Journal: Biology Methods & Protocols

Article Title: Deep learning and transfer learning for brain tumor detection and classification

doi: 10.1093/biomethods/bpae080

Figure Lengend Snippet: Comparison of the proposed model to other brain tumor classification models, including explainability methods.

Article Snippet: The GradCAM (gC) MATLAB function was used to generate heat maps overlaying select images of each training category representing the intensity of the network’s sensitivity to a given area [ ].

Techniques: Comparison, Biomarker Discovery

Journal: Biology Methods & Protocols

Article Title: Deep learning and transfer learning for brain tumor detection and classification

doi: 10.1093/biomethods/bpae080

Figure Lengend Snippet: Flow chart showing pipeline of study methods, starting from dataset acquisition (left), through network training, and finally XAI interpretation of the network’s internal state. Important regions of the MRI are shown by GradCAM, while feature visualization and representation are shown by DDI and feature space

Article Snippet: The GradCAM (gC) MATLAB function was used to generate heat maps overlaying select images of each training category representing the intensity of the network’s sensitivity to a given area [ ].

Techniques:

Journal: Biology Methods & Protocols

Article Title: Deep learning and transfer learning for brain tumor detection and classification

doi: 10.1093/biomethods/bpae080

Figure Lengend Snippet: Image sensitivity maps for T1Net and ExpT1Net showing transfer learning effect. Sensitivity maps were generated using GradCAM, on both softmax and fully connected layers. The underlying sample images are the same as those from Figure 5 . Warmer hues (more red) indicate higher saliency, or more importance to classification outcome

Article Snippet: The GradCAM (gC) MATLAB function was used to generate heat maps overlaying select images of each training category representing the intensity of the network’s sensitivity to a given area [ ].

Techniques: Generated

Journal: Biology Methods & Protocols

Article Title: Deep learning and transfer learning for brain tumor detection and classification

doi: 10.1093/biomethods/bpae080

Figure Lengend Snippet: Image sensitivity maps for T2Net and ExpT2Net showing transfer learning effect. Sensitivity maps were generated using GradCAM, on both softmax and fully connected layers. The underlying sample images are the same as those from Figure 5 . Warmer hues (more red) indicate higher saliency, or more importance to classification outcome

Article Snippet: The GradCAM (gC) MATLAB function was used to generate heat maps overlaying select images of each training category representing the intensity of the network’s sensitivity to a given area [ ].

Techniques: Generated