Skip to content
Publications

Article · Integrative dentistry and maxillofacial surgery · December 20, 2024

Evaluation of the diagnostic value of texture analysis of CT images in differentiating benign and malignant tumors of the jaws

Abdullaeva L.Sh., Marat Khodjibekov, Yulduz M. Khodjibekova, Полатова Джамила Шагайратовна, Pattakhov A.Sh.

Abstract

Purpose of the study: to determine the diagnostic value of using radiomics in differentiating benign and malignant formations of the jaws on MSCT images.Material and methods: A radiomics analysis of CT images was retrospectively performed in 65 patients with histologically and clinically verified tumors of the jaws (28 benign and 37 malignant tumors). For texture analysis, the LIFEx 7.10 program (C. Nioche, F. Orlhac etc.) was used.Results: Using the Kruskal-Wallis test, 1 of 39 texture parameters - DISCRETIZED_AUC_CSH - showed statistically significant differences between benign and malignant tumors of the jaws (p<0.05). A predictive model was built from the selected texture feature using logistic regression. PredDis= 1/ (1 + exp (-1.02914-0.0044970* DISCRETIZED_AUC_CSH)). The regression values calculated from the model were normalized in the range from 0 to 1 using logit transformation and were used as texture heterogeneity indices.Conclusion: Texture analysis of CT images allows noninvasive prediction of benign or malignant nature of tumors of the upper and lower jaws. The texture heterogeneity index calculated from the logistic texture model has the highest prognostic accuracy. Texture analysis transforms standard computed tomography into a multiparametric study, complementing the qualitative assessment of anatomical details of the visualized formation with a quantitative functional indicator characterizing intratumor spatial heterogeneity.

Radiomics and Machine Learning in Medical ImagingImage Retrieval and Classification TechniquesDental Radiography and Imagingradiomicstexture analysisjaw neoplasms

More by Marat Khodjibekov