Design and in vitro simulations of nanocomposite drug delivery systems for use in brain tumor treatments

The aim of this study was to develop a nano-drug system that can be used for passive targeting in brain tumor treatment. Hydroxyapatite nanocrystals (n-HAP) produced by hydrothermal process and graphene oxide (GO) produced by hummers method were used to increase the carrier capacity of the nano drug system and to activate drug release kinetics and drug loading capacity. Analyses for nanocomposite drug carrier systems were performed by FTIR, XRD and SEM. Paclitaxel (PTX), a chemotherapeutic drug used in the treatment of brain cancer, was loaded into HAP nanocrystals and its activity on brain cancer cells was investigated. When PTX was 1 and 2 mg, the efficiency (EE) and loading (LC) capacities were 79.17-72.24% and 80.01-80.27% for the H-n-HAP crystal structure alone, respectively, whereas when 1 and 2 mg of PTX was loaded into the H-n-HAP crystal structure with GO, the EE and LC were 88.57-81.57% and 90.84-110.57%, respectively. Here, PTX release profiles were found to be in accordance with the Hixson model. According to Fick's law, the release profile was observed with k=1.89, n=0.21, SSD=0.04, R2=0.997, FIC=2.03, SD=0.004. For H-n-HAP/GO/PTX, 59.8% was found at 10% concentration, 46.6% for PTX and 49.5% for H-n-HAP/GO/PTX at 50% concentration. An application that automatically detects brain tumors from MRI images using these developed nanocomposite smart drug delivery systems can save time for healthcare professionals, as well as allowing earlier diagnosis of cancer and minimizing the human error rate. In this thesis, Classical U-Net, ResNet50 Convolutional Neural Networks are applied for brain tumor detection. In addition, tumor segmentation was performed with these methods and predictions were generated. The results are given in terms of accuracy, dice coefficient, sensitivity, specificity, necrotic dice coefficient, edema dice coefficient, enhancing dice coefficient and these ratios are compared.