European Radiology: Radiomics makes the identification of multiple myeloma and metastases no longer a headache!

Multiple myeloma (MM) and metastases are the two most common vertebral tumors , both of which can present as soft tissue masses or vertebral destruction
.

Due to the similar age of onset and overlapping symptoms and imaging findings, clinical misdiagnosis may occur

Multiple myeloma (MM) and metastases are the two most common vertebral tumors , both of which can present as soft tissue masses or vertebral destruction

Radiomics, an emerging imaging technology at this stage , can non-invasively assess tumors and their microenvironment and quantify spatial heterogeneity

Recently, a study published in the journal European Radiology used the most common imaging features to establish a vertebral MRI-based radiomic model for the identification of MM and metastases , and compared and analyzed the performance of the model with different EPV values.
, which provides technical support for rapid and accurate clinical diagnosis and treatment .

This study retrospectively analyzed conventional MRI images (T1WI and fat-suppressed T2WI) of 103 MM patients and 138 patients with metastases

The AUC of the 20EPV-model, 15EPV-model and CSF-model (AUC=0.

Figure ROC curves of models built based on different numbers of features in the validation set .
The AUC values ​​of the 15EPV-6-model, 10EPV-8-model and 5EPV-16-model were larger than those of the 2EPV-CFS-model (P < 0.
001) .
The AUC values ​​of the 10EPV-8-model and the 5EPV-16-model were comparable (P = 0.
480) .
 

Figure ROC curves of models built based on different numbers of features in the validation set .

The AUC values ​​of the 15EPV-6-model, 10EPV-8-model and 5EPV-16-model were larger than those of the 2EPV-CFS-model (P < 0.
001) .
The AUC values ​​of the 10EPV-8-model and the 5EPV-16-model were comparable (P = 0.

The radiomics model developed in this study based on conventional vertebral MRI (T1WI and FS-T2WI) can distinguish MM from metastases well .

Original source :

Jianfang Liu , Wei Guo , Piaoe Zeng ,et al .

Vertebral MRI-based radiomics model to differentiate multiple myeloma from metastases: influence of features number on logistic regression model performance.
DOI: 10.
1007/s00330-021-08150-y

Jianfang Liu Wei Guo Piaoe Zeng ,et al 10.
1007/s00330-021-08150-y 10.
1007/s00330-021-08150-yLeave a message here