Insights into Imaging最新文献

Objective: To develop an ensemble machine learning (eML) model using multiphase computed tomography (MPCT) for distinguishing between gastric ectopic pancreas (GEP) and gastric stromal tumors (GIST) in lesions < 3 cm.

Methods: In this study, we retrospectively collected MPCT images from 138 patients between April 2017 and June 2023 across two centers. Cohort 1 comprised 94 patients divided into a training cohort and an internal validation cohort, while the 44 patients from Cohort 2 constituted the external validation cohort. Deep learning (DL) models were constructed based on the lesion region, and radiomics features were extracted to develop radiomics models, which were later integrated into the fusion model. Model performance was assessed through the analysis of the area under the receiver operating characteristic curve (AUROC). The diagnostic efficacy of the optimal model was compared with that of a radiologist. Additionally, the radiologist with the assistance of the eML model provides a secondary diagnosis, to assess the potential clinical value of the model.

Results: After evaluation using an external validation cohort, the radiomics model demonstrated the highest performance in the venous phase, achieving AUROC of 0.87. The DL model showed optimal performance in the non-contrast phase, with AUROC of 0.81. The eML achieved the best performance across all models, with AUROC of 0.90. The use of eML-assisted analysis resulted in a significant improvement in the junior radiologist's accuracy, rising from 0.77 to 0.93 (p < 0.05). However, the senior radiologist's accuracy, while improving from 0.86 to 0.95, did not exhibit a statistically significant difference.

Conclusion: eML model based on MPCT can effectively distinguish between GEPs and GISTs < 3 cm.

Critical relevance statement: The multiphase CT-based fusion model, incorporating radiomics and DL technology, proves effective in distinguishing between GEP and gastric stromal tumors, serving as a valuable tool to enhance diagnoses and offering references for clinical decision-making.

Key points: No studies yet differentiated these tumors via radiomics or DL. Radiomics and DL methodologies unveil potentially distinct phenotypes within lesions. Quantitative analysis on CT for GIST and ectopic pancreas. Ensemble learning aids accurate diagnoses, assisting treatment decisions.

Objective: To establish a practical risk stratification system (RSS) based on ultrasonography (US) and clinical characteristics for predicting soft tissue masses (STMs) malignancy.

Methods: This retrospective multicenter study included patients with STMs who underwent US and pathological examinations between April 2018 and April 2023. Chi-square tests and multivariable logistic regression analyses were performed to assess the association of US and clinical characteristics with the malignancy of STMs in the training set. The RSS was constructed based on the scores of risk factors and validated externally.

Results: The training and validation sets included 1027 STMs (mean age, 50.90 ± 16.64, 442 benign and 585 malignant) and 120 STMs (mean age, 51.93 ± 17.90, 69 benign and 51 malignant), respectively. The RSS was constructed based on three clinical characteristics (age, duration, and history of malignancy) and six US characteristics (size, shape, margin, echogenicity, bone invasion, and vascularity). STMs were assigned to six categories in the RSS, including no abnormal findings, benign, probably benign (fitted probabilities [FP] for malignancy: 0.001-0.008), low suspicion (FP: 0.008-0.365), moderate suspicion (FP: 0.189-0.911), and high suspicion (FP: 0.798-0.999) for malignancy. The RSS displayed good diagnostic performance in the training and validation sets with area under the receiver operating characteristic curve (AUC) values of 0.883 and 0.849, respectively.

Conclusion: The practical RSS based on US and clinical characteristics could be useful for predicting STM malignancy, thereby providing the benefit of timely treatment strategy management to STM patients.

Critical relevance statement: With the help of the RSS, better communication between radiologists and clinicians can be realized, thus facilitating tumor management.

Key points: There is no recognized grading system for STM management. A stratification system based on US and clinical features was built. The system realized great communication between radiologists and clinicians in tumor management.

