Publications by authors named "Tiberiu Marita"
The aim of this study was to evaluate the feasibility of a noninvasive and low-operator-dependent imaging method for carotid-artery-stenosis diagnosis. A previously developed prototype for 3D ultrasound scans based on a standard ultrasound machine and a pose reading sensor was used for this study. Working in a 3D space and processing data using automatic segmentation lowers operator dependency.
View Article and Find Full Text PDFArticle Synopsis
- The research evaluated Mask R-CNN and U-Net models for automatically segmenting ultrasound images of dental arches to assist in periodontal diagnosis.
- It involved a 3D ultrasound investigation on 52 teeth, where initial image segmentation was performed by a less experienced operator and subsequently corrected using 3D reconstructions.
- Results showed that while the original dataset contained 3417 images, the corrected dataset (2135 images) yielded better prediction accuracy, demonstrating the effectiveness of the quality correction method.
The aim of this study was to develop and evaluate a 3D ultrasound scanning method. The main requirements were the freehand architecture of the scanner and high accuracy of the reconstructions. A quantitative evaluation of a freehand 3D ultrasound scanner prototype was performed, comparing the ultrasonographic reconstructions with the CAD (computer-aided design) model of the scanned object, to determine the accuracy of the result.
View Article and Find Full Text PDFHepatocellular carcinoma (HCC) is one of the leading causes of cancer-related deaths worldwide, with its mortality rate correlated with the tumor staging; i.e., early detection and treatment are important factors for the survival rate of patients.
View Article and Find Full Text PDFArticle Synopsis
- Deep-learning methods have shown better performance in computer vision tasks when large datasets are available, but there's a question regarding their effectiveness in medical imaging, particularly with limited data.
- The study proposes a lightweight multi-resolution Convolutional Neural Network (CNN) for classifying ultrasound images between Hepatocellular Carcinoma (HCC) and cirrhotic parenchyma (PAR).
- Results indicate that the deep-learning model outperforms conventional machine-learning methods, achieving higher classification accuracy for the ultrasound binary classification task.