A Quantitative Multivariate Microscopic Analysis for Identifying Changes of Glioblastoma Cancer Cells due to Thermochemoradiation Therapy

Document Type : Original Article


1 Applied Biotechnology Research Center, Baqiyatallah University of Medical Sciences, Tehran, Iran

2 Department of Surgery, School of Medicine, Baqiyatallah Hospital, Baqiyatallah University of Medical Sciences, Tehran, Iran

3 Department of Medical Physics and Imaging, Urmia University of Medical Sciences, Urmia, Iran


Introduction: Although radiation is recognized as the most effective nonsurgical treatment, the outcomes and control rates are generally poor. However, a combination of radiation therapy with hyperthermia and chemotherapy can improve the efficacy of treatment. The aim was to explore the potential of morphological and gradient-based features on microscopic images in improving the identification accuracy of subtle differences in cell structure during different treatments.
Materials and Methods: Fifty single-cell images were used for each group and treatment regimen. The groups were individually subjected to: 1) hyperthermia at 43°C; 2) temozolomide (TMZ) chemotherapy at 10% inhibitory concentration; 3) radiotherapy at 2Gy; 4) combination of TMZ chemotherapy and hyperthermia; 5) combination of radiotherapy and hyperthermia; 6) combination of TMZ chemotherapy and radiotherapy; and 7) combination of TMZ chemotherapy, radiotherapy, and hyperthermia. Morphological and gradient-based features were extracted from each cell. The area under the receiver operating characteristic curve (AUC) was calculated for each significant feature to evaluate the performance of cell change detection.
Results: According to AUCs, gradient-based features showed superior performance to morphological features in identifying cell changes during all treatment regimens in all groups. In this regard, the AUC of the gradient-mean feature exceeded 0.599 for all groups. The ratio of maximum to minimum cell diameter was the best morphological feature, with an AUC above 0.588 for all groups.
Conclusions: Quantitative analysis of features is a reliable indicator of damage, with the potential to characterize cell changes during treatment regimens.