OBJECTIVE. The purpose of our study was to evaluate whether sonographic characterization of focal splenic lesions could be improved by using low mechanical index pulse-inversion sonography after sulfur hexafluoride-filled microbubble injection.MATERIALS AND METHODS. One hundred forty-seven splenic lesions (68 benign, 79 malignant) in 147 patients (81 men, 66 women; mean age, 51 years) underwent baseline gray-scale sonography and sulfur hexafluoride–enhanced low-acoustic-power pulse-inversion sonography (mechanical index < 0.1). Two site investigators assessed in consensus lesion and splenic enhancement during arterial and parenchymal phases. Four readers (readers 1 and 2, blinded; and readers 3 and 4, unblinded to clinical data) independently reviewed baseline and contrast-enhanced sonograms and provided confidence rating for diagnosis of malignancy or benignancy. Accuracy, sensitivity, specificity, positive and negative predictive values, and areas under the receiver operating characteristic curves (Az) ...

Background: Machine learning (ML) algorithms have been trained to early predict critical in-hospital events from COVID-19 using patient data at admission, but little is known on how their performance compares with each other and/or with statistical logistic regression (LR). This prospective multicentre cohort study compares the performance of a LR and five ML models on the contribution of influencing predictors and predictor-to-event relationships on prediction model´s performance.Methods: We used 25 baseline variables of 490 COVID-19 patients admitted to 8 hospitals in Germany (March–November 2020) to develop and validate (75/25 random-split) 3 linear (L1 and L2 penalty, elastic net [EN]) and 2 non-linear (support vector machine [SVM] with radial kernel, random forest [RF]) ML approaches for predicting critical events defined by intensive care unit transfer, invasive ventilation and/or death (composite end-point: 181 patients). Models were compared for performance (area-under-the-receiver-operating characteristic-curve [AUC], Brier score) and predictor importance (performance-loss metrics, partial-dependence profiles).Results: Models performed close with a small benefit for LR (utilizing restricted cubic splines for non-linearity) and RF (AUC means: 0.763–0.731 [RF–L1]); Brier scores: 0.184–0.197 [LR–L1]). Top ranked predictor variables (consistently highest importance: C-reactive protein) were largely identical across models, except creatinine, which exhibited marginal (L1, L2, EN, SVM) or high/non-linear effects (LR, RF) on events.Conclusions: Although the LR and ML models analysed showed no strong differences in performance and the most influencing predictors for COVID-19-related event prediction, our results indicate a predictive benefit from taking account for non-linear predictor-to-event relationships and effects. Future efforts should focus on leveraging data-driven ML technologies from static towards dynamic modelling solutions that continuously learn and adapt to changes in data environments during the evolving pandemic.Trial registration number: NCT04659187.

Zusammenfassung: Ziel: Ziele dieser Arbeit waren die Identifikation und die Validierung von unterschiedlichen tumorspezifischen Kontrastmittelmustern für die Differenzialdiagnose zwischen benignen vaskulären Neoplasien (Typ Hämangiom/Hamartom) und malignen Milzherden in der Kontrastmittelsonografie (CEUS). Material und Methoden: 136 Milzläsionen (58 benigne vaskuläre Neoplasien, 78 maligne Läsionen) wurden jeweils in der B-Bild-Sonografie (US) und nachfolgend in der Kontrastmittelsonografie nach Sulfur-Hexafluoride-Injektion untersucht und dokumentiert. Zwei On-site-Reader analysierten das Kontrastmittelverhalten der Milzläsionen während der arteriellen und parenchymalen Phase. Mittels univariater und multivariater Analyse wurden tumorspezifische Kontrastmittelmuster, die signifikant mit einer der beiden Tumordiagnosen korreliert waren, identifiziert. Zwei geblindete Off-site-Reader benutzten diese tumorspezifischen Kontrastmittelmuster als Diagnosekriterien und stellten unabhängig voneinander anhand der Befunde der B-Bild-Sonografie und der Kontrastmittelsonografie eine Diagnose der Milzherde. Die diagnostische Treffsicherheit und Confidence (Az-Werte) sowie die Übereinstimmung der Off-site-Reader (қ-Werte) wurden kalkuliert. Referenzstandards waren Histopathologie oder CT und/oder MRI mit klinischem Verlauf. Ergebnisse: In der multivariate Analyse waren arterielles Hyper- oder Isoenhancement unabhängige Prediktoren für eine benigne vaskuläre Neoplasie (Odds Ratio, 3,558; p < 0,0017). In der Subgruppe der iso- oder hypoechogenen Läsionen war arterielles Hyperenhancement praktisch beweisend für eine benigne vaskuläre Neoplasie (Odds Ratio, 21,333; p < 0,001). In der geblindeten Situation verbesserte sich die diagnostische Treffsicherheit und Confidence (Az-Werte) für beide Off-site Reader signifikant von 63,2 % und 70,6 % (0,785 und 0,818) in der B-Bild-Sonografie auf 87,5 % und 88,2 % (0,915 und 0,908) in der Kontrastmittelsonografie (p < 0,001). Die Interreaderübereinstimmung verbesserte sich ebenfalls (қ = 0,52 für B-Bild-Sonografie vs. қ = 0,88 nach Kontrastmittelsonografie). Schlussfolgerung: Die Kontrastmittelsonografie verbessert die Differenzialdiagnose zwischen benigen vaskulären Neoplasien und malignen Tumoren in der Milz und scheint besonders hilfreich bei unklaren, zufällig im Ultraschall entdeckten hypoechogenen Milzherden.

