Transfer Learning With Adam Gold Rush Optimization for Endometrial Disease Classification Using Histopathological...

Area of Science:

• Oncology
• Medical Imaging
• Artificial Intelligence

Background:

• Endometrial cancer, a significant condition affecting female reproductive organs, necessitates early and accurate diagnosis for improved survival rates.
• Current diagnostic methods for endometrial cancer can be enhanced through advanced computational techniques applied to histopathological images.

Purpose of the Study:

• To propose a novel Transfer Learning based Convolution Neural Network with Adam Gold Rush Optimization (TL-CNN_AdGRO) for the classification of endometrial cancer.
• To evaluate the performance of the proposed TL-CNN_AdGRO model in accurately identifying endometrial cancer from histopathological images.

Main Methods:

• Histopathological images undergo preprocessing using an Adaptive Weighted Mean Filter (AWMF).
• Endometrial cancer segmentation is performed using Directional Connectivity Network (DConn-Net).
• Feature extraction includes Local Boundary Summation Pattern (LBSP) and Local Gaber Binary Pattern Histogram Sequence Features (LGBPHS), followed by classification using TL-CNN trained with the AdGRO algorithm.

Main Results:

• The proposed TL-CNN_AdGRO model achieved a high accuracy of 91.876%.
• Superior performance metrics include a True Positive Rate (TPR) of 93.987% and a True Negative Rate (TNR) of 89.876% (K-sample 8).
• The model demonstrated robustness and effectiveness compared to existing methods.

Conclusions:

• The TL-CNN_AdGRO model shows significant promise for the early detection of endometrial cancer.
• This approach offers a robust and effective method for histopathological image analysis in oncology.
• The findings support the clinical utility of advanced AI models for improving cancer diagnostics.