As a non-invasive treatment method, sonodynamic therapy (SDT) has been proven to enhance immune system activity by inducing the generation of reactive oxygen species (ROS). However, the efficacy is constrained by the hypoxic tumor microenvironment (TME) and their intrinsic antioxidant defense mechanisms. To overcome these challenges, we have developed a manganese-based nanoplatform named HMAS@MH, which is responsive to both ultrasound and TME. This nanoplatform utilizes hollow mesoporous manganese dioxide (HM) as a carrier, loaded with oxygen-independent alkyl radical generator AIPH and siNrf2, and coated with hyaluronic acid (HA)-modified erythrocyte membrane (eM). After systemic administration, HMAS@MH preferentially accumulates in tumor regions, benefiting from the extended circulation time afforded by the eM camouflage and the active targeting ability conferred by HA. Under dual stimulation from exogenous ultrasound irradiation and endogenous conditions of low pH and high glutathione (GSH) levels, HMAS@MH decomposes to generate alkyl and hydroxyl radicals, resulting in the accumulation of cytoplasmic double-stranded DNA (dsDNA). Additionally, siNrf2 and Mn2+ significantly blocks the neutralization of free radicals by inhibiting the antioxidant defense system and depleting GSH. As a non-cyclic dinucleotide stimulator of interferon genes (STING) agonist, Mn2+ significantly enhances the sensitivity of cyclic GMP-AMP synthase (cGAS) protein to dsDNA, thereby activating the cGAS-STING pathway and triggering robust anti-tumor immune responses. In conclusion, this sonochemical therapy, combined with Nrf2 silencing, amplifies oxidative damage in tumor cells and boosts anti-tumor immunity, resulting in significant tumor suppression.

Objective: To investigate the diagnostic concordance between CEUS LI-RADS and CT/MR LI-RADS for focal intrahepatic lesions in patients at high risk of hepatocellular carcinoma. Methods: Retrospective analysis of imaging data of patients at high risk of HCC attending our hospital from January 2018 to January 2021, 165 lesions were classified according to CEUS LI-RADS and CT/MR LI-RADS, respectively, using puncture biopsy or surgical pathology as the criteria. The kappa concordance test was used to evaluate the classification results of intrahepatic focal lesions in high-risk groups. Results: The concordance between CEUS and CT/MRI LI-RADS for the classification of intrahepatic focal lesions in the high-risk group of HCC was statistically different (P < 0.001), and the agreement between the two imaging methods was general (kappa = 0.345, P < 0.001). The positive predictive value of CEUS LR-5 for HCC was 95.4%, and the positive predictive value of CT/MRI LR-5 had a positive predictive value of 94.6% for HCC, and the difference between CEUS and CT/MRI LR-5 for HCC was not statistically significant. The positive predictive value of CEUS LR-M for non-HCC malignancies was 34.1%, and the positive predictive value of CT/MRI LR-M for non-HCC malignancies was 76.5%, and the difference between CEUS and CT/MRI LR-M had a statistically significant difference in the positive predictive value for non-HCC malignancies (P < 0.05). Conclusions: The classification results of CEUS and CT/MRI LI-RADS for focal intrahepatic lesions in a high-risk group of HCC were in general agreement, but both had high diagnostic performance, and both had high specificity and positive predictive value using LR-5 as the diagnostic imaging standard for HCC, while the differential diagnostic value of LR-M for non-HCC malignant tumors and HCC still needs to be further investigated.

