Stepwise Transfer Learning for Expert-level Pediatric Brain Tumor MRI Segmentation in a Limited Data Scenario
Aidan Boyd1, Zezhong Ye1, Sanjay P Prabhu1
1From the Artificial Intelligence in Medicine (AIM) Program, Mass General Brigham, Harvard Medical School, Boston, Mass (A.B., Z.Y., Y.Z., A.Z., H.H., R.C., H.J.W.L.A., B.H.K.); Department of Radiation Oncology (A.B., Z.Y., M.C.T., Y.Z., A.Z., H.H., R.C., K.X.L., D.A.H.K., H.J.W.L.A., B.H.K.) and Department of Radiology (H.J.W.L.A.), Brigham and Women's Hospital, Dana-Farber Cancer Institute, Harvard Medical School, 75 Francis St, Boston, MA 02115; Department of Radiology, Boston Children's Hospital, Harvard Medical School, Boston, Mass (S.P.P., S.V., T.Y.P.); Department of Biostatistics and Computational Biology, Harvard T.H. Chan School of Public Health, Boston, Mass (P.J.C.); Center for Data-Driven Discovery in Biomedicine (D3b) (A.N., A.C.R.) and Department of Neurosurgery (A.C.R.), Children's Hospital of Philadelphia, Philadelphia, Pa; Department of Radiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pa (A.N.); Departments of Neurology, Pediatrics, and Neurologic Surgery, University of California, San Francisco, San Francisco, Calif (S.M.); and Department of Radiology and Nuclear Medicine, CARIM & GROW, Maastricht University, Maastricht, the Netherlands (H.J.W.L.A.).
Deep learning models using stepwise transfer learning achieved expert-level segmentation for pediatric brain tumors. This AI approach demonstrated high clinical acceptability and accuracy in MRI segmentation tasks.
Area of Science:
- Artificial Intelligence in Medical Imaging
- Oncology and Radiology
- Pediatric Neuro-oncology
Background:
- Accurate segmentation of pediatric brain tumors on MRI is crucial for diagnosis and treatment planning.
- Limited data scenarios pose challenges for developing high-performing deep learning models.
- Transfer learning offers a promising approach to enhance model performance with smaller datasets.
Purpose of the Study:
- To develop and externally validate a deep learning model for pediatric brain tumor segmentation using stepwise transfer learning.
- To evaluate the clinical acceptability of AI-generated segmentations compared to expert segmentations.
Main Methods:
- Retrospective study utilizing two T2-weighted MRI datasets from a national consortium (n=184) and a pediatric cancer center (n=100).
- Development of deep learning neural networks employing a stepwise transfer learning strategy for pediatric low-grade glioma segmentation.
- External testing and randomized blinded evaluation by three clinicians using Likert scales and Turing tests to assess clinical acceptability.
Main Results:
- The best AI model, utilizing in-domain stepwise transfer learning, achieved a median Dice score coefficient of 0.88.
- External testing demonstrated excellent accuracy, with AI segmentations yielding a mean Dice similarity coefficient of 0.82 against expert references.
- Clinicians rated AI segmentations higher on average (median Likert score 9 vs 7) and found them more clinically acceptable (80.2% vs 65.4%).
Conclusions:
- Stepwise transfer learning facilitates expert-level automated segmentation and volumetric measurement of pediatric brain tumors.
- The developed AI model exhibits a high degree of clinical acceptability, comparable to or exceeding human expert performance.
- This approach holds significant potential for improving efficiency and accuracy in pediatric neuro-oncology imaging analysis.
