The addition of CEUS allows for the correct downgrading of most benign BI-RADS 4A lesions, thereby avoiding unnecessary biopsies. An upgrade from BI-RADS 4A to 4B on CEUS warrants immediate biopsy to ensure prompt diagnosis and treatment. TRIAL REGISTRATION:Chinese Clinical Trial Registry: ChiCTR2100050719.

IntroductionThis study aims to develop and validate machine learning models that integrate multimodal features (BI-RADS terminology, ultrasound imaging, and radiomics) to improve breast mass malignancy risk stratification and compare their diagnostic performance across different BI-RADS categories.MethodsThis retrospective cohort study analyzed data from 2, 685 patients with 3, 703 ultrasound images collected from July 2019 to March 2024 at a single medical center. Patients included women with complete ultrasound images and clear pathological diagnoses. The dataset comprised 2, 069 benign cases (2, 762 images) and 616 malignant cases (941 images), randomly divided into training (n=2, 979 images) and validation (n=724 images) sets. Primary outcomes were diagnostic accuracy and area under the receiver operating characteristic curve (AUC) for distinguishing malignant from benign breast masses. Three machine learning models (Logistic Regression, Support Vector Machine, and Random Forest) were trained using BI-RADS terminology features, ultrasound quantitative features, radiomics features, and combined multimodal features. Performance was evaluated both overall and within specific BI-RADS subcategories (2, 3, 4a, 4b, 4c, and 5).ResultsAmong 2, 685 patients, the Random Forest model using combined multimodal features achieved the highest overall performance with an AUC of 0.850 (95% CI 0.810- 0.875). For single-modality approaches, Logistic Regression performed best with BI-RADS terminology features, with an AUC of 0.820 (95% CI, 0.775-0.856), and radiomics features, with an AUC of 0.740 (95% CI, 0.706-0.780); while Random Forest was optimal for ultrasound imaging features, with an AUC of 0.800 (95% CI, 0.768-0.839). Subgroup analysis revealed excellent performance for BI-RADS categories 2 (AUC, 1.000-1.000) and 3 (AUC, 0.947-0.957), acceptable performance for 5 (AUC, 0.813-0.870) and 4a (AUC, 0.800-0.867), but poor performance for categories 4b (AUC, 0.649-0.709), 4c (AUC, 0.551-0.623).DiscussionThis study demonstrates that machine learning models integrating multimodal ultrasound features can effectively stratify breast mass malignancy risk, with the Random Forest model using combined features showing superior performance. The approach shows particular strength in BI-RADS categories 2, 3, 5 and 4a, suggesting potential clinical utility for reducing unnecessary biopsies and improving diagnostic confidence. However, performance limitations in higher-risk categories (4b, 4c) indicate need for further model refinement and multicenter validation before clinical implementation.

[This corrects the article DOI: 10.3389/fonc.2026.1782135.].

Objectives This study aimed to develop and validate an ultrasound‐based IHC4‐associated radiomic model for predicting late recurrence in ER‐positive breast cancer and to evaluate its potential to support risk stratification and guide decisions on extended endocrine therapy. Methods In this retrospective multicenter study, patients were divided into a training cohort, an internal validation cohort, and two external validation cohorts. Radiomic features associated with the immunohistochemical four‐marker (IHC4) score were selected to construct a support vector machine (SVM)‐based IHC4‐associated radiomic model for generating an IHC4‐associated radiomic score. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC). Late distant recurrence (DR) was defined as the primary endpoint, and progression‐free survival (PFS) was defined as the secondary endpoint. The prognostic value of the score was further evaluated using Cox regression analysis. Results A total of 523 patients were included in this study. Seven IHC4‐associated radiomic features were selected to construct the IHC4‐associated radiomic model. The model demonstrated consistent performance across cohorts, with AUCs of 0.81, 0.84, 0.79, and 0.81 in the training, internal validation, and two external validation cohorts, respectively. The IHC4‐associated radiomic score stratified late DR risk across the training, internal validation, and external validation cohorts. In the secondary exploratory PFS analysis, the score remained associated with PFS in multivariable Cox analysis (HR = 4.246, 95% CI: 1.749–10.307, P = .001). Conclusions The IHC4‐associated radiomic model may serve as a noninvasive biomarker for predicting prognosis in ER‐positive breast cancer.

