Publications by authors named "Cosmas Mwikirize"
Time-aware deep neural networks for needle tip localization in 2D ultrasound.
Int J Comput Assist Radiol Surg
May 2021
Article Synopsis
- Accurate needle placement is crucial for procedures like biopsies and regional anesthesia, where ultrasound guidance can help, but deep insertions can obscure visibility.
- A new algorithm enhances needle tip visibility in ultrasound frames by detecting subtle intensity changes, using a hybrid deep neural network to predict needle tip location even when the shaft is not visible.
- Tests with various tissues and different needle sizes showed a significant 30% improvement in tip localization accuracy and a fast response time, indicating the method's potential for clinical use.
Learning needle tip localization from digital subtraction in 2D ultrasound.
Int J Comput Assist Radiol Surg
June 2019
Article Synopsis
- This paper introduces a method for accurately localizing needle tips during ultrasound-guided interventions, addressing challenges presented by low intensity of needle visibility.
- The approach merges a digital subtraction technique to enhance subtle intensity changes from tip movement with a deep learning model to achieve reliable needle tip detection.
- Testing on an extensive dataset of needle images showed that the method achieved a minimal localization error (0.72 mm) while processing images at a rate of approximately 10 frames per second, outperforming existing techniques.
Convolution neural networks for real-time needle detection and localization in 2D ultrasound.
Int J Comput Assist Radiol Surg
May 2018
Purpose: We propose a framework for automatic and accurate detection of steeply inserted needles in 2D ultrasound data using convolution neural networks. We demonstrate its application in needle trajectory estimation and tip localization.
Methods: Our approach consists of a unified network, comprising a fully convolutional network (FCN) and a fast region-based convolutional neural network (R-CNN).
Signal attenuation maps for needle enhancement and localization in 2D ultrasound.
Int J Comput Assist Radiol Surg
March 2018
Purpose: We propose a novel framework for enhancement and localization of steeply inserted hand-held needles under in-plane 2D ultrasound guidance.
Methods: Depth-dependent attenuation and non-axial specular reflection hinder visibility of steeply inserted needles. Here, we model signal transmission maps representative of the attenuation probability within the image domain.