Discovery Radiomics for Computed Tomography Cancer Detection

Objective: Lung cancer is the leading cause for cancer related deaths. As such, there is an urgent need for a streamlined process that can allow radiologists to provide diagnosis with greater efficiency and accuracy. A powerful tool to do this is radiomics. Method: In this study, we take the idea of radiomics one step further by introducing the concept of discovery radiomics for lung cancer detection using CT imaging data. Rather than using pre-defined, hand-engineered feature models as with current radiomics-driven methods, we discover custom radiomic sequencers that can generate radiomic sequences consisting of abstract imaging-based features tailored for characterizing lung tumour phenotype. In this study, we realize these custom radiomic sequencers as deep convolutional sequencers using a deep convolutional neural network learning architecture based on a wealth of CT imaging data. Results: To illustrate the prognostic power and effectiveness of the radiomic sequences produced by the discovered sequencer, we perform a classification between malignant and benign lesions from 93 patients with diagnostic data from the LIDC-IDRI dataset. Using the clinically provided diagnostic data as ground truth, proposed framework provided an average accuracy of 77.52% via 10-fold cross-validation with a sensitivity of 79.06% and specificity of 76.11%. We also perform quantitative analysis to establish the effectiveness of the radiomics sequences. Conclusion: The proposed framework outperforms the state-of-the art approach for lung lesion classification. Significance: These results illustrate the potential for the proposed discovery radiomics approach in aiding radiologists in improving screening efficiency and accuracy.