Hormone receptor-positive/human epidermal growth factor receptor 2-negative (HR+/HER2-) early breast cancer (EBC) patients face long-term recurrence risk despite standard treatment. Current prognostic tools relying on clinicopathological factors or multigene assays have limited accuracy or accessibility. In this study, we developed a multimodal recurrence risk prediction (MRRP) model integrating routinely available clinical data, including whole-slide images (WISs), ultrasound (US) imaging and diagnostic reports, and structured clinical parameters. The MRRP model employs a hierarchical transformer-based fusion framework with innovative intra- and intermodality cross-attention mechanisms to dynamically integrate diverse feature representations. Using a well-curated cohort of 768 HR+/HER2- EBC patients with long-term follow-up, MRRP demonstrated superior prognostic performance (C-index = 0.840) compared to single-modality models, with robust time-dependent AUCs exceeding 0.85 at 3, 5, and 7 years. Ablation studies highlighted the central role of pathology features and the complementary value of US and clinical data. We further validated the optimal query selection strategies and evaluated different pretrained encoders, revealing complex modality interactions. To address real-world challenges of missing modality data, a learnable compensation mechanism was implemented, improving model robustness. Our study provides a clinically practical, AI-driven tool for precise risk stratification in HR+/HER2- EBC patients, facilitating individualized treatment and surveillance decisions without reliance on costly multi-omics data.

Background:Enlargement of cervical lymph nodes (CLNs) is a common clinical response to lesions in the neck as well as in other parts of the body. Accurate qualitative diagnosis of lymph nodes can provide important reference information for clinical decision-making. While histopathological diagnosis remains the gold standard for differentiating benign from malignant CLNs, it is an invasive procedure. Ultrasonography serves as a non-invasive imaging modality widely employed in clinical practice for the preoperative evaluation and qualitative assessment of CLNs; however, its diagnostic accuracy is operator-dependent. In this study, we investigated ultrasound image features between benign and malignant CLNs and developed deep learning (DL) models for the qualitative diagnosis of CLNs. Methods:Patients with pathologically confirmed CLNs via ultrasound-guided biopsy from January 2020 to December 2023 were retrospectively included. The gold standard was histopathological diagnosis. Ultrasound features of CLNs were documented, and their value in differentiating benign from malignant CLNs was assessed using univariate analysis. DL models were developed to qualitatively diagnose the benign and malignant CLNs. Model performance was evaluated using receiver operating characteristic curves, accuracy curves, recall curves, and loss curves. Results:A total of 3,014 CLNs from 2,697 patients were included in this study, with 1,489 classified as benign cases and 1,525 as malignant cases. Almost all DL models demonstrated satisfactory performance in qualitative diagnosis of CLNs, achieving area under the curve (AUC) values ranging from 0.56 to 0.81, with the VGG16 model exhibiting the best performance with an AUC of 0.81 [95% confidence interval (CI): 0.77-0.86], accuracy of 0.73, sensitivity of 0.71, and specificity of 0.74. In comparison to ultrasonography, the VGG16, ResNet101, and ResNet50 models showed significantly superior predictive performance (P<0.05). Conclusions:DL models utilizing ultrasound images demonstrated promising performance in the qualitative diagnosis of CLNs. This approach enhanced the diagnostic accuracy of preoperative ultrasound assessment, thereby allowing a subset of patients to avoid unnecessary biopsies and optimizing clinical decision-making.