Background and aim:Deep learning has improved medical image analysis but often produces opaque decisions and correlation-driven predictions that may diverge from clinical reasoning. We hypothesize that a physiology-informed hybrid framework, which explicitly models placenta-pulmonary interactions and integrates multimodal data, could provide interpretable and reliable guidance for assessing fetal lung maturity (FLM) and optimizing antenatal glucocorticoids (GCs). Materials and methods:In a prospective cohort study involving 320 pregnancies-including 160 with hypertensive disorders of pregnancy (HDP)-each with weekly acquisitions from 28 to 36 weeks, we combined 2D/3D ultrasound, shear-wave elastography, Doppler, and maternal plasma metabolomics. A biophysical placenta-pulmonary coupling model used the umbilical artery pulsatility index (PI) and a metabolomic hypoxia-steroid score to represent placental reserve, while backscatter integrals and elastography were used to characterize fetal lung properties. Constrained by this model, a dual-branch network was developed: (i) a cross-modal attention Transformer with self-supervised contrastive learning to generate unsupervised FLM stages from fused representations and (ii) a spatiotemporal convolution-LSTM network to predict individualized GC dosing and the optimal administration window. A composite loss penalized both projected respiratory distress syndrome (RDS) risk and the biomarker-derived neurotoxicity index. Results:The cross-modal representations clustered into four distinct maturity stages matching biochemical benchmarks, with an inter-stage silhouette score of 0.72. A downstream classifier achieved 92.3% accuracy in discriminating early from late maturity. The dosing branch predicted the GC dose within ±0.5 mg of clinically prescribed regimens and reduced projected RDS risk by 27% compared to standard dosing, while maintaining the biomarker-derived neurotoxicity index below the prespecified threshold. Conclusion:A mechanism-guided, multimodal AI framework constrained by placenta-pulmonary physiology transforms imaging features into traceable decision pathways that align with clinical cognition. This interpretable framework may enable non-invasive FLM staging and individualized GC therapy, providing hypothesis-generating decision support that warrants external validation and prospective trials.

Patients with Hashimoto's thyroiditis (HT) frequently present with concurrent nodular lesions such as nodular goiter and thyroid cancer (especially papillary thyroid carcinoma, PTC), and their risk of PTC is significantly higher than that of non-HT individuals. Patients with HT exhibit varying degrees of glandular fibrosis, leading to some areas having a "nodular-like" appearance. Some of these nodules may display features suggestive of malignancy, which can be difficult to distinguish using conventional ultrasound. Multimodal ultrasound combined with fine-needle aspiration biopsy (FNAB) has emerged as the most effective method currently for identifying the benign or malignant nature of nodules in the context of HT, owing to its advantages of cost-effectiveness, convenience, and reproducibility. Artificial intelligence (AI) has been increasingly applied in the medical field, but its accuracy in diagnosing HT-associated thyroid nodules (TNs) still requires further refinement. This article reviews the progress of multimodal ultrasound technology combined with AI in assessing and diagnosing the benign and malignant nature of TNs in patients with HT.

Parathyroid carcinoma (PC) is a rare endocrine malignancy characterized by uncontrolled secretion of parathyroid hormone (PTH). This article presents a case of a giant PC that was entirely located within the thyroid gland and was initially misdiagnosed as a thyroid nodule. Our case is noteworthy for the absence of clinical symptoms and the markedly elevated serum calcium and PTH levels. We provide a detailed description of the patient's imaging findings and systematically review the diagnostic and therapeutic modalities of PC.

Follicular thyroid adenoma and follicular thyroid carcinoma are highly similar in terms of cell morphology. As preoperative differentiation relies on histopathology, a large number of patients with benign nodules undergo unnecessary surgery. This narrative review summarizes the value and limitations of ultrasonography in distinguishing between these two conditions. Evidence suggests that follicular thyroid adenoma often presents as well-defined, isoechoic nodules with predominantly peripheral vascularity, while follicular thyroid carcinoma often exhibits a taller-than-wide shape, irregular margins, disordered internal vascularity, and sonographic signs suggestive of capsular interruption or vascular invasion. However, there is considerable overlap between the two in terms of echogenicity, calcification, and vascularity parameters, and the traditional thyroid imaging reporting and data system has limited discriminatory efficacy for follicular tumors (area under the curve: 0.6-0.75). Emerging techniques such as contrast-enhanced ultrasound, elastography, and artificial intelligence have demonstrated improved diagnostic accuracy (reportedly up to 91.2%); however, challenges remain, including limited sample size and poor model interpretability. Future research should focus on multimodal fusion models, ultrasound omics, and molecular marker integration to achieve noninvasive and precise preoperative classification. This review was guided by the Scale for the Assessment of Narrative Review Articles.

