AI-enhanced simultaneous multiparametric 18F-FDG PET/MRI for accurate breast cancer diagnosis

• DOI: 10.1007/s00259-021-05492-z
• Publication Date: 2021-08-10T05:02:31Z
• Authors: V. Romeo; P. Clauser; S. Rasul; P. Kapetas; P. Gibbs; P. A. T. Baltzer; M. Hacker; R. Woitek; T. H. Helbich; K. Pinker
• Keywords: Breast MRI; McNemar's test; Multiparametric MRI

To assess whether a radiomics and machine learning (ML) model combining quantitative parameters features extracted from simultaneous multiparametric 18F-FDG PET/MRI can discriminate between benign malignant breast lesions.A population of 102 patients with 120 lesions (101 19 benign) detected on ultrasound and/or mammography was prospectively enrolled. All underwent hybrid for diagnostic purposes. Quantitative were DCE (MTT, VD, PF), DW (mean ADC contralateral parenchyma), PET (SUVmax, SUVmean, SUVminimum lesions, as well SUVmean the T2-weighted images. Radiomics DCE, T2-weighted, ADC, Different models developed using fine Gaussian support vector algorithm which explored different combinations to obtain highest accuracy in discriminating fivefold cross-validation. The performance best ML compared that expert reader review McNemar's test.Eight developed. integrated MTT images obtained cancer diagnosis (AUC 0.983), although its not significantly higher than 0.868) (p = 0.508).A accurately lesions.