Cancer Imaging最新文献

Objective: The presence of calcification, especially microcalcification, is often associated with an increased risk of malignancy and closely linked to papillary thyroid carcinoma (PTC), the most common type of thyroid cancer. However, existing diagnostic ultrasound (US) imaging has critical limitations such as inability to detect subtle calcifications via standard static imaging, leading to 15-20% delayed PTC treatment or unnecessary fine-needle aspiration. This study aimed to develop a calcification-optimized, interpretable deep learning (DL) model based on dynamic ultrasound videos to determine the malignancy nature of calcified thyroid nodules.

Design and methods: This study retrospectively collected ultrasound dynamic video data from 1,257 patients, containing 2,319 thyroid nodules across six hospitals between January 2020 and October 2023. Various DL models were constructed with optimization specifically implemented on the 3D InceptionResNetV2 network by including a calcification attention module to enhance sensitivity to micro-calcifications. Model performance was compared not only with those trained on 2D static ultrasound images, but also against diagnoses from four clinicians (2 junior and 2 senior radiologists). The dataset was split into training (70%, 1,623 videos), validation (10%, 232 videos), internal test (10%, 232 videos), and external test (10%, 232 videos) sets.

Results: On the external test set, the optimized 3D InceptionResNetV2 model trained with dynamic videos outperformed the other four 3D DL models across all metrics: AUROC of 0.916, sensitivity of 0.860, and specificity of 0.834. Its AUROC was significantly higher than that of radiologists (0.916 versus 0.638; p < 0.0001). Additionally, with the assistance of the optimized model, radiologists' diagnostic accuracy improved by 16.9% (junior) and 11.1% (senior) in the external cohort. 3D Grad-CAM further confirmed the model focused on calcified regions (consistent with clinical diagnostic logic) by generating interpretable heatmaps.

Conclusion: A calcification-optimized DL model trained on dynamic ultrasound videos was proposed to efficiently and accurately predict the benign/malignant nature of calcified nodules. This tool shows promises as a non-invasive, interpretable tool for early PTC detection, supporting timely diagnosis and treatment planning.

Objective: To evaluate the efficiency of diffusion weighted imaging (DWI), the mean apparent diffusion coefficient (ADCmean) and the minimum apparent diffusion coefficient (ADCmin) values with different b-values (800 s/mm² and 1000 s/mm²) in the diagnosis and staging of endometrial carcinoma (EC).

Methods: Preoperative DWI images of 412 patients with EC and 134 patients with benign endometrial lesions were analyzed retrospectively. The performance of DWI images, ADCmean and ADCmin values with different b-values (800 s/mm² and 1000 s/mm²) for the diagnosis and staging (deep myometrial invasion, cervical stromal invasion and lymph node metastasis) of EC was assessed by using receiver operating characteristic curve (ROC). The comparison between AUCs was performed using the DeLong test, and a P value < 0.05 was considered statistically significant.

Results: The area under the curves (AUCs) of DWI protocol with b = 1000 s/mm² for qualitative assessment of EC diagnosis and staging (0.850, 0.837, 0.906, and 0.820 for diagnosis, deep myometrial invasion, cervical stromal invasion and lymph node metastasis, respectively) were higher than those of b = 800 s/mm² (0.821, 0.795, 0.860 and 0.814, respectively) (all p < 0.05). The AUCs for the quantitative assessment of EC diagnosis and staging with ADCmean and ADCmin values with b = 1000 s/mm² were higher than those of b = 800 s/mm² (all p < 0.05).

Conclusions: DWI images, ADCmean and ADCmin values with b = 1000 s/mm² had higher performance than those of b = 800 s/mm² in the assessment of EC diagnosis and staging. This study highlights the potential of using b = 1000 s/mm² as an optimized protocol for EC assessment in clinical practice.

Background: High-spatial-resolution T2-weighted imaging (HR-T2WI) has been demonstrated to overestimate the staging of early rectal cancer, which could lead to missed opportunities for organ-preserving treatments. This study aimed to investigate the value of diffusion kurtosis imaging (DKI) in distinguishing between T0-T1 and T2 rectal tumors.

Methods: A total of 138 patients with pathologically confirmed T0-T2 rectal tumors who underwent surgery between 2018 and 2023 were included. The pathological findings on tumor staging obtained from surgical specimens were used as the reference standard. The depth of tumor invasion was assessed using HR-T2WI. Kurtosis and diffusivity from DKI and apparent diffusion coefficient (ADC) from diffusion-weighted imaging were measured for the entire tumor. Diffusion parameters were compared between pT0-T1 and pT2 tumors. Multivariable logistic regression and receiver operating characteristic curve analyses were conducted to evaluate the diagnostic performance of significant individual parameters and their combinations in determining pT0-T1 tumors.