At present, there are no widely accepted criteria for the use of radiofrequency ablation (RFA) for the treatment of colorectal liver metastases (CLM) in the context of effective modern-agent therapies. We aimed to define selection criteria for patients with liver-limited CLM who may benefit from adding RFA to systemic therapy with respect to long-term disease control.

To prospectively evaluate the role of real-time ultrasonography (US)-computed tomography (CT) fusion imaging (US-CT) in comparison with US seeing separate CT images (US + CT) and multidetector-row CT (MDCT) for the correct staging of hepatic metastatic involvement in patients with colorectal cancer. Sixty-four patients with newly diagnosed colorectal cancer and who were referred for abdominopelvic staging before primary tumor resection underwent same-day MDCT, US + CT, and US-CT. Examinations were evaluated on-site by 2 investigators in consensus. Investigators recorded the size and location of detected lesions on segmental liver maps, classified them as being benign, malignant, or indeterminate, and finally assessed the M stage of the liver as being M0, M1, or Mx (indeterminate). All patients underwent surgical exploration including intraoperative US. Reference standard diagnosis was based on findings at surgery, intraoperative US, histopathology, and MDCT follow-up imaging. Differences among investigated modalities were analyzed using McNemar's test. The reference standard verified 109 (45 < or = 1 cm) hepatic lesions in 25 patients, including 65 (25 < or = 1 cm) metastases in 16 patients (M1). Regarding the 45 < or = 1 cm liver lesions, rates for detection were significantly higher (P < 0.05) for MDCT (80%, 36/45) and US-CT (77.8%, 35/45) than for US + CT (64.4%, 29/45); the rate for correct classification by US-CT (71.1%, 32/45) was significantly higher than for US + CT (48.9%, 22/45) and MDCT (31.1%, 14/45) (all P < 0.05). On patient-based analysis, specificity of MDCT (85.4%, 41/48) was significantly lower (P < 0.05) than for US-CT (97.9%, 47/48) and US + CT (93.7%, 45/48); the positive predictive value of MDCT (63.1%, 12/19) was not significantly different (P = 0.27) compared with US + CT (82.3%, 14/17) but significantly lower (P < 0.05) than for US-CT (93.7%, 15/16). In 13 patients (59 lesions) with only benign (stage M0) or coexistent benign and malignant lesions (stage M1), indeterminate lesion ratings and indeterminate liver stagings (Mx) occurred both significantly lower (P < 0.05) with US-CT (3.4%, 2/59; and 0%, 0/13) than with US + CT (11.9%, 7/59; and 23.1%, 3/13) or with MDCT (30.5%, 18/59; and 53.8%, 7/13). Based on these initial diagnostic experiences, complementary US-CT fusion imaging of small CT-indeterminate liver lesions may have value in staging patients with colorectal cancer, focusing on patients who were likely to harbor only benign or coexisting benign and malignant liver lesions and in whom change of M staging would change the clinical management.