Objective: To construct and validate an interpretable machine learning model based on ultrasound radiomics for the preoperative prediction of human epidermal growth factor receptor 2 (HER2) expression status in patients with bladder cancer. Methods: This retrospective study included 270 patients with pathologically confirmed bladder cancer, all of whom underwent conventional ultrasound examination preoperatively. Radiomic features were extracted from ultrasound images and subjected to feature selection using the intraclass correlation coefficient (ICC), Spearman correlation analysis, and least absolute shrinkage and selection operator (LASSO) regression. Four machine learning algorithms, namely, Light Gradient Boosting Machine (LightGBM), Random Forest (RF), eXtreme Gradient Boosting (XGBoost), and K-Nearest Neighbors (KNN), were employed to construct predictive models. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity. The SHAP (SHapley Additive exPlanations) method was applied for interpretability analysis of the optimal model. Results: A total of 1,288 radiomic features were extracted, of which 5 optimal features were retained for model construction after feature selection. In the test set, the XGBoost model achieved the best performance, with an AUC of 0.817 (95% CI: 0.634–0.999). Conclusion: The interpretable machine learning model based on ultrasound radiomics can preoperatively and noninvasively predict HER2 expression status in bladder cancer, providing valuable reference information for clinical decision-making in individualized ADC-targeted therapy strategies.

Objectives To evaluate whether a multiparametric approach combining grayscale ultrasound (2D‐US), contrast‐enhanced ultrasound (CEUS), and serum anti‐thyroid peroxidase antibody (TPO‐Ab) improves the differentiation of benign from malignant thyroid nodules (TNs) in patients with Hashimoto's thyroiditis (HT) and to assess the stability of the combined diagnostic model through internal validation. Methods This retrospective study enrolled 600 HT patients with 650 pathologically confirmed TNs. All patients underwent 2D‐US; CEUS was performed in 475, TPO‐Ab testing in 452, and shear wave elastography (SWE) in an exploratory subset of 88 patients. The primary combined diagnostic model was developed using logistic regression (LR) on the subset of patients with complete 2D‐US, CEUS, and TPO‐Ab data. Internal validation was performed via stratified 5‐fold cross‐validation. Six machine learning classifiers were additionally compared to assess model robustness. Results Among single modalities, 2D‐US showed the highest cross‐validated area under the curve (AUC) (LR: 0.831). CEUS and TPO‐Ab alone yielded limited discrimination (AUC 0.671 and 0.621, respectively). The combined model integrating 2D‐US, CEUS, and TPO‐Ab achieved a cross‐validated AUC of 0.849 with sensitivity of 77.3% and specificity of 77.8%, outperforming any single modality. Among the 6 classifiers evaluated, LR demonstrated the most stable performance and the best balance between sensitivity and specificity. Conclusion Integrating multimodal ultrasound with serum TPO‐Ab significantly improves the differentiation of TNs in HT, with the combined LR model demonstrating robust performance after internal validation. This multiparametric approach may serve as a supplementary tool for risk stratification in surgical candidates, but prospective external validation is required before clinical implementation.

To assess the malignancy risk of peripheral lung lesions with non-diagnostic percutaneous transthoracic needle biopsy (PTNB) findings and to evaluate contrast-enhanced ultrasound (CEUS) guidance for reducing non-diagnostic outcomes. This retrospective study included patients who underwent PTNB at three institutions between October 2018 and March 2023. PTNB results were categorized as diagnostic (malignancy or specific benignity) or non-diagnostic (nonspecific benignity, atypical epithelia, insufficient samples, or sampling failure). The risk factors for malignancy and the effectiveness of CEUS for guiding PTNB were analyzed. This study included 1591 peripheral lung lesions from 1558 patients (median age: 63 y; range: 18–87 y). Among these patients, 1072 were male, and 486 were female. Approximately 40

Abstract Background Hepatic alveolar echinococcosis (HAE) is associated with significant disability and mortality. After infection, there is a prolonged asymptomatic latent period, and most patients present at a late stage when they seek medical treatment. The radiological features of HAE can resemble those of liver malignancies, posing diagnostic challenges. Case summary A 52-year-old male was admitted to our hospital with discomfort and pain in the upper right abdomen persisting for more than six months without an apparent trigger. Abdominal ultrasound revealed a heterogeneous mass in the right posterior lobe of the liver, characterised by an irregular shape and indistinct boundaries. Enhanced ultrasound demostrated slight hyper-enhancement of the peripheral area during the arterial phase and hypo-enhancement in the delayed phase, while the central area showed no enhancement. CT and MRI scans revealed a large abnormal lesion in the right posterior lobe and porta hepatis of the liver, accompanied by marked dilation of the intrahepatic bile ducts and multiple enlarged lymph nodes at the porta hepatis. The lesion’s edges exhibited mild enhancement in the delayed phase, whereas the centre showed no significant enhancement. Serum CA199 levels were markedly elevated, exceeding the upper reference limit by more than sevenfold. The patient’s serum rEm18-ELISA test for alveolar cyst was negative. The pathological diagnosis was obtained via ultrasound-guided liver biopsy. The initial biopsy showed no clear evidence of tumour lesions. A repeat biopsy revealed hepatic alveolar echinococcosis, with Candida infection in the necrotic tissue. Four months later, the patient underwent surgical treatment at another hospital. Pathological examination confirmed hepatic alveolar echinococcosis, and the patient made a good recovery. Conclusion Hepatic alveolar echinococcosis can closely mimic perihilar cholangiocarcinoma, exhibiting almost identical imaging features. Initial biopsies may yield false-negative results due to sampling necrotic tissue. Therefore, it is essential to consider the patient’s epidemiological background, maintain a high index of suspicion, and perform repeat biopsies to ensure an accurate diagnosis. Furthermore, it is crucial to inhance heathcare professionals’ proficiency in both clinical and technical aspects, promote effective interdepartmental communication, and reduce the incidence of misdiagnosis.

