Search

Objective: Deploying an automatic segmentation model in practice should require rigorous quality assurance (QA) and continuous monitoring of the model's use and performance, particularly in high-stakes scenarios such as healthcare. Currently, however, tools to assist with QA for such models are not available to AI researchers. In this work, we build a deep learning model that estimates the quality of automatically generated contours. Methods: The model was trained to predict the segmentation quality by outputting an estimate of the Dice similarity coefficient given an image contour pair as input. Our dataset contained 60 axial T2-weighted MRI images of prostates with ground truth segmentations along with 80 automatically generated segmentation masks. The model we used was a 3D version of the EfficientDet architecture with a custom regression head. For validation, we used a fivefold cross-validation. To counteract the limitation of the small dataset, we used an extensive data augmentation scheme capable of producing virtually infinite training samples from a single ground truth label mask. In addition, we compared the results against a baseline model that only uses clinical variables for its predictions. Results: Our model achieved a mean absolute error of 0.020 ± 0.026 (2.2% mean percentage error) in estimating the Dice score, with a rank correlation of 0.42. Furthermore, the model managed to correctly identify incorrect segmentations (defined in terms of acceptable/unacceptable) 99.6% of the time. Conclusion: We believe that the trained model can be used alongside automatic segmentation tools to ensure quality and thus allow intervention to prevent undesired segmentation behavior. [ABSTRACT FROM AUTHOR]

Objectives: Radiomic involves testing the associations of a large number of quantitative imaging features with clinical characteristics. Our aim was to extract a radiomic signature from axial T2-weighted (T2-W) magnetic resonance imaging (MRI) of the whole prostate able to predict oncological and radiological scores in prostate cancer (PCa). Methods: This study included 65 patients with localized PCa treated with radiotherapy (RT) between 2014 and 2018. For each patient, the T2-W MRI images were normalized with the histogram intensity scale standardization method. Features were extracted with the IBEX software. The association of each radiomic feature with risk class, T-stage, Gleason score (GS), extracapsular extension (ECE) score, and Prostate Imaging Reporting and Data System (PI-RADS v2) score was assessed by univariate and multivariate analysis. Results: Forty-nine out of 65 patients were eligible. Among the 1702 features extracted, 3 to 6 features with the highest predictive power were selected for each outcome. This analysis showed that texture features were the most predictive for GS, PI-RADS v2 score, and risk class; intensity features were highly associated with T-stage, ECE score, and risk class, with areas under the receiver operating characteristic curve (ROC AUC) ranging from 0.74 to 0.94. Conclusions: MRI-based radiomics is a promising tool for prediction of PCa characteristics. Although a significant association was found between the selected features and all the mentioned clinical/radiological scores, further validations on larger cohorts are needed before these findings can be applied in the clinical practice. Key Points: • A radiomic model was used to classify PCa aggressiveness. • Radiomic analysis was performed on T2-W magnetic resonance images of the whole prostate gland. • The most predictive features belong to the texture (57%) and intensity (43%) domains. [ABSTRACT FROM AUTHOR]

Objectives: To assess the diagnostic accuracy of MRI performed using surface coils, with and without contrast medium, in predicting thyroid and cricoid cartilage infiltration in laryngeal tumours, and to investigate whether the radiologist's experience influences diagnostic accuracy.Methods: We retrospectively enrolled patients with biopsy-proven laryngeal cancer who had undergone preoperative staging MRI and open surgery. Two radiologists with different experience (senior vs. junior) reviewed the MR images without (session A1) and with contrast medium (session A2) separately. We calculated the accuracy of MRI with and without contrast medium in detecting infiltration of the thyroid and cricoid cartilages. Interobserver agreement was calculated by Cohen's Kappa (k).Results: Forty-two patients were enrolled, for a total of 62 cartilages. In session A1 the senior and junior radiologists showed an accuracy of 85% and 71%, respectively, with k = 0.53 (0.33-0.72). In session A2 the senior and junior radiologists showed an accuracy of 84% and 77%, respectively, with k = 0.68 (0.49-0.86).Conclusions: Staging of laryngeal tumours with surface coil MRI showed good diagnostic accuracy in assessing cartilaginous infiltration. We observed similar values of diagnostic accuracy for the analysis performed with and without contrast medium for the senior radiologist.Key Points: • Surface coil MRI demonstrated good accuracy in assessing laryngeal cartilage invasion. • The radiologist's experience can influence the diagnostic accuracy. • Gadolinium administration may increase interobserver concordance. [ABSTRACT FROM AUTHOR]

