International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 05 | May 2025
p-ISSN: 2395-0072
www.irjet.net
BRAIN HEMORRHAGE DETECTION USING DEEP LEARNING N Mohan Reddy1, P Sathish2, P Sumanth3, Poonamalli Lokesh4, Mr. L Vijaya Kumar5, Prof. M. E. Palanivel6 1234 UG Scholar, Dept. of CSE(AI), Sreenivasa Institute of Technology and Management Studies, Chittoor, A.P, India
5 Asst. Professor, Dept. of CSE(AI), Sreenivasa Institute of Technology and Management Studies, Chittoor, A.P, India 6 Professor, Dept. of CSE(AI), Sreenivasa Institute of Technology and Management Studies, Chittoor, A.P, India
---------------------------------------------------------------------***--------------------------------------------------------------------Abstract - Brain hemorrhage is a life-threatening medical condition that necessitates rapid and accurate diagnosis to prevent severe complications or fatalities. Current detection methods, such as manual analysis of CT scans by radiologists, are often time-intensive and susceptible to human error, especially in high-pressure emergency situations. This project aims to develop an advanced, automated system for detecting brain hemorrhages using deep learning techniques. The system will utilize robust neural network models such as ResNet and MobileNet to analyze medical images and accurately distinguish between normal and hemorrhagic cases. A dataset comprising both normal and hemorrhage images will be used to train the model, enabling it to learn intricate patterns and subtle variations effectively. The proposed system leverages the superior feature extraction capabilities of deep learning to achieve high precision and reliability in diagnosis. By automating the detection process, the system can significantly reduce diagnostic time and provide consistent, unbiased results, thereby supporting healthcare professionals in making timely and informed decisions. This project has the potential to improve early diagnosis, facilitate prompt treatment, and ultimately enhance patient outcomes in critical care scenarios.
susceptible to human error practically high pressure environment such as emergency rooms. In rural the resource-constrained settings the ability of expired radiologists is often limited which can delay critical diagnosis.To address these challenges artificial intelligence and deep learning techniques has emerged on promising tools in medical images analysis. Deep learning and subfield of machine learning uses neural networks in mimic the human brain’s ability to recognize patterns on features in data. Among the most effective deep learning models are neural networks and mobile net which have shown exceptional performance image classification task. These models can automatically learn to identify hemorrhagic features from medical images offering consistent, fast, and highly accurate predictions. This project focuses on developing an automated system that utilises CNN mobile net architectures to detect brain hemorrhage from CT and MRI images. The system includes modules for image upload, preprocessing, model selection, prediction, and visual explanation through Grad-CAM heatmaps. CNN offers high precision and is suitable for clinical diagnosis with mobile net provides lightweight and real time inference ideal for mobile and emergency scenarios. By integrating deep learning with medical imaging system aims to reduce diagnosis time minimize human errors and increasing accessibility to quality healthcare especially in underserved areas. The overall goal is to create a reliable, interpretable, and scalable diagnostic tool that supports medical professionals in making faster and more accurate decisions for patients suffering from brain hemorrhage.
Key Words: Brain hemorrhage, deep learning, automated diagnosis, medical imaging, feature extraction, hemorrhage detection, healthcare innovation, neural networks
1.INTRODUCTION Brain hemorrhage is a critical and life threatening medical conditions causing by internal bleeding within the brain tissues or surrounding areas. These type of conditions can arise due to various reasons such as trauma, and high blood pressure, aneurysm rupture, or the excessive uses of blood thinking medications. If not diagnosis and treat promptly and brain hemorrhages can lead to irreversible brain damage, long term disability or death early detection and intervention are vital to improving several rates and preventing complications. Currently the primary method of diagnosis brain hemorrhage involves analyzing brain image scans such as computed tomography or magnetic resonance imaging. These scans are interpreted manually by radiologist who look for sign internal breeding, abnormalities, or swelling. While the process is generally accurate it is a time consuming requires specialized expertise, and the
© 2025, IRJET
|
Impact Factor value: 8.315
1.1 LITERATURE SURVEY Automated detection of brain hemorrhage has become a growing research area within the field of medical image analysis especially with the emergence of deep learning and computer visual techniques. Various researchers have explored mission learning and deep learning models analysis CT and MRI images to improve the accuracy and speed of diagnosis. 1. Chilamkurthy et al. (2018) proposed a deep learning model to detect critical finding in headless cities scans. Their system based on CNN achieved performance comparable school export radiologist especially in identifying intracranial hemorrhages.
|
ISO 9001:2008 Certified Journal
|
Page 572