Brain MRI Study Assesses Impact of AI in Differentiating Radionecrosis from Neoplastic Progression of Metastasis
Convolutional neural network-enabled segmentation of brain MRI offered a 25.7 percent higher specificity than a radiomic model for differentiating radionecrosis and metastatic progression in patients treated with stereotactic radiosurgery for brain metastases.
While overlapping features on magnetic resonance imaging (MRI) may make it challenging to differentiate between radionecrosis and metastatic progression in patients treated with stereotactic radiosurgery (SRS), artificial intelligence (AI) may have an impact in resolving this diagnostic challenge.
In a recent study presented at the Radiological Society of North America (RSNA) conference, researchers reviewed preoperative MRI data and histological findings for 96 patients (mean age of 56.4) with brain metastases who had stereotactic radiosurgery and subsequent neurosurgery for suspected recurrence or symptomatic disease recurrence. There was a total of 104 brain lesions in the cohort with 29 cases revealing sole radionecrosis, according to the study.
The use of adjunctive AI may enhance MRI capability for differentiating between radionecrosis and metastatic progression in patients treated with stereotactic radiosurgery, according to recent research presented at the RSNA conference. (Images courtesy of RSNA.)
The researchers found that the combined assessment of 3D T1 post-contrast MRI (T1c) and 3D T2 fluid-attenuated inversion recovery (FLAIR) imaging assessment via a radiomics model yielded an area under the receiver operating characteristic curve (AUROC) of 81 percent, sensitivity of 86.7 percent, accuracy of 79 percent and specificity of 60 percent.
In contrast, the study authors noted that combined assessment of T1ce and FLAIR imaging via convolutional neural network-enabled brain segmentation yielded similar sensitivity (87.8 percent) but an 8 percent higher AUROC (89 percent), 7.9 percent higher accuracy (86.9 percent) and 25.7 percent higher specificity (85.7 percent) for differentiating between radionecrosis and metastatic progression.
“Integrating artificial intelligence models with conventional MRI offers a promising non-invasive approach, potentially revolutionizing diagnostic methodologies in this challenging clinical context,” wrote lead study author Gaia Ressa, M.D., who is affiliated with the Department of Biomedical Sciences at Humanitas University in Milan, Italy, and colleagues.
(Editor’s note: For additional coverage from the RSNA conference, click here.)
While noting that post-surgical histological determination is currently the gold standard for differentiating between radionecrosis and metastatic progression, the researchers maintained that their study findings may have a significant impact upon the monitoring of patients who have undergone stereotactic radiosurgery for brain metastases.
“(Artificial intelligence models) have the potential to enhance patient management strategies and reduce unnecessary neurosurgical interventions,” emphasized Ressa and colleagues.
Reference
1. Ressa G, Levi R, Savini G, et al. Artificial intelligence accurately differentiates radionecrosis from true progression in brain metastasis treated with stereotactic radiosurgery: analysis of 104 histologically assessed lesions. Poster presented at the Radiological Society of North America (RSNA) 2024 110th Scientific Assembly and Annual Meeting Dec. 1-5, 2024. Available at: https://www.rsna.org/annual-meeting .
Considering Breast- and Lesion-Level Assessments with Mammography AI: What New Research Reveals
June 27th 2025While there was a decline of AUC for mammography AI software from breast-level assessments to lesion-level evaluation, the authors of a new study, involving 1,200 women, found that AI offered over a seven percent higher AUC for lesion-level interpretation in comparison to unassisted expert readers.
Can CT-Based Deep Learning Bolster Prognostic Assessments of Ground-Glass Nodules?
June 19th 2025Emerging research shows that a multiple time-series deep learning model assessment of CT images provides 20 percent higher sensitivity than a delta radiomic model and 56 percent higher sensitivity than a clinical model for prognostic evaluation of ground-glass nodules.
FDA Clears Ultrasound AI Detection for Pleural Effusion and Consolidation
June 18th 2025The 14th FDA-cleared AI software embedded in the Exo Iris ultrasound device reportedly enables automated detection of key pulmonary findings that may facilitate detection of pneumonia and tuberculosis in seconds.