Focal liver lesions are frequently encountered during imaging studies, and hemangiomas represent the most common solid liver lesion. Liver hemangiomas usually show characteristic imaging features that enable characterization without the need for biopsy or follow-up. On the other hand, there are many benign and malignant liver lesions that may show one or more imaging features resembling hemangiomas that radiologists must be aware of. In this article we will review the typical imaging features of liver hemangiomas and will show a series of potential liver hemangiomas' mimickers, giving radiologists some hints for improving differential diagnoses. CRITICAL RELEVANCE STATEMENT: The knowledge of imaging features of potential liver hemangiomas mimickers is fundamental to avoid misinterpretation. KEY POINTS: Liver hemangiomas typically show imaging features that enable avoiding a biopsy. Many benign and malignant liver lesions show imaging features resembling hemangiomas. Radiologists must know the potentially misleading imaging features of hemangiomas' mimickers.

Objectives: To noninvasively estimate three breast cancer biomarkers, estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) and enhance performance and interpretability via multi-task deep learning.

Methods: The study included 388 breast cancer patients who received the 3D whole breast ultrasound system (3DWBUS) examinations at Xijing Hospital between October 2020 and September 2021. Two predictive models, a single-task and a multi-task, were developed; the former predicts biomarker expression, while the latter combines tumor segmentation with biomarker prediction to enhance interpretability. Performance evaluation included individual and overall prediction metrics, and Delong's test was used for performance comparison. The models' attention regions were visualized using Grad-CAM + + technology.

Results: All patients were randomly split into a training set (n = 240, 62%), a validation set (n = 60, 15%), and a test set (n = 88, 23%). In the individual evaluation of ER, PR, and HER2 expression prediction, the single-task and multi-task models achieved respective AUCs of 0.809 and 0.735 for ER, 0.688 and 0.767 for PR, and 0.626 and 0.697 for HER2, as observed in the test set. In the overall evaluation, the multi-task model demonstrated superior performance in the test set, achieving a higher macro AUC of 0.733, in contrast to 0.708 for the single-task model. The Grad-CAM + + method revealed that the multi-task model exhibited a stronger focus on diseased tissue areas, improving the interpretability of how the model worked.

Conclusion: Both models demonstrated impressive performance, with the multi-task model excelling in accuracy and offering improved interpretability on noninvasive 3DWBUS images using Grad-CAM + + technology.

Critical relevance statement: The multi-task deep learning model exhibits effective prediction for breast cancer biomarkers, offering direct biomarker identification and improved clinical interpretability, potentially boosting the efficiency of targeted drug screening.

Key points: Tumoral biomarkers are paramount for determining breast cancer treatment. The multi-task model can improve prediction performance, and improve interpretability in clinical practice. The 3D whole breast ultrasound system-based deep learning models excelled in predicting breast cancer biomarkers.

Objective: Past decades of research into contrast media injections and optimization thereof in radiology clinics have focused on scan acquisition parameters, patient-related factors, and contrast injection protocol variables. In this review, evidence is provided that a fourth bucket of crucial variables has been missed which account for previously unexplained phenomena and higher-than-expected variability in data. We propose how these critical factors should be considered and implemented in the contrast-medium administration protocols to optimize contrast enhancement.

Methods: This article leverages a combination of methodologies for uncovering and quantifying confounding variables associated with or affecting the contrast-medium injection. Engineering benchtop equipment such as Coriolis flow meters, pressure transducers, and volumetric measurement devices are combined with small, targeted systematic evaluations querying operators, equipment, and the physics and fluid dynamics that make a seemingly simple task of injecting fluid into a patient a complex and non-linear endeavor.

Results: Evidence is presented around seven key factors affecting the contrast-medium injection including a new way of selecting optimal IV catheters, degraded performance from longer tubing sets, variability associated with the mechanical injection system technology, common operator errors, fluids exchanging places stealthily based on gravity and density, wasted contrast media and inefficient saline flushes, as well as variability in the injected flow rate vs. theoretical expectations.

Conclusion: There remain several critical, but not commonly known, sources of error associated with contrast-medium injections. Elimination of these hidden sources of error where possible can bring immediate benefits and help to drive standardized and optimized contrast-media injections.

Critical relevance statement: This review brings to light the commonly neglected/unknown factors negatively impacting contrast-medium injections and provides recommendations that can result in patient benefits, quality improvements, sustainability increases, and financial benefits by enabling otherwise unachievable optimization.

Key points: How IV contrast media is administered is a rarely considered source of CT imaging variability. IV catheter selection, tubing length, injection systems, and insufficient flushing can result in unintended variability. These findings can be immediately addressed to improve standardization in contrast-enhanced CT imaging.