Results: Kurtosis was lower in pT0-T1 than in pT2 rectal tumors (0.799 vs. 0.950, P < 0.001), while diffusivity and ADC were higher than those of pT2 rectal cancer (1.732 × 10^-3 mm²/s vs. 1.368 × 10^-3 mm²/s, P < 0.001; 1.316 × 10^-3 mm²/s vs. 1.043 × 10^-3 mm²/s, P < 0.001). Diffusivity demonstrated the highest diagnostic efficacy in differentiating pT0-T1 from pT2 rectal tumors, with an AUC of 0.810, which was higher that of HR-T2WI (AUC = 0.752, P < 0.001) and kurtosis (AUC = 0.729, P = 0.007), but showed no difference compared to ADC (AUC = 0.785, P = 0.087). A multivariable logistic regression model incorporating HR-T2WI and diffusivity improved diagnostic performance compared with all other individual parameters, achieving an AUC of 0.885 (all P < 0.05).

Conclusion: The combination of HR-T2WI and diffusivity can effectively detect pT0-T1 rectal tumors. HR-T2WI combined with diffusivity derived from DKI may serve as a potential biomarker for early assessment of rectal tumors, offering valuable insights for selecting suitable candidates for organ-preserving surgery.

Background: Lymphovascular invasion (LVI) status in rectal cancer (RC) without lymph node metastasis (LNM) can significantly influence the patient's treatment decisions. This study aims to develop and validate a combined model based on MRI radiomics features integrated with clinical immune-inflammatory biomarkers for the prediction of LVI status in RC without LNM. The Shapley Additive Explanation (SHAP) method was employed to visualize the prediction process and enhance interpretability for clinical application.

Methods: We retrospectively collected data from 257 RC patients without LNM from two centers. Univariate and multivariate logistic regression analyses were performed on clinical data to identify independent predictors of LVI. Volumes of interest were manually delineated on T2WI and ADC sequences, and corresponding radiomic features were extracted. A combined model was constructed by combining rad-score and clinical immune-inflammatory biomarkers, and the SHAP was used to visualize the prediction process.

Results: The area under the curve (AUC) of the combined model was based on intratumoral features (training vs. testing vs. validation datasets: 0.813 vs. 0.854 vs. 0.807). The AUC of the combined model was based on both intra- and peritumoral features (training vs. testing vs. validation datasets: 0.855 vs. 0.841 vs. 0.860). After comparison, the combined model (C + Q) based on intra- and peritumoral MRI radiomics features integrated with clinical immune-inflammatory biomarkers demonstrated better predictive performance.

Conclusion: The combined model (C + Q) has great potential in the non-invasive prediction of LVI in RC without LNM, providing a basis for stratified management and individualized treatment decisions for RC patients.

Introduction: Disappearing colorectal liver metastases (DLM) frequently occur during chemotherapy. However, DLM is not equivalent to pathologically complete response. This study aimed to investigate the effect of radiographic DLM on microwave ablation (MWA) in patients with synchronous colorectal liver metastases (CRLM).

Methods: A retrospective review was performed for patients who accepted MWA following pre-ablation chemotherapy from January 2014 to December 2021. DLM was defined as undetectable tumors on pre-ablation contrast-enhanced imagings compared to the initial ones. Overall survival (OS) and intrahepatic progression-free survival (ihPFS) were analyzed and compared between patients with and without DLM. Univariate and multivariate cox regression were used to identify risk factors for OS and ihPFS. A propensity score matching (PSM) analysis was used to balance the patient demographics.

Results: Sixty-eight patients with DLM and 97 without DLM were included. The 1-year, 3-year, and 5-year ihPFS rates were significantly lower for patients with DLM compared to those without DLM before and after PSM (55.7%, 36.8%, and 30.6% vs. 70.8%, 59.3%, and 52.0% before PSM, respectively, p = 0.012; 44.9%, 31.8%, and 21.2% vs. 72.3%, 58.8%, and 47.5% after PSM, respectively, p = 0.039). Twenty-three (33.8%) patients with DLM had DLM-site recurrences during follow-up. The OS was not statistically different between the two groups both before and after PSM (p-value = 0.11 and 0.49). Multivariable cox regression revealed DLM (HR = 2.2; 95% CI = 1.1-4.1; p-value = 0.009) was a risk factor for poor ihPFS.

Conclusion: Patients with DLM presented worse ihPFS, suggesting that to eradicate visible tumors before disappearance may be advantageous when synchronous CRLM is ablatable.

Purpose: To evaluate the accuracy of fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG PET/CT) in detecting metastatic uveal melanoma (UM) using both per-patient and per-lesion analyses, while also characterizing lesion detectability across various metastatic sites.

Methods: In this retrospective study conducted from January 2011 to September 2024, UM participants underwent PET/CT scans for follow-up or suspected recurrence. The lesion uptake were quantified by maximum standardized uptake value (SUVmax). Pathology and clinical follow-up served as reference standard.

