Radiomics analysis in differentiating osteosarcoma and chondrosarcoma based on T2-weighted imaging and contrast-enhanced T1-weighted imaging

Abstract

This study was performed to investigate the diagnostic value of radiomics models constructed by fat suppressed T2-weighted imaging (T2WI-FS) and contrast-enhanced T1-weighted imaging (CET1) based on magnetic resonance imaging (MRI) for differentiation of osteosarcoma (OS) and chondrosarcoma (CS). In this retrospective cohort study, we included all inpatients with pathologically confirmed OS or CS from Second Xiangya Hospital of Central South University (Hunan, China) as of October 2020. Demographic and imaging variables were extracted from electronic medical records and compared between OS and CS group. Totals of 530 radiomics features were extracted from CET1 and T2WI-FS sequences based on MRI. The least absolute shrinkage and selection operator (LASSO) method was used for screening and dimensionality reduction of the radiomics model. Multivariate logistic regression analysis was performed to construct the radiomics model, and receiver operating characteristic curve (ROC) was generated to evaluate the diagnostic accuracy of the radiomics model. The training cohort and validation cohort included 87 and 29 patients, respectively. 8 CET1 features and 15 T2WI-FS features were screened based on the radiomics features. In the training group, the area under the receiver-operator characteristic curve (AUC) value for CET1 and T2WI-FS sequences in the radiomics model was 0.894 (95% CI 0.817-0.970) and 0.970 (95% CI 0.940-0.999), respectively. In the validation group, the AUC value for CET1 and T2WI-FS sequences in the radiomics model was 0.821 (95% CI 0.642-1.000) and 0.899 (95% CI 0.785-1.000), respectively. In this study, we developed a radiomics model based on T2WI-FS and CET1 sequences to differentiate between OS and CS. This model exhibits good performance and can help clinicians make decisions and optimize the use of healthcare resources.

Keywords: Chondrosarcoma; Differential diagnosis; Magnetic resonance imaging; Osteosarcoma; Radiomics.

Fig. 1

The flowchart of eligible patients.

Fig. 2

Basic procedures for the study. (1) Screen patients suitable for this study from all OS and CS patients in our hospital and obtain CET1 and T2WI-FS sequence data; (2) experienced radiologist outlines the tumor on CET1 and T2WI-FSMR images; (3) extract 530 initial radiomics features from the tumor area delineated by the original MR image; (4) select radiomics features by least absolute shrinkage and selection operator (LASSO) logistic regression; (5) use ROC analysis to evaluate the established radiomics model.

Fig. 3

Radiomics features were selected from CET1 and T2WI-FS using the LASSO binary logistic regression model. (a,c) is the tuning parameter (λ) in the LASSO model selected by ten cross-validations based on the minimum criterion. The lower x axis represents log (λ), the upper x axis represents the number of elements, and the y axis represents the binomial deviation. The dashed vertical line indicates the deviation value for each model with a given λ. The vertical dashed black line indicates the optimal value of λ. When the optimal λ is selected as 0.0008 in (a), log (λ) value is −3.13, and when the optimal λ is selected as 0.0031 in (c), log value (λ) is −2.506352. (b,d) Shows the LASSO coefficient profiles of 530 radiomics features. (b) Shows 15 selected features with non-zero coefficients in the T2WI-FS sequence and (d) shows 8 selected features with non-zero coefficients in the CET1 sequence.

Fig. 4

Violin plot of 8 screened radiomics features based on CET1 sequence: OS represents osteosarcoma and CS represents chondrosarcoma. The violin plot analysis compared the difference and distribution of 8 radiomics features extracted from CET1 sequences of OS and CS by LASSO method.

Fig. 5

Violin plot of 15 screened radiomics features based on T2-FS sequence: OS represents osteosarcoma and CS represents chondrosarcoma. The violin plot analysis compared the difference and distribution of 15 radiomics features extracted from T2-FS sequences of OS and CS by LASSO method.

Fig. 6

ROC curves for the training and validation groups of the radiomics model. ROC curves for the T2WI-FS and CET1 based radiomics models for OS and CS in the training (a) and validation (b) groups. The AUC of CET1 and T2WI-FS in the training group was 0.894 (95% CI 0.817–0.970) and 0.970 (95% CI 0.940–0.999), respectively; and the AUC in the validation group was 0.821 (95% CI 0.642–1.000) and 0.899 (95% CI 0.785–1.000), respectively.