Texture Analysis of CT Images in Differential Diagnosis of Non-Small Cell Lung Cancer
Background. In current clinical practice, the information contained in computed tomography (CT) images of lung cancer is not used to its full extent – only a few semantic characteristics (e.g. size, contours, nature of contrast agent accumulation, etc.). Today, researchers are attempting to transfor...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Luchevaya Diagnostika, LLC
2025-07-01
|
| Series: | Вестник рентгенологии и радиологии |
| Subjects: | |
| Online Access: | https://www.russianradiology.ru/jour/article/view/938 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Background. In current clinical practice, the information contained in computed tomography (CT) images of lung cancer is not used to its full extent – only a few semantic characteristics (e.g. size, contours, nature of contrast agent accumulation, etc.). Today, researchers are attempting to transform CT image data into quantitative indicators describing the shape and texture of lung cancer, as well as to link these indicators with clinical data. This approach is called “radiomics” and is a developing field in medicine.Objective: to analyze publications on differential diagnosis of non-small cell lung cancer (NSCLC) using texture analysis as well as to assess the possibilities and prospects of this method in increasing information content of CT studies.Material and methods. The literature review presents data obtained from available sources in PubMed, ScienceDirect and Google Scholar databases, published up to and including the end of 2024, found using the key words and phrases in Russian and English languages: “NSCLC”, “lung adenocarcinoma”, “squamous cell lung cancer”, “computed tomography”, “radiomics”, “texture analysis”, “differential diagnostics”.Results. The literature review describes the methods of texture analysis at all stages. Based on the results of the studied scientific works, the authors conclude that the use of texture analysis allows non-invasively predicting the histological form of NSCLC with sensitivity 72–83%, specificity 67–92%, and accuracy 74–86%. Conclusion. The use of texture analysis, according to published studies, is a promising method for differential diagnosis of histological forms of NSCLC (up to AUC ~0.7–0.9), however, the difference in methods and the lack of standardization of texture analysis require additional research. |
|---|---|
| ISSN: | 0042-4676 2619-0478 |