Differentiation of recurrent glioblastoma from radiation necrosis using diffusion radiomics with machine learning model development and external validation

Abstract

The purpose of this study was to establish a high-performing radiomics strategy with machine learning from conventional and diffusion MRI to differentiate recurrent glioblastoma (GBM) from radiation necrosis (RN) after concurrent chemoradiotherapy (CCRT) or radiotherapy. Eighty-six patients with GBM were enrolled in the training set after they underwent CCRT or radiotherapy and presented with new or enlarging contrast enhancement within the radiation field on follow-up MRI. A diagnosis was established either pathologically or clinicoradiologically (63 recurrent GBM and 23 RN). Another 41 patients (23 recurrent GBM and 18 RN) from a different institution were enrolled in the test set. Conventional MRI sequences (T2-weighted and postcontrast T1-weighted images) and ADC were analyzed to extract 263 radiomic features. After feature selection, various machine learning models with oversampling methods were trained with combinations of MRI sequences and subsequently validated in the test set. In the independent test set, the model using ADC sequence showed the best diagnostic performance, with an AUC, accuracy, sensitivity, specificity of 0.80, 78%, 66.7%, and 87%, respectively. In conclusion, the radiomics models models using other MRI sequences showed AUCs ranging from 0.65 to 0.66 in the test set. The diffusion radiomics may be helpful in differentiating recurrent GBM from RN..

Quality of Radiomics Research on Brain Metastasis: A Roadmap to Promote Clinical Translation

“…Radiomics involves the exploitation of MRI data to extract high-dimensional quantitative imaging features, which can be used to support clinical decision-making [ 5 6 ]. Although previous radiomics studies in the field of neuro-oncology have mainly focused on gliomas [ 7 8 9 10 11 12 ], there has been an increase in the number of studies on brain metastases. Indeed, radiomics studies have demonstrated promising results in the discrimination of brain metastasis from other tumors [ 13 14 15 16 17 18 19 20 ], identification of primary tumor types in patients with brain metastases [ 21 22 23 24 ], prediction of specific genetic mutations [ 25 26 27 28 29 ], prediction of survival [ 30 31 ], differentiation between radiation necrosis and brain metastasis [ 32 33 34 35 ], and prediction of outcome after radiosurgery [ 36 37 38 39 40 41 ].…”

Radiomics-Based Detection of Radionecrosis Using Harmonized Multiparametric MRI

“…This result seems more consistent with clinical observations since conventional radiological features based on CE-T1w and FLAIR MRI have been shown to have low sensitivity in discriminating radionecrosis from recurrence [ 29 , 30 ]. In [ 13 ], several radiomic models based on CE-T1w, T2, and diffusion images were investigated to differentiate radionecrosis from tumor progression in patients treated with radiotherapy. In the study by Park et al, the model with the best discriminating performance was based on diffusion images.…”

“…Recently, radiomics analysis has been shown to provide useful insights for decision-making in patients bearing GBM under therapies [ 10 , 11 , 12 , 13 ]. While neuroradiologists routinely use perfusion and diffusion images to identify radionecrosis, none of these studies investigated such sequences, instead using anatomical images only (CE-T1w and/or FLAIR images) to detect radionecrosis automatically.…”

Systemic Therapy Type and Timing Effects on Radiation Necrosis Risk in HER2+ Breast Cancer Brain Metastases Patients Treated With Stereotactic Radiosurgery

“…Specifically for PET, fluciclovine is an amino acid radiotracer that has recently demonstrated good initial results for distinguishing RN from recurrent tumor among patients with brain metastases who were treated with SRS ( 28 , 29 ). Furthermore, there also have been efforts to use radiomics and data mining to develop models that can effectively differentiate tumor from RN ( 30 ).…”