ResNet for Histopathologic Cancer Detection, the Deeper, the Better?
Journal Title: Journal of Data Science and Intelligent Systems - Year 2024, Vol 2, Issue 4
Abstract
Histopathological image classification has become one of the most challenging tasks for researchers, due to the varied categories and detailed differences within diseases. In this study, we investigate the critical role of network depth in histopathological image classification, utilizing deep residual convolutional neural networks (ResNet). We evaluate the efficacy of two transfer learning strategies using ResNet with varying layers (18, 34, 50, 152) pretrained on ImageNet. Specifically, we analyze whether a deeper network or the fine-tuning of all layers in pre-trained ResNets enhances performance compared to freezing most layers and training only the last layer. Conducted on Kaggle's dataset of 220,025 labeled histopathology patches, our findings reveal that increasing the depth of ResNet does not guarantee better accuracy (ResNet-34 AUC: 0.992 vs. ResNet-152 AUC: 0.989). Instead, dataset-specific semantic features and the cost of training should guide model selection. Furthermore, deep ResNet outperforms traditional logistic regression (ResNet AUC: up to 0.992 vs. logistic regression AUC: 0.775), showcasing superior generalization and robustness. Notably, the strategy of freezing most layers doesn't improve the accuracy and efficiency of transfer learning and the performance of both transfer strategies depends largely on the types of data. Overall, both methods produce satisfactory results in comparison to models trained from scratch or conventional machine learning models.
Authors and Affiliations
Ziying Wang, Jinghong Gao, Hangyi Kan, Yang Huang, Furong Tang, Wen Li, Fenglong Yang
Keywords
Related Articles
- EP ID EP752194
- DOI 10.47852/bonviewJDSIS3202744
- Views 29
- Downloads 0
Intra-annual National Statistical Accounts Based on Machine Learning Algorithm
The methods used for forecasting financial series are based on the concept that a pattern can be identified in the data and distinguished from randomness by smoothing past values. This smoothing process eliminates random...
A Comparative Analysis of Feature Eliminator Methods to Improve Machine Learning Phishing Detection
This Machine-learning-based phishing detection employs statistical models and algorithms to assess and recognise phishing attacks. These algorithms can learn patterns and features that distinguish between phishing and no...
Bootstrap Methods for Canonical Correlation Analysis of Functional Data
The bootstrap method is a very general resampling procedure for investigating the distributional property of statistics. In this paper, we present two bootstrap methods with the aim of studying the functional canonical c...
Dynamic Neural Network Architecture Design for Predicting Remaining Useful Life of Dynamic Processes
The prediction of remaining useful life is critical in predictive health management. This is done to reduce the expenses associated with operation and maintenance by avoiding errors and failures in dynamic processes. Rec...
Chemical Engineering Numerical Analysis with R: Peng-Robinson Equation of State
Likely, many text on MATLAB, C++, FORTRAN and Python programming languages exist in chemical engineering libraries, discussing their applications for chemical engineering numerical analysis. R programming language, which...