Table 26 Evaluation on the domain expert framework.

From: A federated learning-based privacy-preserving image processing framework for brain tumor detection from CT scans

Evaluation criteria

Mean score (out of 5)

Std. Dev.

Min

Max

Interpretation

Tumor boundary accuracy

4.6

0.49

4

5

High agreement on precise tumor localization

Segmentation interpretability

4.5

0.53

4

5

Tumor regions are clearly distinguishable from healthy tissue

Classification confidence

4.2

0.63

3

5

Confidence levels are consistent with expert assessments

Image quality post-processing

4.7

0.45

4

5

Filtering techniques enhanced structural clarity effectively

Clinical usability of visual outputs

4.1

0.57

3

5

Outputs are understandable and helpful in supporting diagnosis

Overall diagnostic utility

4.4

0.51

4

5

Experts endorse framework as a strong clinical decision support