Pulse diagnosis has been a requisite facet in traditional Chinese medicine as well as in western medicine, yet the prognosis of lung cancer hinged on wrist pulse analysis entails the quotidian approaches. In spite of diagnosing the lung cancer, in traditional methods, the identification stratagem is divaricated into assorted steps: analysis of signals procured, synthetic extraction and selection of features, and subsequently the classification. However, the vague and mundane feature selection and signal analysis steers to the inadequate classification accuracy due to intrinsic deficiencies. In this study, we have proposed a novel deep convolutional neural network (DCNN) based approach to discern the lung cancer against the acquired wrist pulse signals. In order to ensnare the features, vanquishing the overfitting, a 1- dimensional 15-layers DCNN model is devised hinged on 1-D convolutional, batch normalization, and pooling layers. Considering the instinctive feature extraction, from the experimental data comprised of 45,969 samples of 16 lung cancer and 20 healthy individuals, assorted units are heaped in the lodged DCNN. The experimental comparison with the stateof- art deep neural networks (DNNs) and conventional methods evinced that our lodged approach conquer the deficiencies of conventional signal processing and manual feature selection approaches. Finally, the results, with the validation precision of 97.67%, outperform the recent existing approach for lung cancer recognition.