RESUMEN
Background: We explored whether a model based on contrast-enhanced computed tomography radiomics features and clinicopathological factors can evaluate preoperative lymphovascular invasion (LVI) in patients with gastric cancer (GC) with Lauren classification. Methods: Based on clinical and radiomic characteristics, we established three models: Clinical + Arterial phase_Radcore, Clinical + Venous phase_Radcore and a combined model. The relationship between Lauren classification and LVI was analyzed using a histogram. Results: We retrospectively analyzed 495 patients with GC. The areas under the curve of the combined model were 0.8629 and 0.8343 in the training and testing datasets, respectively. The combined model showed a superior performance to the other models. Conclusion: CECT-based radiomics models can effectively predict preoperative LVI in GC patients with Lauren classification.
Asunto(s)
Neoplasias Gástricas , Humanos , Neoplasias Gástricas/diagnóstico por imagen , Neoplasias Gástricas/cirugía , Neoplasias Gástricas/patología , Estudios Retrospectivos , Tomografía Computarizada por Rayos X/métodos , Metástasis Linfática , Medios de ContrasteRESUMEN
BACKGROUND: There is a demand for additional alternative methods that can allow the differentiation of the breast tumor into molecular subtypes precisely and conveniently. PURPOSE: The present study aimed to determine suitable optimal classifiers and investigate the general applicability of computer-aided diagnosis (CAD) to associate between the breast cancer molecular subtype and the extracted MR imaging features. METHODS: We analyzed a total of 264 patients (mean age: 47.9 ± 9.7 years; range: 19-81 years) with 264 masses (mean size: 28.6 ± 15.86 mm; range: 5-91 mm) using a Unet model and Gradient Tree Boosting for segmentation and classification. RESULTS: The tumors were segmented clearly by the Unet model automatically. All the extracted features which including the shape features,the texture features of the tumors and the clinical features were input into the classifiers for classification, and the results showed that the GTB classifier is superior to other classifiers, which achieved F1-Score 0.72, AUC 0.81 and score 0.71. Analyzed the different features combinations, we founded that the texture features associated with the clinical features are the optimal features to different the breast cancer subtypes. CONCLUSION: CAD is feasible to differentiate the breast cancer subtypes, automatical segmentation were feasible by Unet model and the extracted texture features from breast MR imaging with the clinical features can be used to help differentiating the molecular subtype. Moreover, in the clinical features, BPE and age characteristics have the best potential for subtype.