Comparative Analysis of State of the Art Deep Learning Models for Lung Cancer Detection

Abstract

This study is based on the reviews recent advancements in deep learning (DL) techniques applied to lung cancer detection from 2021 to 2025. Models such as lungs cancer prediction-Convolutional Neural Networks (LCP-CNN), Inception V3, EfficientNet-B3, and Convolutional Neural Networks Long Short-term Memory (CNN–LSTM) are achieved previous development model accuracies in between 86% and 99%, using deep learning techniques for diagnostic methods. Public datasets, including Kaggle and LIDC-IDRI, were most frequently used for train, test and validation, supporting model generalization and reliability. This Research trends show a peak in 2023 with increased use of hybrid and ensemble deep learning model integrating (CNN) and Vision Transformers (Vitis). Overall, the study concludes that deep learning based diagnostic systems significantly improve the accuracy and automation of lung cancer detection, reducing diagnostic errors and supporting early medical intervention. In this study the Convolutional Neural Network techniques has achieved the highest accuracy from other deep learning techniques.