Work

A retrospective study for the computed tomography radiomics changes of bone and muscles in patients with chronic obstructive pulmonary disease

Year: 2026

Type: article

Abstract: To investigate the computed tomography (CT) radiomics changes of bone and muscles in patients with chronic obstructive pulmonary disease (COPD). 102 COPD patients and 102 healthy controls from center A, and 96 COPD patients and 136 healthy controls from center B were included in the study. By using ‘TotalSegmentator’ of 3D slicer to process chest CT data, multiple organs were automatically segmented, including: spleen, liver, Th1-11 vertebra, humerus, scapula, spinal cord, deep back muscle and sternum. 1953 radiomics features were extracted from these organs. The dimensionality reduction of features were maximum relevancy and minimum redundancy (mRMR) and Least Absolute Selection and Shrinkage Operator (LASSO). The Receiver Operating Characteristic (ROC) curve was used for determine performance in identification of COPD. Area under curve (AUC) and accuracy were calculated for radiomics score and features. 190 CT features were significantly different between COPD and healthy cases. The AUC of radiomics score of Th11, Th10, left scapula and left deep back muscle were 0.791, 0.755, 0.722 and 0.714 respectively. By using mRMR, 30 features were selected from 190 features. By using LASSO, eighteen features had non-zero coefficients. In the 18 features, Th11, Th10, left scapula and left deep back muscle had more than one features. In the external data, the radiomics score of multiple organs had AUC of 0.863(0.814,0.907), sensitivity of 0.719 (0.623, 0.811), specificity of 0.897(0.841, 0.945), and accuracy of 0.816 (0.766, 0.867). CT radiomics is feasible in evaluating bone and muscles changes of patients with COPD. Th11, Th10, scapula and deep back muscle are more likely influenced by COPD.

Source: BMC Pulmonary Medicine

DOI: 10.1186/s12890-026-04890-z

Authors: Ruimin Guo, Ting Li, Nadeer M. Gharaibeh, Gang Wu, Hongwei Ren

Institutions: Tongji Hospital, Wuhan University of Science and Technology, Huazhong University of Science and Technology, Central Hospital of Wuhan

Language: English

Topic: Radiomics and Machine Learning in Medical Imaging

Subfield: Radiology, Nuclear Medicine and Imaging

Field: Medicine

Domain: Health Sciences

SDG: Good health and well-being

Open Access status: gold

Estimated APC paid: $3,090