Discussion of a Simple Method to Generate Descriptive Images Using Predictive ResNet Model Weights and Feature Maps for Recurrent Cervix Cancer DOI Creative Commons
Destie Provenzano, Jeffrey Wang, Sharad Goyal

et al.

Tomography, Journal Year: 2025, Volume and Issue: 11(3), P. 38 - 38

Published: March 20, 2025

Background: Predictive models like Residual Neural Networks (ResNets) can use Magnetic Resonance Imaging (MRI) data to identify cervix tumors likely recur after radiotherapy (RT) with high accuracy. However, there persists a lack of insight into model selections (explainability). In this study, we explored whether features could be used generate simulated images as method explainability. Methods: T2W MRI were collected for twenty-seven women cancer who received RT from the TCGA-CESC database. Simulated generated follows: [A] ResNet was trained recurrent cancer; [B] evaluated on subjects obtain corresponding feature maps; [C] most important maps determined each image; [D] combined across all [E] final image reviewed by radiation oncologist and an initial algorithm likelihood recurrence. Results: (93% accuracy) images. passed through identified non-recurrent radiotherapy. A characteristics aggressive Cervical Cancer. These also contained multiple not considered clinically relevant. Conclusion: This simple able that mimicked tumor useful evaluating explainability predictive assist radiologists identification predict disease course.

Language: Английский

Discussion of a Simple Method to Generate Descriptive Images Using Predictive ResNet Model Weights and Feature Maps for Recurrent Cervix Cancer DOI Creative Commons
Destie Provenzano, Jeffrey Wang, Sharad Goyal

et al.

Tomography, Journal Year: 2025, Volume and Issue: 11(3), P. 38 - 38

Published: March 20, 2025

Background: Predictive models like Residual Neural Networks (ResNets) can use Magnetic Resonance Imaging (MRI) data to identify cervix tumors likely recur after radiotherapy (RT) with high accuracy. However, there persists a lack of insight into model selections (explainability). In this study, we explored whether features could be used generate simulated images as method explainability. Methods: T2W MRI were collected for twenty-seven women cancer who received RT from the TCGA-CESC database. Simulated generated follows: [A] ResNet was trained recurrent cancer; [B] evaluated on subjects obtain corresponding feature maps; [C] most important maps determined each image; [D] combined across all [E] final image reviewed by radiation oncologist and an initial algorithm likelihood recurrence. Results: (93% accuracy) images. passed through identified non-recurrent radiotherapy. A characteristics aggressive Cervical Cancer. These also contained multiple not considered clinically relevant. Conclusion: This simple able that mimicked tumor useful evaluating explainability predictive assist radiologists identification predict disease course.

Language: Английский

Citations

0