BIOMEDICAL ENGINEERING & BIOMEDICAL PROJECTS
1. A DEEP NEURAL NETWORK BASED DIABETIC RETINOPATHY CLASSIFICATION
As a result of diabetes mellitus, which causes lesions on the retina of the eyes, diabetic retinopathy is a degenerative condition that affects the eyes. For diabetic patients, especially the working-age population in poor countries, diabetic retinopathy is thought to be the main cause of blindness. Since the condition is irreversible, treatment focuses on maintaining the patient's level of eyesight. Regular screening is crucial for diabetic patients to ensure that DR is detected at an early stage. DR detection traditionally involves a physician’s examination of retinal imaging for the shape and appearance of different types of lesions. Traditional screening of retinal diseases requires multiple stages of scans followed by filtration techniques to narrow down the subject samples. Optical coherence tomography (OCT) and spatial domain optical coherence tomography (SD-OCT) are examples of scans performed during the screening stage Machine learning-based medical image analysis has proven competency in assessing retinal fundus images, and the utilization of deep learning algorithms has aided the early diagnosis of Diabetic Retinopathy. The paper discusses the available retinal fundus datasets for Diabetic Retinopathy that are used for tasks such as detection and classification. The proposed system is a Deep Learning-based system for classifying DR aneurysms (MA), haemorrhages (HM), and exudates, using Convolutional Neural Network (CNN) algorithm. The system consists of three stages: image pre-processing, feature extraction, and classification. In the first stage, the retinal fundus images are pre-processed to enhance their quality and remove noise. In the second stage, features are extracted from the pre-processed images using CNN algorithm. The system achieves high accuracy in classifying DR aneurysms (MA), haemorrhages (HM), and exudates, and can potentially be used for large-scale screening programs to improve the early detection and treatment of DR. In addition to DR, this system consists of a gas sensor that detects the level of acetone in the patient's breath.
.
2. Digital Breast Tomosynthesis Deep Learning Advancements In Cancer Detection
.
Breast cancer is a prevalent and potentially life-threatening disease, necessitating early detection for effective treatment. In this context, leveraging advanced technologies such as machine learning offers promising avenues for improving diagnostic accuracy. This project focuses on the development and implementation of a Convolutional Neural Network (CNN) algorithm to enhance the detection of breast cancer cells from histopathological images. The proposed CNN model is designed to automatically learn hierarchical features from microscopic images, enabling it to discern subtle patterns indicative of cancerous cells. The dataset used for training and evaluation encompasses a diverse range of breast tissue samples, ensuring the model's robustness across various pathological presentations. The project aims to contribute to the existing body of knowledge by providing a reliable and efficient tool for aiding pathologists in the early detection of breast cancer. The performance of the CNN algorithm is assessed through rigorous experimentation and comparison with existing methods, demonstrating its potential to significantly improve the accuracy and speed of breast cancer diagnosis. This research not only addresses a critical healthcare challenge but also underscores the transformative impact of machine learning in advancing medical diagnostics.
3. DETECTION OF PANCREATIC CANCER USING MACHINE LEARNING CNN
Pancreatic cancer is one of the most lethal forms of cancer, with a high mortality rate primarily due to late-stage diagnosis. Early detection plays a crucial role in improving patient outcomes by enabling timely intervention and treatment. In this study, we propose a novel approach for the detection of pancreatic cancer using Convolutional Neural Networks (CNNs). Leveraging the capabilities of deep learning, our CNN-based model analyzes medical imaging data, specifically computed tomography (CT) scans, to accurately identify signs of pancreatic malignancy.The proposed CNN architecture is trained on a large dataset of annotated CT scans obtained from patients with confirmed pancreatic cancer and healthy controls. Through extensive training and validation, the model learns to differentiate between normal pancreatic tissue and cancerous lesions, capturing subtle structural and textural features indicative of malignancy. Additionally, the CNN is optimized to minimize false positives and negatives, ensuring high sensitivity and specificity in cancer detection.To evaluate the performance of our approach, we conduct rigorous testing using independent datasets and benchmark against existing diagnostic methods. Our results demonstrate the efficacy of the CNN-based model in detecting pancreatic cancer with high accuracy and reliability. Moreover, we explore the potential of integrating our automated detection system into clinical workflows to aid radiologists and oncologists in early cancer diagnosis.Overall, our study presents a promising advancement in the field of pancreatic cancer detection, offering a non-invasive and efficient approach for early diagnosis. By harnessing the power of deep learning and medical imaging, our CNN-based method has the potential to revolutionize pancreatic cancer screening and improve patient outcomes through timely intervention and treatment.
4.AI ENHANCED EYE TRACKING USING DEEP LEARNING IN CONVOLUTIONAL NEURAL NETWORK
The Intelligent Eyeball Controlled Wheelchair for Paralyzed Patients using Arduino is a novel assistive technology designed to improve the mobility of people who are paralyzed or have limited motor function. The system is based on a combination of an Arduino microcontroller and a camera that tracks eye movements. By monitoring the movement of the user's eyes, the system is able to control the movement of the wheelchair in real time. The system consists of a camera mounted on a pair of glasses worn by the user. The camera captures images of the user's eyes, which are then processed by an Arduino microcontroller. The microcontroller processes the eye movement data and uses it to control the movement of the wheelchair. The wheelchair is equipped with motors and wheels, which are controlled by the microcontroller. The microcontroller is programmed to interpret the user's eye movements and translate them into commands for the wheelchair's motors. The system can be programmed to recognize a range of eye movements, such as up, down, left, right, and blink, which can be used to control the wheelchair's movement. The Intelligent Eyeball Controlled Wheelchair for Paralyzed Patients using Arduino has the potential to significantly improve the mobility and quality of life of people who are paralyzed or have limited motor function. The system is affordable, easy to use, and can be customized to meet the needs of individual users. It has the potential to revolutionize the field of assistive technology and provide a new level of independence to people with disabilities.
5. Intelligent System For Alzheimer’s Risk Assessment.
The aim of this project is to develop a deep learning model based on Convolutional Neural Networks (CNNs) to accurately diagnose Alzheimer's disease (AD) using Magnetic Resonance Imaging (MRI) scans. AD is a progressive and irreversible brain disorder that affects memory and cognitive function, and its diagnosis is challenging due to the complexity and subtlety of its symptoms. In this project, we will start by collecting a large dataset of MRI scans from patients with AD, as well as healthy individuals, to train and validate the CNN model. The data will be pre-processed and augmented to improve the performance of the model. Then, we will design a CNN architecture that is capable of learning discriminative features from the MRI scans and classifying them into AD or non-AD. We will evaluate the performance of the model using various metrics such as accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC-ROC). We will also compare our model's performance with other state-of-the-art methods for AD diagnosis.