Breast screening reduces cancer-specific mortality but can also precipitate avoidable harms through over-detection of benign abnormalities and subsequent over-surveillance. Across mammography and digital breast tomosynthesis (DBT), ultrasound and magnetic resonance imaging (MRI), gains in sensitivity are often offset by reduced specificity, driving false-positive recalls, benign-biopsy burden and resource strain. Within breast imaging reporting and data system (BI-RADS)–guided decision-making, Category 3 and Category 4A trigger short-interval follow-up or biopsy despite low event rates, amplifying anxiety and cost. Artificial intelligence (AI) offers a practical route to mitigate these drawbacks. Prospective and real-world studies indicate that AI-assisted reading can maintain or improve cancer detection while lowering recall rates and workload. AI models also support finer risk stratification—particularly for BI-RADS 4 lesions—thereby reducing unnecessary interventions. This review synthesises evidence on the performance and limitations of mainstream screening technologies, delineates the multidimensional impact of over-detection, and evaluates the capacity of AI to rebalance sensitivity and specificity, optimise follow-up intervals and support risk-adapted workflows. A patient-centred, evidence-driven strategy that integrates validated AI with clearly defined decision thresholds and effective patient-provider communication can maximise benefit while minimising harm. This review critically evaluates the causes and consequences of over-detection and over-surveillance in breast cancer screening and highlights how AI can advance radiologic decision-making through improved lesion stratification and more efficient, personalised follow-up strategies.

BackgroundFor breast cancer, developing non-invasive methods to accurately predict axillary lymph node (ALN) status before surgery has become a general trend. This study aimed to develop and evaluate a nomogram to predict the probability of ALN metastasis (ALNM) preoperatively based on clinicopathological and ultrasonography (US) features.MethodsPatients diagnosed with breast cancer by preoperative histopathologic biopsy in West China Hospital from 1 August, 2022 to 31 January, 2024 and undergoing surgical treatment with preoperative US in West China Hospital were prospectively included. Preoperative clinicopathological and US features, along with postoperative pathological ALN status, were collected. Patients included were randomly divided into a training set and a test set (7:3). In the training cohort, the independent predictors of ALNM were obtained by univariate and multivariate binary logistic regression analyses and were used to develop a binary logistic regression model presented as a nomogram. Model performance was evaluated by receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis (DCA).ResultsA total of 610 patients were included for analysis: 427 in the training set and 183 in the test set. Molecular subtypes, tumor infiltration of the subcutaneous layer, tumor infiltration of the retromammary space, lymph node (LN) short axis, LN long/short (L/S) axis ratio, LN corticomedullary demarcation, and LN cortical thickness evenness were independent predictors of ALNM. The nomogram showed good discrimination with an area under the ROC curve (AUC) of 0.854 for the training set and 0.822 for the test set, presented good agreement between predicted and observed probabilities, and acquired net benefit across a wide threshold range.ConclusionsThe nomogram demonstrated strong discrimination, calibration, and clinical net benefit to assist clinical decisions.

This study aimed to evaluate the predictive value of implementing machine learning models based on ultrasound radiomics and clinicopathological features in the survival analysis of triple-negative breast cancer (TNBC) patients. All patients, including retrospective cohort (training cohort, n = 306; internal validation cohort, n = 77) and prospective external validation cohort (n = 82), were diagnosed as locoregional TNBC and underwent pre-intervention sonographic evaluation in this multi-center study. A thorough chart review was conducted for each patient to collect clinicopathological and sonographic features, and ultrasound radiomics features were obtained by PyRadiomics. Deep learning algorithms were utilized to delineate ROIs on ultrasound images. Radiomics analysis pipeline modules were developed for analyzing features. Radiomic scores, clinical scores, and combined nomograms were analyzed to predict 2-year, 3-year, and 5-year overall survival (OS) and disease-free survival (DFS). Receiver operating characteristic (ROC) curves, calibration curves, and decision curves were used to evaluate the prediction performance. Both clinical and radiomic scores showed good performance for overall survival and disease-free survival prediction in internal (median AUC of 0.82 and 0.72 respectively) and external validation (median AUC of 0.70 and 0.74 respectively). The combined nomograms had AUCs of 0.80–0.93 and 0.73–0.89 in the internal and external validation, which had best predictive performance in all tasks (p < 0.05), especially for 5-year OS (p < 0.01). For the overall evaluation of six tasks, combined models obtained better performance than clinical and radiomic scores [AUCs of 0.83 (0.73,0.93), 0.81 (0.72,0.93), and 0.70 (0.61,0.85) respectively]. The combined nomograms based on pre-intervention ultrasound radiomics and clinicopathological features demonstrated exemplary performance in survival analysis. The new models may allow us to non-invasively classify TNBC patients with various disease outcome.

