Related Experiment Video
Updated: Jul 16, 2026
09:06
Whole-brain Segmentation and Change-point Analysis of Anatomical Brain MRI—Application in Premanifest Huntington's Disease
Published on: June 9, 2018
PANDA: Patch-based unsupervised deep learning for brain anomaly detection via age prediction in fetal MRI
Yingqi Hao1,2, Mingxuan Liu2, Juncheng Zhu1,3,4
1Department of Radiology, West China Second University Hospital, Sichuan University, Chengdu, Sichuan Province, China.
Imaging Neuroscience (Cambridge, Mass.)
|July 15, 2026
Summary
This study introduces PANDA, a novel framework for detecting fetal brain anomalies using MRI scans. PANDA enhances anomaly detection by analyzing age differences in brain MRI patches, improving diagnostic accuracy for various conditions.
Area of Science:
- Medical imaging analysis
- Artificial intelligence in healthcare
- Fetal neurodevelopmental disorders
Background:
- Fetal brain anomalies have diverse origins and can lead to severe morbidities.
- Unsupervised anomaly detection offers a promising avenue for automated diagnosis of fetal brain abnormalities.
- Previous methods using absolute age difference (AAD) showed potential but limited performance.
Purpose of the Study:
- To introduce a novel 3D Patch-based brain ANomaly Detection framework via Age prediction (PANDA) for enhanced fetal brain anomaly detection.
- To utilize the maximum AAD across all patches (MaxAAD) as a biomarker for identifying fetal brain anomalies.
- To evaluate PANDA's diagnostic performance on a large clinical cohort.
Main Methods:
- Development of the PANDA framework utilizing 3D image patches.
- Calculation of the maximum absolute age difference (MaxAAD) across all patches as a biomarker.
- Experimental validation on a large dataset of 1,316 fetal MRI scans (711 normal, 605 abnormal).
Main Results:
- PANDA achieved a high diagnostic performance with an AUROC of 0.762 and AUPR of 0.790.
- The framework demonstrated superior performance across subgroup analyses for ventriculomegaly, GMH-IVH, and SEC.
- MaxAAD proved effective as a biomarker for detecting fetal brain anomalies.
Conclusions:
- The PANDA framework significantly enhances the detection of fetal brain anomalies using MRI.
- PANDA offers a robust and accurate approach for automated diagnosis in clinical settings.
- The MaxAAD biomarker shows promise for identifying deviations from normal fetal brain development.
