Purpose: The aim of this study was to propose and investigate gray level histogram and texture features information provided by 18F-FDG PET to assess patients imaging response to radiochemotherapy in non-small cell lung cancer (NSCLC). Methods: Twelve patients with newly diagnosed NSCLC treated with combined radiochemotherapy were involved in this study. Patients were categorized under three headings (non-responders, partial responders and complete responders) by experienced radiologists on the basis of RECIST according PET scans changes between pretreatment and 1 month after treatment. We analyzed the percentage variation of PET density using histogram analysis approach which characterizes global change of tumor region on PET. Texture parameters variation between pretreatment and 1 month after treatment completion which describe local voxel spatial distribution were extracted from Gray Level Co-occurrence Matrix (GLCM). Correlation between characteristics variation and three type response status were analyzed. Results: The uniformity of gray level histogram on the whole and the maximum percentage decrease in histogram was well associated with tumor shrinkage and response status. The above indices derived from histogram were capable to differentiate three groups tumor response to radiochemotherapy. Texture parameters variation (ASM, ENT and IDM) were able to differentiate the 3 response groups considering a high correlation with response status. Conclusion: We demonstrated that histogram and texture analysis methods on baseline 18F-FDG PET scans provided robust, discriminative stratification in assessing response to combined radiochemotherapy and may have a good application prospect in clinical practice.
Read full abstract