Neurovascular coupling contributes to the regulation of cognitive function in Parkinson’s disease (PD). However, the correlation between structural and functional integrity of neurovascular unit in substantia nigra and cognitive impairment in PD remains poorly understood. Using super-resolution ultrasound imaging (SRUS), this study aimed to explore the interplay among dopaminergic neurons, microcirculation, and cognitive function in 6-hydroxydopamine-induced PD rat model. The model exhibited impaired motor and cognitive abilities, along with substantia nigra hyperechogenicity detected via transcranial sonography. SRUS revealed reduced microvascular density, complexity, and velocity, alongside increased vessel tortuosity in the substantia nigra. These changes were accompanied by elevated expression of MMP9, CD4, Iba 1, and GFAP, and decreased levels of TH, GLUT-1 and Laminin. Levodopa treatment prevented dopaminergic neurons degeneration, reduced substantia nigra hyperechogenicity, restored microvascular structure and function, as well as alleviated neuroinflammation, which may contribute to improved cognitive performance. These findings suggest that SRUS can effectively detect microvascular alterations in the substantia nigra in a 6-hydroxydopamine-induced PD rat model. Dysfunction neurovascular coupling, likely mediated by dopaminergic neuronal injury, may play a role in PD-related cognitive impairment.

The American College of Radiology Breast Imaging Reporting and Data System (ACR BI-RADS) 6th Edition’s updated definition of mixed solid and cystic breast lesions (MSCBLs) reflects the diagnostic challenges in assessing these lesions. This study aimed to develop and validate a scoring system based on ultrasound (US) features and clinical factors to differentiate benign from malignant MSCBLs. This retrospective multicenter study included 499 MSCBLs from three medical centers, which were divided into a training cohort (n = 396, 79.4

Metanephric adenoma (MA) is a rare primary renal epithelial tumor classified as a metanephric neoplasm. Its clinical symptoms and imaging features are non-specific, making preoperative diagnosis challenging and often leading to misdiagnosis as renal cell carcinoma (RCC). This report details the contrast-enhanced ultrasound (CEUS) findings of a 54-year-old male patient with a MA: 15 s after contrast agent injection, the lesion showed heterogeneous hypoenhancement, exhibiting progressive enhancement. The contrast agent began to fade at 50s, resulting in lower enhancement. Combined with a literature review, it aims to provide more comprehensive information for the imaging diagnosis of MA.

This study developed and validated a deep learning model for diagnosing lymphadenopathy (LA) using B-mode ultrasound (BUS) and color Doppler flow imaging (CDFI) videos. A retrospective and prospective study was conducted from January 2016 to August 2025, including 7371 patients (3824 male [51.9

