Detection of Brain Tumor using Convolution Neural Network through Detection of Metastasis

• Published in: Devices for Integrated Circuit pp. 589 - 593
• Main Authors: Santra, Soumen, Sardar, Suman, Sil, Swarnadeep, Sarkar, Rudra, Deyasi, Arpan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 05.04.2025
• Subjects: Accuracy; Brain modeling; Brain tumor; Brain tumors; Convolution; Convolution neural network; Integrated circuit modeling; Neural networks; Pathology; Sensitivity; Testing; testing accuracy; Training; training accuracy; validation accuracy
• ISSN: 2996-3044
• DOI: 10.1109/DevIC63749.2025.11012428

A novel model using convolution neural network has been developed for near accurate detection of brain tumor by carefully preprocessing the data and addressing class imbalance. Dataset was gathered between December 2017 and March 2018, consists of 7,001 JPG MRI pictures, 3,800 of which show a healthy brain, and 3,201 of which show a brain with tumors; results 94.14% accuracy in training, 94.33% accuracy in testing and 95.12% accuracy in validation. Insignificant losses are reported in all the cases, speaks in favor of quality of segregation between pathological and normal tissues. Present study provides 92.39% sensitivity, 96.63% specificity and 97.03% precision which clearly outperforms peer-reviewed works carried out using CNN; and therefore, justifies novelty of the work.