Efficient radio surgery segmentation of different brain tumors using deep learning

Abstract

Author(s): Vijay Upadhye, Tarang Bhatnagar, Avadhesh Kumar, Iype Cherian, Sandeep Kumar C, Shikhar Gupta

Background: A feature of medical image processing poses a significant barrier in brain tumor segmentation. Patients' chances of receiving effective therapy and of surviving their brain tumors are greatly enhanced by early diagnosis. Separating brain tumors by hand from the enormous amount of MRI images produced in clinical practice to make a cancer diagnosis is a difficult and time-consuming task. Deep Learning (DL) models are being used to categorize medical pictures semantically at a quick pace.

Methods: In this work, used clinical stereotactic radio surgical dataset as a baseline to compare cutting-edge DL segmentation methods. Herein, proposed a novel DL based Hyper Automated Densely Fused Convolutional Neural Network (HADF-CNN) approach for brain tumor segmentation. To analyze the efficiency of the proposed method, initially, data were collected and preprocessed using Median Filter (MF) to remove unwanted noise. Next, the brain data is segmented using the suggested methodology. The categorization of brain tumors is done using the Deep Support Vector Machine (DSVM).

Results: To compared the proposed model performance with certain conventional segmentation methods. According to experimental data, the suggested approach performs better than traditional approaches.

Conclusion: Based on the data in this study, might draw the conclusion that DL has potential to help with brain lesion segmentation.

Share this article