Kidney transplant recipients (KTRs) carry an elevated risk of cancer-related mortality. The cumulative incidence of de novo post-transplant malignancy (DPTM) reaches 10% at 10 years, with renal cell carcinoma (RCC) arising in native kidneys being the predominant urologic malignancy. This study presents three KTRs who developed native kidney RCC 6-15 years post-transplantation. Notably, Case 1 demonstrated a 14.7 cm mass at diagnosis, secondary to non-adherence to protocol-based native kidney surveillance. Histopathological confirmation of RCC was established in all cases through ISUP/WHO-graded surgical specimens and immunophenotypic profiling. KTRs exhibit elevated native kidney RCC risk, often with nonspecific clinical presentations. Our findings emphasize the critical role of systematic imaging protocols, particularly ultrasonography and contrast-enhanced ultrasound (CEUS), in early tumor detection. Implementing these strategies may improve survival and reduce disease burden in this high-risk population.

This study aimed to develop an early predictive model for neoadjuvant therapy (NAT) response in breast cancer by integrating multimodal ultrasound (conventional B-mode, shear-wave elastography, and contrast-enhanced ultrasound) and radiomics with clinical-pathological data, and to evaluate its predictive accuracy after two cycles of NAT. This retrospective study included 239 breast cancer patients receiving neoadjuvant therapy, divided into training (n = 167) and validation (n = 72) cohorts. Multimodal ultrasound—B-mode, shear-wave elastography (SWE), and contrast-enhanced ultrasound (CEUS)—was performed at baseline and after two cycles. Tumors were segmented using a U-Net-based deep learning model with radiologist adjustment, and radiomic features were extracted via PyRadiomics. Candidate variables were screened using univariate analysis and multicollinearity checks, followed by LASSO and stepwise logistic regression to build three models: a clinical-ultrasound model, a radiomics-only model, and a combined model. Model performance for early response prediction was assessed using ROC analysis. In the training cohort (n = 167), Model_Clinic achieved an AUC of 0.85, with HER2 positivity, maximum tumor stiffness (Emax), stiffness heterogeneity (Estd), and the CEUS “radiation sign” emerging as independent predictors (all P < 0.05). The radiomics model showed moderate performance at baseline (AUC 0.69) but improved after two cycles (AUC 0.83), and a model using radiomic feature changes achieved an AUC of 0.79. Model_Combined demonstrated the best performance with a training AUC of 0.91 (sensitivity 89.4

ObjectiveThis study aimed to develop a deep learning system to identify and differentiate the metastatic cervical lymph nodes (CLNs) of thyroid cancer.MethodsFrom January 2014 to December 2020, 3059 consecutive patients with suspected with metastatic CLNs of thyroid cancer were retrospectively enrolled in this study. All CLNs were confirmed by fine needle aspiration. The patients were randomly divided into the training (1228 benign and 1284 metastatic CLNs) and test (307 benign and 240 metastatic CLNs) groups. Grayscale ultrasonic images were used to develop and test the performance of the Y-Net deep learning model. We used the Y-Net network model to segment and differentiate the lymph nodes. The Dice coefficient was used to evaluate the segmentation efficiency. Sensitivity, specificity, accuracy, positive predictive value (PPV), and negative predictive value (NPV) were used to evaluate the classification efficiency.ResultsIn the test set, the median Dice coefficient was 0.832. The sensitivity, specificity, accuracy, PPV, and NPV were 57.25%, 87.08%, 72.03%, 81.87%, and 66.67%, respectively. We also used the Y-Net classified branch to evaluate the classification efficiency of the LNs ultrasonic images. The classification branch model had sensitivity, specificity, accuracy, PPV, and NPV of 84.78%, 80.23%, 82.45%, 79.35%, and 85.61%, respectively. For the original ultrasonic reports, the sensitivity, specificity, accuracy, PPV, and NPV were 95.14%, 34.3%, 64.66%, 59.02%, 87.71%, respectively. The Y-Net model yielded better accuracy than the original ultrasonic reports.ConclusionThe Y-Net model can be useful in assisting sonographers to improve the accuracy of the classification of ultrasound images of metastatic CLNs.

Objective:This study is focused on ultrasound multimodality examination, which refers to the combined use of three ultrasound examination modalities, ultrasound (US), acoustic radiation force impulse (ARFI) imaging, and contrast-enhanced ultrasound (CEUS). The purpose of this study is to analyze the value of applying ultrasound multimodality examination in the differential diagnosis of benign and malignant breast non-mass-like lesions (NMLs). Methods:Cases of breast NMLs were analyzed retrospectively, and the nature of all the lesions was verified by pathological examination. Based on the gray-scale ultrasound image characteristics, the cases were classified into types Ⅰ to Ⅴ, and type Ⅰ and type Ⅱ were further classified into 4 subtypes, Ⅰa, Ⅰb, Ⅱa, and Ⅱb, according to whether there was also calcification, and the proportion of malignant cases in each subtype was statistically analyzed. Logistic regression models of US, US+ARFI, US+CEUS, and US+ARFI+CEUS for the diagnosis of malignant cases were established, ROC curves were drawn, the area under the curve (AUC) was calculated, and comparisons were made accordingly. The detection rate of malignant NMLs without calcification (atypical malignant NMLs) by the combination examination of US, ARFI, and CEUS was analyzed. Results:A total of 407 cases were included in the study. All subjects were female, aged 22 to 81 years, with the average age being (47.0±11.0) years. There were 220 benign cases and 187 malignant cases. Ranked from the highest to the lowest, the malignancy proportion of the different types was Ⅰb>Ⅱb>Ⅲ>Ⅴ>Ⅰa>Ⅱa>Ⅳ. The malignant proportion of the low echo area with calcification was significantly higher than that of the lesions without calcification. The AUC (95% confidence interval [CI]) for diagnosing malignant cases with the logistic regression models of US, US+ARFI, US+CEUS, and US+ARFI+CEUS were 0.895 (0.862-0.927), 0.908 (0.878-0.937), 0.921 (0.893-0.948), and 0.927 (0.902-0.952), respectively. Comparison of the AUC of the 4 regression models showed significant differences (P<0.001). The detection rate of US for NMLs without calcification was 80.7%. When US was used in combination with ARFI and CEUS, 86.4% of the malignant NMLs lesions without calcification could be detected if the lesion CEUS score was 4 or 5 points or if shear-wave velocity (SWV)≥4.28 m/s. Conclusion:Breast NMLs with calcification show high risks of malignancy, and a pathological examination is always recommended for a conclusive diagnosis. Ultrasound multimodality examination can improve the diagnostic accuracy of breast NML without calcification.

Background: Low nuclear grade ductal carcinoma in situ (DCIS) patients can adopt proactive management strategies to avoid unnecessary surgical resection. Different personalized treatment modalities may be selected based on the expression status of molecular markers, which is also predictive of different outcomes and risks of recurrence. DCIS ultrasound findings are mostly non mass lesions, making it difficult to determine boundaries. Currently, studies have shown that models based on deep learning radiomics (DLR) have advantages in automatic recognition of tumor contours. Machine learning models based on clinical imaging features can explain the importance of imaging features. Methods: The available ultrasound data of 349 patients with pure DCIS confirmed by surgical pathology [54 low nuclear grade, 175 positive estrogen receptor (ER+), 163 positive progesterone receptor (PR+), and 81 positive human epidermal growth factor receptor 2 (HER2+)] were collected. Radiologists extracted ultrasonographic features of DCIS lesions based on the 5th Edition of Breast Imaging Reporting and Data System (BI-RADS). Patient age and BI-RADS characteristics were used to construct clinical machine learning (CML) models. The RadImageNet pretrained network was used for extracting radiomics features and as an input for DLR modeling. For training and validation datasets, 80% and 20% of the data, respectively, were used. Logistic regression (LR), support vector machine (SVM), random forest (RF), and eXtreme Gradient Boosting (XGBoost) algorithms were performed and compared for the final classification modeling. Each task used the area under the receiver operating characteristic curve (AUC) to evaluate the effectiveness of DLR and CML models. Results: In the training dataset, low nuclear grade, ER+, PR+, and HER2+ DCIS lesions accounted for 19.20%, 65.12%, 61.21%, and 30.19%, respectively; the validation set, they consisted of 19.30%, 62.50%, 57.14%, and 30.91%, respectively. In the DLR models we developed, the best AUC values for identifying features were 0.633 for identifying low nuclear grade, completed by the XGBoost Classifier of ResNet50; 0.618 for identifying ER, completed by the RF Classifier of InceptionV3; 0.755 for identifying PR, completed by the XGBoost Classifier of InceptionV3; and 0.713 for identifying HER2, completed by the LR Classifier of ResNet50. The CML models had better performance than DLR in predicting low nuclear grade, ER+, PR+, and HER2+ DCIS lesions. The best AUC values by classification were as follows: for low nuclear grade by RF classification, AUC: 0.719; for ER+ by XGBoost classification, AUC: 0.761; for PR+ by XGBoost classification, AUC: 0.780; and for HER2+ by RF classification, AUC: 0.723. Conclusions: Based on small-scale datasets, our study showed that the DLR models developed using RadImageNet pretrained network and CML models may help predict low nuclear grade, ER+, PR+, and HER2+ DCIS lesions so that patients benefit from hierarchical and personalized treatment

Background: Using meta-analysis to evaluate the diagnostic value of contrast-enhanced ultrasound (CEUS) in the diagnosis of papillary thyroid microcarcinoma (PTMC). Methods: For this systematic review and meta-analysis, we searched PubMed, Cochrane Library, Web of Science, WanFang Data, VPCS Data, and China National Knowledge Infrastructure electronic databases for diagnostic studies on PTMC by CEUS from January 2013 to November 2022. Data were not available or incomplete such as case reports, nonhuman studies, etc, were excluded. Random-effects meta-analyses were used to evaluate the diagnostic accuracy of CEUS in diagnosing PTMC. The quality of the evidence was assessed with the QUADAS-2 scale. This study is registered on PROSPERO, number CRD42023409417. Results: Of 1064 records identified, 33 were eligible. The results showed that the pooled sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, and diagnostic odds ratio of CEUS in diagnosing PTMC were 0.84 (95% confidence interval [CI] = 0.83–0.86), 0.82 (95% CI = 0.80–0.83), 3.90 (95% CI = 3.23–4.72), 0.21 (95% CI = 0.18–0.25), and 20.01 (95% CI = 14.97–26.74), respectively, and the area under the summary receiver operating characteristic curve was 0.8930 (the Q index was 0.8239). The Deek funnel plot indicated publication bias (P ˂.01). Conclusion: This meta-analysis provides an overview of diagnostic accuracy of CEUS in diagnosing PTMC which indicates CEUS has a good diagnostic value for PTMC. The limitations of this study are publication bias and strong geographical bias.