Academic Radiology最新文献

Rationale and objectives: To develop radiomics and deep learning models for differentiating malignant and benign soft tissue tumors (STTs) preoperatively based on fat saturation T2-weighted imaging (FS-T2WI) of patients.

Materials and methods: Data of 115 patients with STTs of extremities and trunk were collected from our hospital as the training set, and data of other 70 patients were collected from another center as the external validation set. Outlined Regions of interest included the intratumor and the peritumor region extending outward by 5 mm, then the corresponding radiomics features were extracted respectively. Deep learning was performed using pretrained 3D ResNet algorithms, and deep learning features were extracted from the entire FS-T2WI of patients. Recursive feature elimination and least absolute shrinkage and selection operator were used to select the radiomics and deep learning features with predictive value. Five machine learning algorithms were applied to build radiomics models, the area under the ROC curve (AUC) in the validation set were used to evaluate the diagnostic performance, and decision curve analysis (DCA) was used to evaluate the clinical benefit of models.

Results: Based on 20 selected deep learning and radiomics features, the deep learning radiomics (DLR) model had the best predictive performance in the validation set, with an AUC of 0.9410. DCA and calibration curves showed that the DLR model had better clinical net benefit and goodness of fit.

Conclusion: By extracting more features from FS-T2WI, the DLR model is a noninvasive, low-cost, and highly accurate preoperative differential diagnosis of benign and malignant STTs.

Rationale and objectives: To summarize the clinical features and ultrasonic characteristics of patients with cystic neutrophilic granulomatous mastitis (CNGM), and to enhance the understanding of CNGM in clinical practice.

Materials and methods: This study retrospectively analyzed the demographic data, clinical symptoms, and ultrasonic characteristics of 141 patients diagnosed with CNGM through pathological examination. This study was approved by the Medical Ethical Committee of Beijing Hospital of Traditional Chinese Medicine (2023BL02-054-01).

Results: CNGM is more common in women of childbearing age. Clinically, the majority of CNGM patients presented with breast lumps, redness, swelling, ulceration, nipple discharge, and nipple retraction. Some patients also exhibited systemic symptoms such as erythema nodosum, fever, and cough. Ultrasonically, CNGM was characterized by irregular, hypoechoic areas with unclear margins. Additional features included liquefaction areas within the hypoechoic areas, subcutaneous extension, tissue thickening, ductal dilatation, and axillary lymph node enlargement. Among the 141 cases, CNGM patients were classified into four types based on the nature and extent of the lesions: 87 cases of diffuse abscess type, 49 cases of patchy hypoechoic type, 2 cases of localized hypoechoic type, and 3 cases of localized abscess type.

Conclusion: This study describes the clinical features, ultrasound manifestations and their various subtypes, and microbial identification of patients with CNGM, which will help to better identify CNGM as a potential diagnostic entity with important implications for subsequent treatment.

Data availability statement: The original contributions presented in the study are included in the article. Further inquiries can be directed to the corresponding authors.

Rationale and objectives: Accurate preoperative pathological staging of gastric cancer is crucial for optimal treatment selection and improved patient outcomes. Traditional imaging methods such as CT and endoscopy have limitations in staging accuracy.

Methods: This retrospective study included 691 gastric cancer patients treated from March 2017 to March 2024. Enhanced venous-phase CT and endoscopic images, along with postoperative pathological results, were collected. We developed three modeling approaches: (1) nine deep learning models applied to CT images (DeepCT), (2) 11 machine learning algorithms using handcrafted radiomic features from CT images (HandcraftedCT), and (3) ResNet-50-extracted deep features from endoscopic images followed by 11 machine learning algorithms (DeepEndo). The two top-performing models from each approach were combined into the Integrated Multi-Modal Model using a stacking ensemble method. Performance was assessed using ROC-AUC, sensitivity, and specificity.

Results: The Integrated Multi-Modal Model achieved an ROC-AUC of 0.933 (95% CI, 0.887-0.979) on the test set, outperforming individual models. Sensitivity and specificity were 0.869 and 0.840, respectively. Various evaluation metrics demonstrated that the final fusion model effectively integrated the strengths of each sub-model, resulting in a balanced and robust performance with reduced false-positive and false-negative rates.

