Automated Segmentation of Head CT Scans for Computerassisted Craniomaxillofacial Surgery Applying a Hierarchical Patchbased Stack of Convolutional Neural NetworksInternational Journal of Computer Assisted Radiology
CARDIOWATCH: EARLY DETECTION USING WEARABLE ULTRASOUND
Abstract for the project work:
The project focuses on the development of a novel wearable ultrasound device for the early detection of cardiovascular diseases (CVD) through routine monitoring of the carotid arteries. Utilizing B-mode ultrasound technology, the device aims to monitor intima-media thickness (IMT), a proven indicator of CVD, in an easy-to-use and cost-effective manner. The proposed system includes hardware components integrated into a neck brace-like structure, with ultrasonic transducer array probes for image acquisition and an image processing algorithm for analysing ultrasound images to detect plaque build-up in the carotid artery. By quantifying biomarkers such as lumen diameter, the algorithm can predict atherosclerosis and CVD, enabling early detection and intervention. The project's innovative approach offers a portable, non- invasive, and economical solution for point-of-care monitoring and autonomous anomaly detection, potentially revolutionizing the early prediction, diagnosis, and monitoring of CVDs and related conditions. Literature survey: -----------------------SURVEY 1--------------------------------------------------------------- V. Nambi et al., “Carotid intima-media thickness and presence or absence of plaque improves prediction of coronary heart disease risk”, evaluated the impact of adding carotid intima-media thickness (CIMT) and plaque information to traditional risk factors in predicting coronary heart disease (CHD) risk. The results showed that incorporating CIMT and plaque data improved CHD risk prediction, especially in the intermediate risk groups. The addition of plaque information had a more significant effect on improving risk prediction in women compared to men. Overall, the study suggests that CIMT and plaque assessment can enhance CHD risk prediction and should be considered in clinical practice for individuals with intermediate risk. -----------------------SURVEY 2--------------------------------------------------------------- Tomohisa Nezu et al., “Carotid Intima-Media Thickness for Atherosclerosis”, in this article authors discussed about the assessment of CVDs form the CIMT. This article discusses the use of carotid intima-media thickness (IMT) to assess cardiovascular disease (CVD) risk. Previously, increased IMT was considered a strong predictor of future heart attacks and strokes. It was easy and non-invasive to measure, making it widely used. Studies even showed that lowering IMT through medication seemed to benefit patients. While increased IMT is still linked to CVD, it may not be the best predictor of future events. Decreasing IMT with medication didn't necessarily translate to fewer heart attacks. Additionally, other factors like the presence and characteristics of plaque buildup in the carotid artery might be more important. Current guidelines don't recommend routine IMT measurement for initial CVD risk assessment. Carotid IMT and the plaque formation adds more weight to the presence of CVDs rather than accounting the IMT alone. ---------------------------------------SURVEY 3-------------------------------------------------- Apostolopoulos, T., et al., "Ultrasound-Based Image Analysis for Predicting Carotid Artery Stenosis Risk: Problem, Techniques, Datasets, and Future Directions." in this article Extracting the region of interest (ROI) from US images is crucial. Techniques include active contours, level sets, and thresholding methods. In Feature Extraction: Geometric features like lumen and plaque area, intima-media thickness (IMT), and wall thickness-to- lumen ratio (WT/LR) are extracted to quantify arterial morphology. Additionally, texture features can capture plaque composition. Supervised learning algorithms like Support Vector Machines (SVM) and Random Forests can be trained on labeled data to classify images as normal or stenotic. Deep learning models, particularly convolutional neural networks (CNNs), are increasingly used for automated feature extraction and classification due to their ability to learn complex relationships from large datasets.
WORKFLOW:
WORK PLAN (role of each member):
1. Hardware Development and Integration: Person A: Responsible for designing and implementing the hardware components of the wearable imaging system, including the transducer, transmit/receive circuitry, and power module. Person B: Focuses on integrating the hardware components into a functional prototype, ensuring proper assembly and functionality of the device. 2. Algorithm Development and Image Processing: Person C: Leads the development of the efficient algorithm for predicting IMT anomalies from ultrasound images, including image analysis and plaque detection algorithms. Person A and Person B: Collaborate with Person C to provide input on hardware requirements for efficient algorithm execution and integration of the algorithm into the device. 3. Testing and Validation: All team members: Collaborate on testing the wearable ultrasound device prototype, conducting validation studies to ensure accurate detection of IMT anomalies and plaque build-up. Person B: Responsible for coordinating user testing and feedback collection to iterate on the design and functionality of the device.
Automated Segmentation of Head CT Scans for Computerassisted Craniomaxillofacial Surgery Applying a Hierarchical Patchbased Stack of Convolutional Neural NetworksInternational Journal of Computer Assisted Radiology