Results: Fifty-five participants (mean age, 49.2 ± 12.7; 26 females) were evaluated, and the average recurrent time was 30.7 months (IQR, 18.0-89.2). On per-patient level, 31 patients (56%) were confirmed to have metastatic lesions through pathology or clinical follow-up, of which 28/31 (90.3%) patients were successfully detected by ¹⁸F-FDG PET/CT and 3/31 (9.7%) patients with liver metastases were missed. Seventeen of 31 patients (54.8%) had multiple organ involvement. On per-lesion level, a total of 270 lesions were comfirmed, of which 245 (90.7%) were detected by ¹⁸F-FDG PET/CT, including metastasis to liver (103 of 128, 80.5%), bone (64 of 64, 100%), lymph node (24 of 24, 100%), lung (33 of 33, 100%), and other uncommen sites (21 of 21, 100%). The detection ability of ¹⁸F-FDG for liver metastases was positively correlated with the diameter of the lesions (r² = 0.671, p = 0.000). ¹⁸F-FDG successfully detected all bone, lymph node, and lung metastases, with 30 of 64 (46.9%) bone metastases showing no changes on CT and 12 of 24 (50%) lymph node metastases being less than 10 mm, making them prone to misdiagnosis on CT.

Conclusion: ¹⁸F-FDG PET/CT may be a useful diagnostic tool in detecting metastatic UM, especially for early bone metastases and small lymph nodes. Added contrast-enhanced computed tomography (CT) or magnetic resonance imaging (MRI) may be still needed for tiny liver metastases detection.

Background: To evaluate the feasibility and diagnostic utility of a deep learning (DL)-based super-resolution (SR) reconstruction algorithm applied to pancreatobiliary MRI for assessing pancreatic intraductal papillary mucinous neoplasms (IPMNs).

Methods: This retrospective study included 162 patients with presumed pancreatic IPMN (≥ 1 cm) who underwent pancreatobiliary MRI between May 2019 and May 2022. Two portal venous phase (PVP) images of dynamic T1-wegithed imaging were sequentially acquired: early PVP image obtained using standard compressed sensing (CS)-volumetric interpolated breath-hold examination (VIBE) (standard CS-VIBE) and late PVP image obtained using CS-VIBE with DL-based SR reconstruction algorithm to generate 1 mm-thickness images (DL-SR CS-VIBE). Arterial phase and 3-min delayed phase were also acquired using DL-SR CS-VIBE. The image quality of standard and DL-SR CS-VIBE PVP sequences was compared using Wilcoxon signed-rank test. The diagnostic performance of full-sequence pancreatobiliaryMRI including DL-SR CS-VIBE for predicting malignant IPMN was assessed using multi-reader multi-case analysis. Diagnostic accuracy was assessed using receiver operating characteristic analysis, while sensitivity and specificity were estimated with corresponding 95% confidence intervals.

Results: Among 162 patients, 15 had malignant IPMN, while 147 had benign IPMN. DL-SR CS-VIBE demonstrated significantly better overall image quality (3.73 ± 0.33 vs. 3.22 ± 0.43) and cystic lesion conspicuity (3.37 ± 0.50 vs. 2.71 ± 0.52) than standard CS-VIBE (all Ps < 0.001). The area under the ROC curve (AUC) for predicting malignant IPMN was 0.858 (95% CI: 0.807, 0.909). Using the presence of high-risk stigmata as an indicator of test-positive, pooled sensitivity and pooled specificity of pancreatobiliary MRI including DL-SR CS-VIBE for malignant IPMN were 71.1% (95% confidence interval [CI]: 55.7, 83.6) and 82.8% (95% CI: 78.9, 86.2), respectively. Among MRI features, diagnostic accuracy was highest for mural nodules ≥ 5 mm (AUC, 0.736) and main pancreatic duct size ≥ 10 mm (AUC, 0.720).

Conclusion: Pancreatobiliary MRI with DL-SR CS-VIBE enhances image quality and lesion conspicuity, offering promising diagnostic accuracy for malignant IPMN, though further studies with larger cohorts are needed to refine these findings and evaluate clinical impact.

Purpose: T1rho imaging showed potential applications in cancer imaging but little research explored the underlying biological processes that contribute to the T1rho values in cancer. This study aimed to investigate the potential associations between quantitative imaging biomarkers from T1rho imaging and the well-established diffusion weighted imaging (DWI), with tumour-stromal, immunohistochemical (IHC), and tumour-infiltration-lymphocytes (TIL) biomarkers in nasopharyngeal carcinoma (NPC).

Methods: Pre-treatment T1rho and DWI imaging of primary NPCs were performed in 50 prospectively recruited patients. The mean T1rho and apparent diffusion coefficient (ADC) of NPC were obtained and correlated with tumour-stromal, IHC, TIL biomarkers using the Pearson Correlation test and the coefficients (R) were calculated.

Results: The mean T1rho values negatively correlated with collagenous stroma-lymphoid stroma (R=-0.314, p = 0.03) and positively correlated with percentage of tumour cells positive for Ki-67 (R = 0.402, p < 0.01), but there were no associations between T1rho values and the other tumour-stromal, IHC or TIL biomarkers (p = 0.16-0.98) or between ADC values and any of these biomarkers (p = 0.07-0.82).

Conclusion: Our results showed the possible underlying biological mechanisms of T1rho imaging in head and neck cancer. T1rho imaging negatively correlated with the ratio of collagenous to lymphoid stroma, and positively correlated with tumour cell proliferation, which are both known to be predictors of outcome, suggesting that T1rho imaging may have a valuable role in head and neck cancer imaging. As this is a preliminary study with small sample size, further studies are encouraged to validate our findings.