Conclusion: The Integrated Multi-Modal Model effectively integrates radiomic and deep learning features from CT and endoscopic images, demonstrating superior performance in preoperative pathological staging of gastric cancer. This multimodal approach enhances predictive accuracy and provides a reliable tool for clinicians to develop individualized treatment plans, thereby improving patient outcomes.

Data availability: The data presented in this study are available on request from the corresponding author. The data are not publicly available due to ethical reasons. All code used in this study is based on third-party libraries and all custom code developed for this study is available upon reasonable request from the corresponding author.

Rationale and objectives: To develop an automatic deep-radiomics framework that diagnoses and stratifies prostate cancer in patients with prostate-specific antigen (PSA) levels between 4 and 10 ng/mL.

Materials and methods: A total of 1124 patients with histological results and PSA levels between 4 and 10 ng/mL were enrolled from one public dataset and two local institutions. An nnUNet was trained for prostate masks, and a feature extraction module identified suspicious lesion masks. Radiomics features were extracted from the biparametric magnetic resonance imaging using these masks. Machine learning models were developed to diagnose prostate cancer (PCa), clinically significant PCa (csPCa), and high-risk csPCa based on radiomics and clinical features. The models were evaluated in both internal and external cohorts. The best model was further compared with PSA density (PSAD), free to total PSA (F/T PSA), and Prostate Imaging Reporting and Data System (PI-RADS) scores in the external cohort.

Results: The models based on both radiomics and clinical features outperformed those based on radiomics or clinical features alone. The top-performing models achieved areas under the curve of 0.80, 0.88, and 0.83 on internal testing, and 0.79, 0.80, and 0.82 on external testing for diagnosing PCa, csPCa, and high-risk csPCa. Our deep-radiomics model surpassed PSAD, F/T PSA, and PI-RADS scores in an external cohort. Decision curve analysis indicated that our model offers greater net benefit than these methods.

Conclusion: The deep-radiomics model automatically segments prostate and suspicious lesions, diagnoses, and stages of PCa in patients with PSA levels between 4 and 10 ng/mL. Our method addresses the shortcomings of manual segmentation and inconsistency, delivering outstanding performance. It provides multilevel predictions to assist clinical decision-making and benefit patients with gray zone PSA.

Rationale and objectives: Mixed ground-glass nodules (mGGNs) are highly malignant and common nonspecific lung imaging findings. This study aimed to explore whether combining quantitative and qualitative spectral dual-layer detector-based computed tomography (SDCT)-derived parameters with serological tumor abnormal proteins (TAPs) and thymidine kinase 1 (TK1) expression enhances invasive mGGN diagnostic efficacy and to develop a joint diagnostic model.

Materials and methods: This prospective study included patients with mGGNs undergoing preoperative triple-phase contrast-enhanced SDCT with TAP and TK1 tests. Based on pathologic invasiveness, mGGNs were classified as noninvasive or invasive adenocarcinomas. To establish the predictive model, 397 patients were divided into training and internal validation cohorts. Another 144 patients comprised the external validation set. A nomogram predicting invasive mGGNs was generated and assessed using receiver operating characteristic curves.

Results: CT100keV_a, Zeff_a, ED_a, TAP, Dsolid, and Internal_bronchial_morphology were identified as independent risk factors for mGGN invasiveness. The SDCT parameter-TAP nomogram combining these six predictors demonstrated satisfactory discrimination capabilities in all three datasets (areas under the curves 0.840-0.911). The optimal training set cutoff was 0.566, yielding an 88.2% sensitivity and 80.4% specificity. Decision curve analysis showed the highest net benefit across a breadth of threshold probabilities, and clinical impact curve analysis confirmed the model's clinical validity. The nomogram had significantly higher discriminative accuracy than any variable alone.

Conclusion: Multiple SDCT-derived parameters predict mGGN invasiveness, with Zeff_a playing a prominent role. The developed SDCT parameter-TAP nomogram has excellent diagnostic performance and high calibration accuracy, facilitating individual noninvasive risk prediction of malignant mGGNs.

Critical relevance statement: Multiple quantitative and functional parameters derived from SDCT can predict the pathological invasiveness of mGGNs, with Zeff_a playing a prominent role. A SDCT parameters-TAP nomogram has excellent diagnostic performance and high calibration accuracy, facilitating noninvasive prediction of individual risks of malignant mGGNs.

