Multicenter Evaluation of AI-generated DIR and PSIR for Cortical and Juxtacortical Multiple Sclerosis Lesion Detection.

Background Cortical multiple sclerosis lesions are clinically relevant but inconspicuous at conventional clinical MRI. Double inversion recovery (DIR) and phase-sensitive inversion recovery (PSIR) are more sensitive but often unavailable. In the past 2 years, artificial intelligence (AI) was used to generate DIR and PSIR from standard clinical sequences (eg, T1-weighted, T2-weighted, and fluid-attenuated inversion-recovery sequences), but multicenter validation is crucial for further implementation. Purpose To evaluate cortical and juxtacortical multiple sclerosis lesion detection for diagnostic and disease monitoring purposes on AI-generated DIR and PSIR images compared with MRI-acquired DIR and PSIR images in a multicenter setting. Materials and Methods Generative adversarial networks were used to generate AI-based DIR (n = 50) and PSIR (n = 43) images. The number of detected lesions between AI-generated images and MRI-acquired (reference) images was compared by randomized blinded scoring by seven readers (all with >10 years of experience in lesion assessment). Reliability was expressed as the intraclass correlation coefficient (ICC). Differences in lesion subtype were determined using Wilcoxon signed-rank tests. Results MRI scans of 202 patients with multiple sclerosis (mean age, 46 years ± 11 [SD]; 127 women) were retrospectively collected from seven centers (February 2020 to January 2021). In total, 1154 lesions were detected on AI-generated DIR images versus 855 on MRI-acquired DIR images (mean difference per reader, 35.0% ± 22.8; P < .001). On AI-generated PSIR images, 803 lesions were detected versus 814 on MRI-acquired PSIR images (98.9% ± 19.4; P = .87). Reliability was good for both DIR (ICC, 0.81) and PSIR (ICC, 0.75) across centers. Regionally, more juxtacortical lesions were detected on AI-generated DIR images than on MRI-acquired DIR images (495 [42.9%] vs 338 [39.5%]; P < .001). On AI-generated PSIR images, fewer juxtacortical lesions were detected than on MRI-acquired PSIR images (232 [28.9%] vs 282 [34.6%]; P = .02). Conclusion Artificial intelligence-generated double inversion-recovery and phase-sensitive inversion-recovery images performed well compared with their MRI-acquired counterparts and can be considered reliable in a multicenter setting, with good between-reader and between-center interpretative agreement. Published under a CC BY 4.0 license. Supplemental material is available for this article. See also the editorial by Zivadinov and Dwyer in this issue.

Identification and Characterization of Leptomeningeal Metastases Using SPINE, A Web-Based Collaborative Platform.

BACKGROUND AND PURPOSE: Leptomeningeal metastases (LMs) carry a poor prognosis. Existing LM scoring systems show limited reproducibility. We assessed the contribution of education level on the reproducibility of LM scoring using structured planning and implementation of new experiments (SPINE), a novel web-based platform. METHODS: Stringent radiological definitions of LM and a customized interactive scoring system were implemented in SPINE. Five patients with brain LM and 3 patients with spine, but no brain LM, were selected. Each patient's baseline post-contrast T1-weighted brain MRI was analyzed by three attending neuroradiologists, two neuroradiology fellows, and two radiology residents. Raters identified and characterized all LMs based on: (1) location (cerebrum, cerebellum, brainstem, ventricle, and/or cranial nerves); (2) shape (nodular and/or linear/curvilinear); (3) size (≥ or <5mm in two orthogonal diameters); (4) spatial extension (focal or diffuse). Inter-rater agreement and association of LM with patient survival were investigated. RESULTS: On average, 6.5 LMs per case were detected. Forty-nine percent of LMs were cerebral, 77.7% were nodular, 86.6% were focal, and 66% were <5 × 5 mm. Agreement on the total number of LMs and the above-mentioned common LM characteristics was higher between attendings (intra-class correlation [ICC] = 0.8-0.94) than fellows (ICC = 0.6-0.82) or residents (ICC = 0.43-0.73). Agreement on ventricular, cranial nerve, and nodular + linear LM was low even between attendings. The number of brainstem LMs showed significant correlation with survival. CONCLUSION: Structured education using SPINE may improve consistency in LM reporting. Future work should address the impact of the presented approach on the reproducibility of longitudinal analyses directly relevant to the assessment of treatment-response.