Performance Comparison to Detect Lung Cancer in Histopathological Images Using ResNet over VGG16
Syed Yamin Khan, Geetha Ramalingam · 2024
Aim: This study compares the accuracy of ResNet’s deep learning with that of the VGG16 to predict lung cancer. Materials and Methods: To delve into this research, we examined 20 samples from each of the ResNet and VGG16 groups. We aimed to assess the precision of Lung Cancer prediction based on Histopathology. We carefully examined 20 samples for each group, using a G power of 80% and an alpha value of 0.005. We carefully gathered the dataset from https://www.w3.org/1999/xlink" xlink:href="https://www.clincalc.com">clincalc.com, employing Python programming, Anaconda software, and Jupiter software. Our main focus was on using areas affected by lung cancer as data models to train the ResNet and VGG16 algorithms. Afterward, we used SPSS software to analyze the statistics, uncovering that the ResNet algorithm showed an accuracy of 82.3%, whereas the VGG16 algorithm performed less impressively at 29.16%. This discrepancy was statistically significant, with P values of 0.001 and P < 0.05. In conclusion, our findings underscored the superior accuracy of the ResNet algorithm in contrast to the VGG16 algorithm in predicting Lung Cancer based on Histopathology.