Rationale and objectives: Prostate cancer (PCa) is the second most common cancer among men and a leading cause of cancer-related mortalities. Radiomics has shown promising performances in the classification of PCa grade group (GG) in several studies. Here, we aimed to systematically review and meta-analyze the performance of radiomics in predicting GG in PCa.

Materials and methods: Adhering to PRISMA-DTA guidelines, we included studies employing magnetic resonance imaging-derived radiomics for predicting GG, with histopathologic evaluations as the reference standard. Databases searched included Web of Sciences, PubMed, Scopus, and Embase. The Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) and METhodological RadiomICs Score (METRICS) tools were used for quality assessment. Pooled estimates for sensitivity, specificity, likelihood ratios, diagnostic odds ratio, and area under the curve (AUC) were calculated. Cochran's Q and I-squared tests assessed heterogeneity, while meta-regression, subgroup analysis, and sensitivity analysis addressed potential sources. Publication bias was evaluated using Deek's funnel plot, while clinical applicability was assessed with Fagan nomograms and likelihood ratio scattergrams.

Results: Data were extracted from 43 studies involving 9983 patients. Radiomics models demonstrated high accuracy in predicting GG. Patient-based analyses yielded AUCs of 0.93 for GG≥2, 0.91 for GG≥3, and 0.93 for GG≥4. Lesion-based analyses showed AUCs of 0.84 for GG≥2 and 0.89 for GG≥3. Significant heterogeneity was observed, and meta-regression identified sources of heterogeneity. Radiomics model showed moderate power to exclude and confirm the GG.

Conclusion: Radiomics appears to be an accurate noninvasive tool for predicting PCa GG. It improves the performance of standard diagnostic methods, enhancing clinical decision-making.

Rationale and objectives: To compare the long-term efficacy of prostatic artery embolization (PAE) with PAE followed by holmium laser enucleation of the prostate (HoLEP) for benign prostatic hyperplasia (BPH) in patients with large prostatic volume (PV>80 cm³), and to identify the appropriate population for PAE+HoLEP.

Methods: From March 2015 to December 2023, 208 consecutive BPH patients were enrolled into two groups: PAE monotherapy (Group A, n=168) and PAE followed by HoLEP (Group B, n=40). Differences in clinical and functional parameters between baseline and each follow-up point were compared. Cumulative clinical success rates were assessed. Predictors of lower urinary tract symptoms (LUTS) recurrence were analyzed using ROC analyses and Cox proportional hazards regression.

Results: The median follow-up times in Group A and B were 36 and 48 months. Both groups showed significant improvements in clinical and functional parameters at each follow-up period compared to baseline (P<0.01). Cumulative clinical success rates in Group A were 95.3%, 91.6%, 80.6%, 68.0%, and 47.9%, compared to 100%, 100%, 100%, 100%, and 85.7% in Group B at 1, 2, 3, 4, and 5 years. Unilateral PAE and PV≥150.3 cm³ were independent predictors of LUTS recurrence in Group A (P<0.001).

Conclusion: PAE monotherapy and combination PAE+HoLEP were effective options for patients with large PV, but the LUTS recurrence rate of PAE increased over time. Unilateral PAE was a significant factor for recurrence. Patients with PV≥150.3 cm³ could be good candidates for a combined approach.

Objectives: There is a burgeoning discrepancy between the procedural competency of graduating diagnostic radiology residents and the needs of our patient population. The causes of this mismatch and opportunities for improvement are explored by the APDR Procedural Competency of Graduating DR Residents Task Force.

Materials and methods: The APDR convened a task force consisting of diverse broad stakeholder viewpoints, drawing from organized radiology, academic and private practices. The task force conducted structured analyses of the drivers contributing to the current state and reviewed relevant resources, conducted membership surveys, and developed consensus statements regarding solutions to the identified problem.

Results: A defined list of procedures a graduating resident is expected to competently perform is established. Key domain-based drivers of the currents state were identified including the ABR initial certification exam structure and content, ACGME practices, creation of the IR-DR residency and ESIR tracks, residency and fellowship training paradigms, and secular trends. The task force offers several best practice recommendations for improving procedural training in DR residency to better meet the needs of the marketplace and our patients.

Conclusion: Armed with a defined list of procedures expected of a general radiologist and best practices for enhancing procedural training in diagnostic residencies, the task force presents a national game-plan for improving our ability to deliver high value diagnostic and interventional services to the communities that need it most.