OBJECTIVES:To evaluate whether a multiparametric approach combining grayscale ultrasound (2D-US), contrast-enhanced ultrasound (CEUS), and serum anti-thyroid peroxidase antibody (TPO-Ab) improves the differentiation of benign from malignant thyroid nodules (TNs) in patients with Hashimoto's thyroiditis (HT) and to assess the stability of the combined diagnostic model through internal validation. METHODS:This retrospective study enrolled 600 HT patients with 650 pathologically confirmed TNs. All patients underwent 2D-US; CEUS was performed in 475, TPO-Ab testing in 452, and shear wave elastography (SWE) in an exploratory subset of 88 patients. The primary combined diagnostic model was developed using logistic regression (LR) on the subset of patients with complete 2D-US, CEUS, and TPO-Ab data. Internal validation was performed via stratified 5-fold cross-validation. Six machine learning classifiers were additionally compared to assess model robustness. RESULTS:Among single modalities, 2D-US showed the highest cross-validated area under the curve (AUC) (LR: 0.831). CEUS and TPO-Ab alone yielded limited discrimination (AUC 0.671 and 0.621, respectively). The combined model integrating 2D-US, CEUS, and TPO-Ab achieved a cross-validated AUC of 0.849 with sensitivity of 77.3% and specificity of 77.8%, outperforming any single modality. Among the 6 classifiers evaluated, LR demonstrated the most stable performance and the best balance between sensitivity and specificity. CONCLUSION:Integrating multimodal ultrasound with serum TPO-Ab significantly improves the differentiation of TNs in HT, with the combined LR model demonstrating robust performance after internal validation. This multiparametric approach may serve as a supplementary tool for risk stratification in surgical candidates, but prospective external validation is required before clinical implementation.

BACKGROUND:Conventional ultrasonography has limitations in distinguishing benign from malignant thyroid nodules classified as ACR TI-RADS 4-5. METHODS:We retrospectively enrolled 152 benign and 228 malignant papillary thyroid carcinomas (all ACR TI-RADS 4-5, confirmed by postoperative pathology). Propensity score matching was performed to balance conventional US features. RESULTS:After matching, 81 pairs of benign and malignant nodules were obtained. In this surgically treated cohort, multivariate analysis revealed that absence of filling defects on CEUS was independently associated with a higher likelihood of malignancy (p < 0.05). Incorporating CEUS features into the diagnostic model improved the AUC from 0.699 to 0.780. Notably, benign nodules with different pathological bases exhibited distinct CEUS manifestations. CONCLUSION:In this retrospective propensity-matched surgical cohort, CEUS provides incremental diagnostic value for ACR TI-RADS 4-5 thyroid nodules beyond conventional US. The combination of CEUS and conventional US improved diagnostic accuracy compared with conventional US alone. However, prospective studies in unselected patients are needed to validate these findings. The complex pathological underpinnings of different benign nodule types may contribute to variations in their CEUS manifestations.

Objective: The objective of this study was to develop and validate a predictive model for MTM-HCC by integrating preoperative ultrasound (US) and contrast-enhanced ultrasound (CEUS) features with relevant clinical characteristics. Methods: This retrospective study analyzed data from patients with histopathologically confirmed hepatocellular carcinoma who underwent preoperative CEUS examination at the Ultrasound Department of the Lanzhou University Second Hospital between December 2021 and March 2025. The study cohort comprised 45 patients diagnosed with MTM-HCC and 194 patients with non-MTM-HCC. Ultrasound and CEUS images were independently reviewed by two senior abdominal radiologists with extensive experience in hepatic imaging, ensuring objective feature assessment. Clinical variables and imaging characteristics were systematically compared between the two groups to identify distinguishing patterns. To evaluate the associations among clinical data, ultrasound-derived features, and MTM-HCC, univariate analyses were first performed, followed by multivariate logistic regression to construct and assess predictive models. Results: A total of 239 patients (mean age: 57.28 ± 9.60 years; 187 males and 52 females) were included in the analysis. Among them, 45 HCC patients (18.8%) were classified as MTM-HCC. Multivariate analysis identified four independent predictors: elevated alpha-fetoprotein (AFP ≥ 467 ng/mL) (OR = 8.5, 95% CI: 4.2-17.30; p < 0.001), presence of non-enhancing necrotic areas (OR = 5.92, 95% CI: 1.82-19.30, p = 0.003), intratumoral arteries (OR = 6.61, 95% CI: 2.28-19.22, p < 0.001), and peritumoral feeding arteries (OR = 3.13, 95% CI: 1.15-8.50, p = 0.025). Conclusions: An integrated prediction model that combines ultrasound imaging and clinical parameters offers a feasible, non-invasive approach for accurate preoperative identification of MTM-HCC.

