PurposeTo develop and validate an end-to-end interactive software that integrateslung ultrasound radiomics and machine learning for the objective assessmentof pediatric pneumonia severity at the point of care.Materials and methodsThis retrospective study included a dataset of lung ultrasound images from293 pediatric patients (157 with mild pneumonia and 136 with severepneumonia). A total of 104 radiomics features were extracted fromclinician-delineated regions of interest. Feature selection was performedusing Least Absolute Shrinkage and Selection Operator regression, and 10machine learning algorithms were constructed and evaluated. The optimalmodel was interpreted using SHapley Additive exPlanations. The finalizedclassifier was integrated into an interactive software platform that guidesusers from image upload to severity prediction.ResultsThe Light Gradient Boosting Machine classifier, which forms the core of thesoftware, demonstrated superior performance on the independent test set,achieving an accuracy of 89.8%, a sensitivity of 91.7%, a specificity of88.6%, and an area under the curve of 90.1%. SHapley Additive exPlanationanalysis identified and ranked the contribution of key predictive features,with morphological characteristics being the most influential.ConclusionsWe present an end-to-end interactive software that successfully leverageslung ultrasound radiomics and machine learning to provide an objective,accurate, and rapid assessment of pediatric pneumonia severity. This toolhas significant potential to standardize diagnosis and support clinicaldecision-making in real-world settings.

This study aimed to assess the feasibility of dynamic contrast-enhanced ultrasound with Sonazoid, its correlation with disease activity, its ability to differentiate between remission/mild activity and moderate/severe activity, and if a region of interest including the mucosa and submucosa was different compared to one encompassing the entire wall. We prospectively studied 48 patients with Crohn's disease who underwent colonoscopy and dynamic contrast-enhanced ultrasound with Sonazoid from 2015 to 2019. We compared the local simple endoscopic score for Crohn's disease in the most affected area with parameters from Vuebox, including peak enhancement, wash-in and wash-out area under the curves, wash-in and wash-out rates, and wash-in perfusion index. Region of interest-1 (ROI-1) included all wall layers, and region of interest-2 (ROI-2) included the mucosa and submucosa. Linear data and normalised data were used. Technical failure or a quality of fit value of<80% was considered a failed examination. The feasibility of the dynamic contrast-enhanced ultrasound examination was 73%. No significant findings were noted in the linear data ( p >0.05). In the normalised data in both ROI-1 and ROI-2, the local simple endoscopic score for Crohn's disease correlated significantly with peak enhancement ( r =0.38 and r =0.35), wash-in rate ( r =0.34 and r =0.34), and wash-in perfusion index ( r =0.40 and r =0.37) and with wash-in area under the curve ( r =0.36) in ROI-2. In patients with moderate/severe disease peak enhancement, wash-in area under the curve, wash-out area under the curve, wash-in rate, and wash-in perfusion index were significantly higher in both region of interests ( p <0.05). All parameters were significantly different in ROI-1 and ROI-2 ( p <0.05). Dynamic contrast-enhanced ultrasound parameters can differentiate between remission/mild and moderate/severe activity in Crohn's disease, but the method has relatively low feasibility.

PurposeTo investigate the potential correlation of dynamiccontrast-enhanced ultrasound (DCE-US) with angiogenesis activity of renalcell carcinoma (RCC).Materials and MethodsPatients with surgery resection andhistopathologically proven RCC lesions were included. B-mode ultrasound(BMUS) and contrast-enhanced ultrasound (CEUS) were performed one weekbefore surgery. SonoVue was injected as the contrast agent. VueBox (Bracco,Italy) was used for the quantitative analysis. According to thehistopathological and immunohistochemical results, patients were classifiedinto two groups: active angiogenesis and inactive angiogenesis. Timeintensity curves (TICs) and quantitative parameters were compared betweentwo groups.ResultsFrom July 2023 to November 2023, a total of 50patients (13 females and 37 males, mean age 61.1±11.1 years) were included.The mean size of the lesions was 39.4±2.7 mm. Patients were classified intothe active angiogenesis group (n=30) and the inactive angiogenesis group(n=20). On BMUS, 68.0% (34/50) of RCCs were visualized as hypoechoic lesionswith ill-defined borders and irregular shapes (P>0.05). During corticalphase of CEUS, 72.6% (23/30) of RCCs with active angiogenesis werevisualized with hyperenhancement (P=0.027). Only 30.0% (9/30) of RCCs withactive angiogenesis showed hypo-enhancement in the parenchymal phase(P>0.05). Compared to the inactive angiogenesis group, TICs of the activeangiogenesis group revealed faster and greater enhancement in the corticalphase, slower decline during the parenchymal phase, and an increased areaunder the curve. Among quantitative parameters, the active angiogenesisgroup showed the higher ratio of wash-in rate and wash-in perfusion index(P<0.05).ConclusionDCE-US analysis has potential value in predictingangiogenesis activity in RCC lesions.

a: Ultrasound revealed a reduction in the hypoechoic ductal area at the site of the original nodule.Long arrow: The restored sinus tract; the restored bone.b: CBCT shows bone regeneration in the defect area. 1

[This corrects the article DOI: 10.1055/a-2318-6654.].

This study aimed to measure the precise distance from the lowest boundary of a rectal tumor to the anal verge (DTAV) in patients with rectal cancer. A retrospective analysis was performed on clinical data from 70 rectal cancer patients. DTAV measurements were collected using transrectal biplane ultrasound, MRI, and colonoscopy. The difference in DTAV measurements between the mean DTAV value obtained by ultrasound (US mean ) and colonoscopy exhibited a difference of 0.22 cm. In contrast, the difference between US mean and MRI was 0.48 cm, while the difference between MRI and colonoscopy was -0.26 cm. The ICC for DTAV measurements demonstrated excellent agreement, with values of 0.948 between US mean and MRI, 0.942 between US mean and colonoscopy, and 0.943 between MRI and colonoscopy. The minimum DTAV value obtained by ultrasound (US min ) was 5.05 cm, the middle DTAV value obtained by ultrasound (US mid ) was 5.10 cm, and the maximum DTAV value obtained by ultrasound (US max ) was 5.30 cm. Notably, the median values of the differences in DTAV measurements between US max and US min , US max and US mid , as well as US mid and US min , were 0.2 cm, 0.1 cm, and 0.1 cm, respectively. Furthermore, the consistency of DTAV measurements between US min and US mid , US max and US mid , as well as US min and US max was excellent, with all ICC values reaching 0.999. Additionally, the radiologist's reassessment of MRI DTAV data showed excellent consistency with the original results, with an ICC value of 0.985. Transrectal biplane ultrasound utilizing EFOV imaging technology exhibited both accuracy and reproducibility for measuring DTAV. This approach provided a highly efficient and practical clinical tool for DTAV measurement.