An automatic lung nodule detection and classification using an optimized convolutional neural network and enhanced k-means clustering

Lung cancer can be lethal if it is not found in the initial phases. Lung cancer, nevertheless, is challenging to identify early due to the dimensions and form of the nodules. Imaging specialists require the assistance of automated instruments for accurate interpretation. Therefore, in this paper, automatic lung nodule classification is proposed. In the proposed methodology, there are four phases, namely pre-processing, segmentation, classification, and severity analysis. As a first step, lung nodule images are collected from the dataset and pre-processed. After pre-processing, segmentation is carried out. For segmentation enhanced k-means clustering algorithm is applied. To identify the object as cancerous or benign, the segmented region is then fed into an optimized convolution neural network (OCNN). Here, adaptive sunflower optimization (ASFO) algorithm is used to pick the hyperparameters effectively to improve the effectiveness of the CNN classifier. Finally, from the segmented region, the severity of the patient is evaluated. The efficiency of the presented technique is analyzed based on various metrics and research work compared with different state-of-art-works. © 2023, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.