Fig 1.
Different views and types of Brain MRI [8].
Table 1.
Summary of related works.
Fig 2.
View-specific integrated segmentation and identification for optimizing NeuroDiagnosis.
Fig 3.
Main components of the proposed methodology.
Table 2.
MRI categories and views distribution.
Fig 4.
(a) Magnetic resonance image and (b) corresponding binary mask pair.
Fig 5.
Model training flowchart.
Fig 6.
Training integrated segmentation-classification model.
Fig 7.
The proposed novel combination of tumor segmentation and classification model.
It combines the fine-tuned U-net segmentation model with a classification header.
Table 3.
Performance metrics of different encoders.
Fig 8.
Confusion matrix for view classification.
Table 4.
Performance comparison of different encoders.
Table 5.
Performance comparison among competing models.
Fig 9.
Summary plots for (a) Dice score and (b) F1 score.
Table 6.
Dice and F1-score comparison across models over multiple runs.
Table 7.
Paired t-test results comparing Vision with CNN, Citnet, and Basic Unet for F1 and Dice metrics.
Fig 10.
Segmentation results for (a) axial, (b) coronal, and (c) sagittal views.
Fig 11.
Effect of Lambda on model performance.
Fig 12.
Effect of Threshold on model performance.
Table 8.
Ablation study results.
Table 9.
Effect of Single Type Single View approach on VISION model.
Table 10.
Effect of All Type Single View approach on VISION model.