Research on liver cancer segmentation method based on PCNN image processing and SE-ResUnet
Lan Zang1,2, Wei Liang1,2, Hanchu Ke3
1State Key Laboratory of Marine Resource Utilization in South China Sea, Hainan University, Haikou, 570228, China.
Scientific Reports
|August 7, 2023
Summary
This study introduces an improved Unet model for more accurate liver cancer segmentation. The enhanced model, utilizing Pulse Coupled Neural Network and SE modules, shows superior performance in identifying cancerous regions.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Liver cancer presents with subtle initial symptoms and complex anatomy, complicating accurate diagnosis.
- Early and precise detection of liver cancer is crucial for improving patient outcomes.
Purpose of the Study:
- To develop an advanced image segmentation model for enhanced liver cancer detection.
- To improve the accuracy and performance of liver cancer segmentation compared to existing methods.
Main Methods:
- A variant Unet network model was proposed for liver cancer segmentation.
- Image preprocessing involved adaptive filtering using the Pulse Coupled Neural Network (PCNN) algorithm.
- The segmentation model integrated the SE module with a residual network and employed bilinear interpolation for Unet's operations.
Main Results:
- The proposed model demonstrated superior segmentation performance and accuracy over the original Unet.
- Key evaluation metrics, including the Dice coefficient and mIoU, showed improvements of at least 2.1%.
Conclusions:
- The developed Unet-based model offers a promising approach for accurate liver cancer segmentation.
- This method has practical applications in clinical settings for aiding physicians in liver cancer diagnosis.