Background:Non-valvular atrial fibrillation (NVAF) carries a high risk of left atrial appendage thrombus (LAAT) and dense spontaneous echo contrast (dense SEC), the primary triggers of cardioembolic stroke. Conventional CHADS2 [congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, prior stroke/transient ischemic attack (TIA) score] and CHA2DS2‑VASc (congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, prior stroke/TIA, vascular disease, age 65-74 years, Sex category score) scores lack left atrial appendage (LAA) morphological features, yielding limited predictive accuracy for dense SEC/LAAT. This study constructed a nomogram based on quantitative LAA parameters derived from three-dimensional transesophageal echocardiography (3D-TEE) to predict dense SEC/LAAT in NVAF patients and compared its performance with the two conventional clinical risk scores. Methods:We retrospectively enrolled 159 NVAF patients who underwent 3D-TEE from July 2024 to December 2025, stratified into a dense SEC/LAAT positive group (n=50) and a negative group (n=109). Variables with severe multicollinearity [variance inflation factor (VIF) ≥10] were excluded. Univariate logistic regression (P<0.10) screened candidate predictors, followed by forward stepwise multivariate logistic regression to identify independent risk factors and construct a nomogram. Model discrimination, calibration and clinical utility were assessed via receiver operating characteristic (ROC) curves, calibration curves, 10-fold cross-validation, decision curve analysis (DCA) and clinical impact curves; inter-model area under the curve (AUC) comparisons used P<0.05 as the statistical significance threshold. Results:Four independent predictors of dense SEC/LAAT were identified: D-dimer >0.550 mg/L [odds ratio (OR) =7.805, 95% confidence interval (CI): 2.044-29.795, P=0.002]; European Heart Rhythm Association (EHRA) score ≥ IIb (OR =6.255, 95% CI: 1.458-26.833, P=0.013); LAA poor echogenicity (OR =23.037, 95% CI: 5.651-93.909, P<0.001); LAA orifice morphology (OR =0.537, 95% CI: 0.296-0.975, P=0.041). The nomogram achieved an original AUC of 0.892 (95% CI: 0.839-0.945) at an optimal cutoff value of 0.29, with sensitivity 0.82, specificity 0.83, accuracy 0.82, and negative predictive value 0.91. It significantly outperformed CHADS2 (AUC =0.629) and CHA2DS2-VASc (AUC =0.606) (all P<0.05). Ten-fold cross-validation yielded an optimism-corrected AUC of 0.868 and a mean Brier score of 0.127; however, marked overfitting was observed (calibration slope =4.175, mean maximum calibration error =0.385). DCA confirmed sustained positive net clinical benefit within the 10-50% threshold probability range. Conclusions:The 3D-TEE-based nomogram shows acceptable discrimination for dense SEC/LAAT in NVAF patients and addresses the limitations of traditional risk scoring systems. Nevertheless, prominent overfitting prevents its direct clinical use without external validation and recalibration; it can serve as an auxiliary research tool for LAAT risk stratification.

Purpose: By integrating imaging data, laboratory markers, and the expression of immunohistochemical (IHC) markers, a mode l was developed to predict tumor progression (TP) following microwave ablation (MWA) in hepatocellular carcinoma (HCC), offering a novel approach for evaluating treat ment efficacy in HCC patients. Patients and Methods: This retrospective study included 91 patients with stage 0 or stage A BCLC with single pathologically confirmed HCC who underwent ultrasound-guided percutaneous MWA. Imaging features, laboratory indicators (such as liver function indicators and tumor markers), and IHC markers were included. The correlation between these three indices and TP after tumor ablation was analyzed, and a prognostic model was constructed to explore the influence of multiple factors on the prognosis of tumors after MWA. Results: Laboratory markers (Age-Platelet Index [API] and elevated pre-/post-ablation AFP/PIVKA-II) were significantly correlated with post-ablation TP (all p< 0.05). IHC marker positivity for GPC-3, Arg-1, HSP 70, and a high Ki-67 index predicted disease progression (all p< 0.05). Univariate analysis identified tumor size, high-risk location, morphology, washout time, API, pre-ablation AFP level > 200 ng/mL, post-ablation persistent AFP/PIVKA-II elevation, GPC-3/Arg-1/HSP 70 expression, and Ki-67 index as TP predictors. Multivariate analysis confirmed irregular morphology, API, AFP > 200 ng/mL, GPC-3/HSP 70 expression, and the Ki-67 index as independent predictors. ROC analysis demonstrated the superior predictive performance of the multifactor model (AUC=0.995). Conclusion: Tumor morphology, preoperative serum AFP levels (> 200 ng/mL), positive expression of GPC-3, HSP 70, and Ki-67 were independent predictors of HCC relapse after ablation. The model constructed in combination with these predictors showed high diagnostic performance in predicting post-ablation progression of HCC.