A predictive model for pain response following radiotherapy for treatment of spinal metastases

Abstract

To establish a predictive model for pain response following radiotherapy using a combination of radiomic and clinical features of spinal metastasis. This retrospective study enrolled patients with painful spine metastases who received palliative radiation therapy from 2018 to 2019. Pain response was defined using the International Consensus Criteria. The clinical and radiomic features were extracted from medical records and pre-treatment CT images. Feature selection was performed and a random forests ensemble learning method was used to build a predictive model. Area under the curve (AUC) was used as a predictive performance metric. 69 patients were enrolled with 48 patients showing a response. Random forest models built on the radiomic, clinical, and ‘combined’ features achieved an AUC of 0.824, 0.702, 0.848, respectively. The sensitivity and specificity of the combined features model were 85.4% and 76.2%, at the best diagnostic decision point. We built a pain response model in patients with spinal metastases using a combination of clinical and radiomic features. To the best of our knowledge, we are the first to examine pain response using pre-treatment CT radiomic features. Our model showed the potential to predict patients who respond to radiation therapy.

Application of Artificial Intelligence Methods for Imaging of Spinal Metastasis

“…Considering these risks, it will be helpful to identify those who have a high likelihood of pain relief from radiotherapy and direct other patients to alternate therapies. Wakabayashi K et al [ 98 ] developed a radiomics model utilising feature subsets, random forests and recursive feature exclusion to predict pain response post-radiotherapy for vertebral metastases. Their study concluded that the model incorporating both clinical and radiomics features was the most effective in predicting response to pain following radiotherapy with an AUC of 0.85 and accuracy of 82.6%, and this was significantly improved ( p = 0.044) when compared to the model only including clinical features with an AUC of 0.70 and accuracy of 65.2%.…”

Differentiation between spinal multiple myeloma and metastases originated from lung using multi-view attention-guided network

“…Compared with current mainstream imaging methods based on visual assessment, machine learning has provided an objective tool in many clinic tasks such as differential diagnosis, assessment of therapeutic response and prognosis. In the field of spinal tumor, radiomics shows its effectiveness in distinguishing spinal primary or metastatic tumor ( 6 , 26 , 27 ), prediction of gene mutation ( 28 , 29 ), early reoccurrence ( 27 ), condition monitoring ( 30 ) and assessment of treatment response ( 31 , 32 ). However, there are limited studies focusing on the differential diagnosis between spinal MM and metastases.…”

Radiomics-based machine learning models to distinguish between metastatic and healthy bone using lesion-center-based geometric regions of interest

“…Our rationale for this approach was that by aggregating features extracted from ROIs with various sizes around the BM centers, we could extract sufficient information about the BMs’ shape, size, and other characteristics and distinguish them from HBs using ML classifiers. Similar feature aggregation approaches were used in other studies 36 , 37 .…”