Özet:

Abstract The classification of Brain tumors is fundamental for the finding of Brain Cancer (BC) in medical services frameworks. AI (artificial intelligence) methods in light of PC-supported analytic frameworks (Miscreants) are generally utilized for the exact location of brain tumors. In any case, because of the issues like the mistake of counterfeit demonstrative frameworks, clinical experts are not really integrating them into the conclusion cycle of Brain tumors (BT). As deep learning (DL) technology got revolutionized greatly in medical fields, the usage of such ideas brings more effectiveness in terms of performance and accuracy. With that said, this research work offers an efficient deep learning-based categorization of brain tumors, with the steps being as follows: a) Data collecting from well-known databases for the brain, lung, and liver, which together comprise 10,000 records, b) Preprocessing using CLAHE ( for brightness enhancing), Thresholding (Grayscale), Filtering (ADF) and skull masking for removal of noise and anomalies from raw images, c) feature extraction using Principle Component Analysis (PCA), d) feature selection using VGG16 network and finally   e) classification using Deep Dense Neural Network (Densenet 164). Experimental tests indicates that the proposed model outperforms better than other state-of-art models under different measures (accuracy: 0.97, sensitivity: 0.98, specificity: 0.98, detection rate: 0.94).

Anahtar Kelimeler: