A Robust Predictive Model Using Combined Residual Network and Long Short – Term Memory for Lung Cancer Identification

Lung cancer is the primary cause of cancer mortality in many nations. Both benign and malignant tumors exist. Malignant tumors are meant to be referred to as "cancer" here. Lung cancer has several potential origins, such as secondhand smoke, active smoking, chemical exposure, and genetics. In this study, we present a Residual Network (ResNet)/Long Short-Term Memory (LSTM) hybrid model for lung cancer categorization on CT scan pictures, and we compare the suggested model's effectiveness to that of UNet Architecture. The ability to accurately identify or forecast early-stage lung cancer is crucial. Future medical expenses might be cut significantly as a result. The suggested ResNet - LSTM hybrid technique aims to enhance overall accuracy in lung cancer prediction. It can make surgery safer or perhaps unnecessary, and it has been shown to improve survival rates. Thus, technologies for identifying lung cancer are not only easy to use, cheap, and convenient, but also very successful methods for initial identification of lung cancer. As a result, the network is crucial to making an accurate diagnosis. According to our findings, the steps involved in the detection process are as follows: image preprocessing, image segmentation, extraction of features, and finally, identifying the neural system. The primary goals of this study are to (1) compare and contrast different methods to identify lung cancer, and (2) analyze the current best strategy. The ResNet-LSTM algorithm has shown outstanding accuracy, suggesting that its usage in lung cancer identification might improve detection rates. Lung cancer sufferers may find it simpler to detect and treat their condition if the cancer can be predicted at an earlier stage. We discovered that the accuracy of a hybrid ResNet-LSTM model for diagnosing lung cancer was 97%, whereas that of a standard UNet Model was 93%. In comparison to the previous model, the new one had better accuracy, precision, and recall. Clinicians can improve the accuracy and timeliness of lung nodule detection and classification using medical imaging technologies powered by deep learning. This study details the most up-to-date imaging methods for spotting lung cancer in its earliest stages using deep learning. When compare to the most recent state-of-the-art methods, the suggested method's effectiveness and dependability shine through in experimental findings.