25 August 2025 in Digital Diagnostics
ВАВасильев Юрий Александрович
Васильев Юрий Александрович
+1
BACKGROUND: Texture analysis improves the diagnostic accuracy of magnetic resonance imaging and differential diagnosis of prostate lesions, which are primarily segmented through manual labeling, resulting in significant inter-expert variability of masks. A consensus-based technique can help reduce inconsistencies in prostate lesion segmentation. However, global scientific studies have not described any standardized, consensus-based labeling protocols. AIM: This study aimed to develop a consensus algorithm for manual labeling of prostate lesions by several independent experts and evaluate inter-expert consistency in lesion segmentation. METHODS: This retrospective study included 60 biparametric magnetic resonance imaging scans of the prostate gland performed according to PI-RADS 2.1 technical specification. The scans showed PI-RADS 3, 4, and 5 lesions. Two independent radiologists manually segmented the prostate lesions using 3D Slicer. Then, the resulting masks were compared using the Dice–Sørensen coefficient (DSC). For lesions with DSC ≥ 0.75, the final mask was based on the overlap between the two original masks. Conversely, for lesions with DSC 0.75, the final mask was determined using the proposed consensus algorithm. RESULTS: The proposed consensus algorithm significantly increased the DSC values, from 0.61 [0.48; 0.73] for primary labeling to 0.74 [0.62; 0.79] for labeling using the proposed algorithm (p = 0.01). CONCLUSION: The proposed consensus-based algorithm for labeling prostate lesions using magnetic resonance imaging data is crucial in addressing inadequate approaches to objective segmentation in research and clinical settings.
5 April 2026 in Siberian Journal of Clinical and Experimental Medicine
Background . Preoperative differentional diagnosis of meningioma grade remains challenging with routine brain magnetic resonance imaging (MRI). The lack of reliable non-invasive tools limits the potential for early risk stratification and treatment planning. Aim : To develop an interpretable classification radiomic model RadMenGR for predicting meningioma grade (Grade I or Grade II) based on contrast-enhanced T1-weighted images. Material and Methods . This retrospective single-center study was conducted using the open-source anonymized dataset MeningiomaSEG-CLASS. 95 patients were included in the analysis (53 with Grade 1 and 42 with Grade 2 tumors). 105 radiomic features were extracted from manually segmented MR images using PyRadiomics. Classification was performed with a Naive Bayes algorithm following feature discretization using the Entropy / MDL method. Diagnostic performance was assessed using the area under the curve (AUC), sensitivity, specificity, and accuracy. Bootstrap analysis with 10,000 iterations and a 95% confidence interval was used for validation. Results . On the validation cohort (n = 46), the ROC-AUC was 0.805 (95% CI: 0.671–0.915). The lower bound of the 95% CI for the AUC exceeded the value under the null hypothesis (AUC = 0.63), confirming the statistical significance of the results (p < 0.05). Conclusion. This study developed an interpretable radiomic classification model for the differential diagnosis of Grade 1 and Grade 2 meningiomas. The application of a Naive Bayes algorithm to features extracted from contrast-enhanced T1-weighted images and transformed using a discretization method enabled the achievement of a significant level of diagnostic accuracy. However, the width of the confidence interval points to a lack of model robustness, necessitating validation on an independent cohort.