Purpose: To investigate whether there is an increased signal intensity (SI) of dentate nucleus (DN) and globus pallidus (GP) on unenhanced T1-weighted magnetic resonance imaging (MRI), in patients who had undergone multiple administrations of gadoxetate disodium.Materials and Methods: We retrospectevely included stage III melanoma patients, who had been previously enrolled in a trial of adjuvant therapy and who had undergone whole-body contrast-enhanced MRIs with gadoxetate disodium every three months for their follow-up. The SI ratios of DN-to-pons and GP-to-thalamus on unenhanced T1-weighted images were calculated. The difference in SI ratios between the first and the last MRI examinations was assessed and a linear mixed model was performed to detect how SI ratios varied with the number of administrations.Results: Eighteen patients were included in our study. The number of gadoxetate disodium administrations ranged from 2 to 18. Paired t-test did not show any significant difference in DN-to-pons (p=0.21) and GP-to-thalamus (p=0.09) SI ratios by the end of the study. DN-to-pons SI ratio and GP-to-thalamus SI ratio did not significantly increase with increasing the number of administrations (p=0.14 and p=0.06, respectively).Conclusion: Multiple administrations of gadoxetate disodium are not associated with increased SI in DN and GP in the brain.Key Points: • Gadolinium may deposit in the human brain after multiple GBCA administrations. • Gadolinium deposition is associated with increased T1W signal intensity • Increase in signal intensity is most apparent within the DN and GP • Multiple administrations of gadoxetate disodium do not increase T1W signal. [ABSTRACT FROM AUTHOR]

Objectives: To assess the independent prognostic value of standardized uptake value (SUV) and apparent diffusion coefficient (ADC), separately and combined, in order to evaluate if the combination of these two variables allows further prognostic stratification of patients with head and neck squamous cell carcinomas (HNSCC).Methods: Pretreatment SUV and ADC were calculated in 57 patients with HNSCC. Mean follow-up was 21.3 months. Semiquantitative analysis of primary tumours was performed using SUVmaxT/B, ADCmean, ADCmin and ADCmax. The prognostic value of SUVmaxT/B, ADCmean, ADCmin and ADCmax in predicting disease-free survival (DFS) was evaluated with log-rank test and Cox regression models.Results: Patients with SUVmaxT/B ≥5.75 had an overall worse prognosis (p = 0.003). After adjusting for lymph node status and diameter, SUVmaxT/B and ADCmin were both significant predictors of DFS with hazard ratio (HR) = 10.37 (95 % CI 1.22-87.95) and 3.26 (95 % CI 1.20-8.85) for SUVmaxT/B ≥5.75 and ADCmin ≥0.58 × 10-3 mm2/s, respectively. When the analysis was restricted to subjects with SUVmaxT/B ≥5.75, high ADCmin significantly predicted a worse prognosis, with adjusted HR = 3.11 (95 % CI 1.13-8.55).Conclusions: The combination of SUVmaxT/B and ADCmin improves the prognostic role of the two separate parameters; patients with high SUVmaxT/B and high ADCmin are associated with a poor prognosis.Key Points: • High SUV maxT/B is a poor prognostic factor in HNSCC • High ADC min is a poor prognostic factor in HNSCC • In patients with high SUV maxT/B , high ADC min identified those with worse prognosis. [ABSTRACT FROM AUTHOR]