Attention-based Saliency Maps Improve Interpretability of Pneumothorax Classification.

Purpose: To investigate the chest radiograph classification performance of vision transformers (ViTs) and interpretability of attention-based saliency maps, using the example of pneumothorax classification. Materials and Methods: In this retrospective study, ViTs were fine-tuned for lung disease classification using four public datasets: CheXpert, Chest X-Ray 14, MIMIC CXR, and VinBigData. Saliency maps were generated using transformer multimodal explainability and gradient-weighted class activation mapping (GradCAM). Classification performance was evaluated on the Chest X-Ray 14, VinBigData, and Society for Imaging Informatics in Medicine-American College of Radiology (SIIM-ACR) Pneumothorax Segmentation datasets using the area under the receiver operating characteristic curve (AUC) analysis and compared with convolutional neural networks (CNNs). The explainability methods were evaluated with positive and negative perturbation, sensitivity-n, effective heat ratio, intra-architecture repeatability, and interarchitecture reproducibility. In the user study, three radiologists classified 160 chest radiographs with and without saliency maps for pneumothorax and rated their usefulness. Results: ViTs had comparable chest radiograph classification AUCs compared with state-of-the-art CNNs: 0.95 (95% CI: 0.94, 0.95) versus 0.83 (95%, CI 0.83, 0.84) on Chest X-Ray 14, 0.84 (95% CI: 0.77, 0.91) versus 0.83 (95% CI: 0.76, 0.90) on VinBigData, and 0.85 (95% CI: 0.85, 0.86) versus 0.87 (95% CI: 0.87, 0.88) on SIIM-ACR. Both saliency map methods unveiled a strong bias toward pneumothorax tubes in the models. Radiologists found 47% of the attention-based and 39% of the GradCAM saliency maps useful. The attention-based methods outperformed GradCAM on all metrics. Conclusion: ViTs performed similarly to CNNs in chest radiograph classification, and their attention-based saliency maps were more useful to radiologists and outperformed GradCAM.Keywords: Conventional Radiography, Thorax, Diagnosis, Supervised Learning, Convolutional Neural Network (CNN) Online supplemental material is available for this article. © RSNA, 2023.

[Ultrasound diagnostics of the liver : Principles and important pathologies]. / Ultraschalldiagnostik der Leber : Grundlagen und wichtige Pathologien.

Diffuse changes in the liver parenchyma, focal lesions and blood flow in hepatic vessels can be assessed using ultrasound. Screening by ultrasound can be used to detect hepatocellular carcinomas as possible malignant sequelae of liver cirrhosis. As metastases are far more frequent than primary malignant liver tumors, secondary malignant neoplasms should be taken into consideration as a differential diagnosis in the presence of focal liver lesions. This particularly concerns patients with a known metastatic disease. Benign focal liver lesions are often incidentally discovered in women of childbearing age. Cysts, hemangiomas and focal nodular hyperplasia mostly show typical morphological features in ultrasound and do not require further follow-up; however, with hepatic adenomas, regular follow-up is recommended due to the risk of bleeding and/or malignant transformation.

Diagnostic Workup for Patients with Solid Renal Masses: A Cost-Effectiveness Analysis.

BACKGROUND: For patients with solid renal masses, a precise differentiation between malignant and benign tumors is crucial for forward treatment management. Even though MRI and CT are often deemed as the gold standard in the diagnosis of solid renal masses, CEUS may also offer very high sensitivity in detection. The aim of this study therefore was to evaluate the effectiveness of CEUS from an economical point of view. METHODS: A decision-making model based on a Markov model assessed expenses and utilities (in QALYs) associated with CEUS, MRI and CT. The utilized parameters were acquired from published research. Further, a Monte Carlo simulation-based deterministic sensitivity analysis of utilized variables with 30,000 repetitions was executed. The willingness-to-pay (WTP) is at USD 100,000/QALY. RESULTS: In the baseline, CT caused overall expenses of USD 10,285.58 and an efficacy of 11.95 QALYs, whereas MRI caused overall expenses of USD 7407.70 and an efficacy of 12.25. Further, CEUS caused overall expenses of USD 5539.78, with an efficacy of 12.44. Consequently, CT and MRI were dominated by CEUS, and CEUS remained cost-effective in the sensitivity analyses. CONCLUSIONS: CEUS should be considered as a cost-effective imaging strategy for the initial diagnostic workup and assessment of solid renal masses compared to CT and MRI.

Contrast-Enhanced Ultrasound (CEUS) for the Evaluation of Bosniak III Complex Renal Cystic Lesions-A 10-Year Specialized European Single-Center Experience with Histopathological Validation.

Background and objectives: The aim of the present retrospective single-center study is to evaluate the diagnostic performance of contrast-enhanced ultrasound (CEUS) for assessing Bosniak III complex renal cystic lesions with histopathological validation. Materials and Methods: 49 patients with CEUS-categorized Bosniak III renal cystic lesions were included in this retrospective study. All patients underwent native B-mode, Color Doppler, contrast-enhanced ultrasound (CEUS) between 2010-2020. Eight and five patients underwent computed tomography (CT) and magnetic resonance imaging (MRI), respectively. Twenty-nine underwent (partial) nephrectomy allowing for histopathological analysis. The applied contrast agent for CEUS was a second-generation blood pool agent. Ultrasonography examinations were performed and interpreted by a single experienced radiologist with more than 15 years of experience (EFSUMB Level 3). Results: CEUS examinations were successfully performed in all included patients without registering any adverse effects. The malignancy rate of CEUS-categorized Bosniak III renal lesions accounted for 66%. Initially, cystic complexity was visualized in native B-mode. In none of the renal lesions hypervascularization was detected in Color Doppler. CEUS allowed for detection of contrast enhancement patterns in all included Bosniak III renal lesions. Delayed wash-out could be detected in 6/29 renal lesions. In two cases of histopathologically confirmed clear-cell RCC, appropriate up-grading from Bosniak IIF to III was achieved by CEUS. Conclusions: CEUS depicts a promising imaging modality for the precise diagnostic workup and stratification of renal cystic lesions according to the Bosniak classification system, thereby helping guidance of adequate clinical management in the future.

Contrast-Enhanced Ultrasound (CEUS) for Follow-Up of Bosniak 2F Complex Renal Cystic Lesions-A 12-Year Retrospective Study in a Specialized European Center.

Bosniak 2F renal cystic lesions feature morphologic characteristics between Bosniak I and III categories, the majority of which remain benign. However, a minor part of Bosniak 2F lesions may progress to malignancy. The purpose of this study was to assess Bosniak 2F cystic lesions during follow-up examinations by CEUS. One-hundred-and-twelve out of 364 patients with Bosniak 2F lesions underwent follow-up CEUS examinations between February 2008 and February 2020. Twelve out of 364 patients underwent renal surgery without follow-up CEUS. The progression rate of Bosniak 2F renal lesions detected by CEUS accounted for 7.1% (8/112 patients) after a mean of 12.9 months. The first follow-up CEUS revealed 75% of progressions (6/8), the remaining 25% (2/8) of progressions were detected during second follow-up CEUS. Underlying clear-cell renal cell carcinoma was histopathologically validated in 5/8 progressive complex cystic renal lesions. Stable sonomorphologic features were observed in 92.1% (104/112 patients). CEUS depicts a promising diagnostic imaging modality in the diagnostic work-up and follow-up of complex renal cystic lesions at higher spatial and temporal resolutions than CT or MRI. Its excellent safety profile, its easy and repeatable accessibility, and low financial costs render CEUS an attractive and powerful alternative imaging tool for monitoring complex renal cystic lesions.