Deep Learning Features Improve The Performance Of A Radiomics Signature For Predicting KRAS Status In Patients With Colorectal Cancer

Background: Colorectal cancer(CRC),the world’s third most common cancer,accounts for about 1 million new cases per year.Although epidermal growth factor receptor(EGFR)inhibitor therapy has significant benefit for patients with CRC,about 40-50% of patients with CRC have mutations in the KRAS gene,which encodes an important component of the EGFR signaling cascade,such mutations make CRC essentially insensitive to EGFR inhibitor therapy.Therefore,pretreatment genetic profiling of tumors can provide precious information for developing targeted therapiesPurpose: We assess the performance of a model combining a deep convolutional neural network(CNN)and a hand-crafted radiomics signature for predicting KRAS status in patients with colorectal cancer(CRC).Materials and Methods: The primary cohort consisted of 279 patients with clinicopathologically confirmed CRC between April 2011 and April 2015.An independent validation cohort of 119 patients from May 2015 to April 2016 was used to confirm the combined model’s predictive performance.Portal venous phase computed tomographic(CT)images were analyzed to extract traditional hand-crafted radiomics features as well as deep learning features.A Wilcoxon rank sum test,the minimum redundancy maximum relevance(m RMR)algorithm,and multivariable logistic regression analysis were used to select features and build a radiomics signature.A combined model was then developed using multivariable logistic regression analysis.The goodness fit of the model was measured by the C-index.A calibration curve was used to show the consistency between observed status and predicted status.To investigate the clinical usefulness of combined model,we adopted decision curve analysis to estimate the standard net benefits(s NB)at different threshold probabilities.Results: In total,2634 hand-crafted features and 2208 deep learning features were extracted from the ROIs of each patients in the primary cohort.Four hand-crafted features and six deep learning features with high predictive value for KRAS status in CRC were selected in primary dataset.The C-index of hand-crafted radiomics signature’s discriminative ability was 0.719(95% confidence interval,CI: 0.658-0.776)for the primary cohort and 0.720(95% CI: 0.625-0.813)for the validation cohort.The C-index of the deep radiomics signature’s discriminative ability was 0.754(95% CI: 0.696-0.813)for the primary cohort and 0.786(95% CI: 0.702-0.863)for the validation cohort.The combined model,which merged the handcrafted radiomics features and deep radiomics features,achieve a C-index of 0.815(95% CI: 0.766-0.868)for the primary cohort and 0.832(95% CI: 0.762-0.905)for the validation cohort.Conclusions: This study presents a model that incorporates the hand-crafted and deep radiomics signature,which can be used for individualized preoperative prediction of KRAS mutations in patients with CRC.