Shell feature: a new radiomics descriptor for predicting distant failure after radiotherapy in non-small cell lung cancer and cervix cancer

TL;DR

This study introduces a novel shell feature derived from PET images that improves the prediction of distant failure in NSCLC and cervical cancer patients after radiotherapy, outperforming traditional radiomics features.

Contribution

The paper presents a new tumor shell feature for radiomics that enhances prediction accuracy of distant failure in lung and cervical cancers, validated with PET imaging data.

Findings

Shell feature achieved higher AUC than other features in both cancers.

Shell feature significantly outperformed traditional radiomics features.

The method demonstrated robust predictive performance across patient cohorts.

Abstract

Purpose To develop and demonstrate a novel tumor shell feature for predicting distant failure in non-small cell lung cancer (NSCLC) and cervical cancer (CC) patients. Patients and Methods The shell predictive model was constructed using pretreatment positron emission tomography (PET) images from 48 NSCLC patients received stereotactic body radiation therapy (SBRT) and 52 CC patients underwent external beam radiation therapy and concurrent chemotherapy followed with high-dose-rate intracavitary brachytherapy. A shell feature, consisting of outer voxels around the tumor boundary, was extracted from a series of axial PET slices. The hypothesis behind this feature is that non-invasive and invasive tumors may have different morphologic patterns in the tumor periphery, in turn reflecting the differences in radiological presentations in the PET images. The utility of the shell was evaluated by the support vector machine (SVM) classifier in comparison with intensity, geometry, gray level co-occurrence matrix (GLCM) based texture, neighborhood gray tone difference matrix (NGTDM) based texture, and a combination of these four features. The results were assessed in terms of accuracy, sensitivity, specificity, and the area under the receiver operating curve (AUC). Results For NSCLC, the AUC achieved by the shell feature was 0.82 while the highest AUC achieved by the other features was 0.76. Similarly, for CC, the AUC achieved by the shell feature was 0.83 while the highest AUC achieved by the other features was 0.76. Also, the difference in performance between shell and the other features was significant (P < 0.005) in all cases. Conclusions We propose a boundary-based shell feature that correlates with tumor metastasis. The shell feature showed better predictive performance than all the other features for distant failure prediction in both NSCLC and CC.