Cookies
Usamos cookies para melhorar nosso site e a sua experiência. Ao continuar a navegar no site, você aceita a nossa política de cookies. Ver mais
Fechar
  • Menu
Publicações

Publicações por Teresa Finisterra Araújo

2015

Optical Flow Based Approach for Automatic Cardiac Cycle Estimation in Ultrasound Images of the Carotid

Autores
Araujo, T; Aresta, G; Rouco, J; Ferreira, C; Azevedo, E; Campilho, A;

Publicação
IMAGE ANALYSIS AND RECOGNITION (ICIAR 2015)

Abstract
This paper proposes a method to detect a reference frame in an ultrasound video of the carotid artery. This reference frame, usually located at the end of the diastole, is used as the location to measure several vascular biomarkers. Our approach is based on the analysis of the movement of the carotid walls in ultrasound images using an optical flow technique. A periodic movement resembling heart beat is observed in the resulting signals. The comparison of these signals with electrocardiograms validates the proposed method for detecting the reference frame.

2017

Classification of breast cancer histology images using Convolutional Neural Networks

Autores
Araujo, T; Aresta, G; Castro, E; Rouco, J; Aguiar, P; Eloy, C; Polonia, A; Campilho, A;

Publicação
PLOS ONE

Abstract

2017

Improving convolutional neural network design via variable neighborhood search

Autores
Araujo, T; Aresta, G; Almada Lobo, B; Mendonca, AM; Campilho, A;

Publicação
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Abstract
An unsupervised method for convolutional neural network (CNN) architecture design is proposed. The method relies on a variable neighborhood search-based approach for finding CNN architectures and hyperparameter values that improve classification performance. For this purpose, t-Distributed Stochastic Neighbor Embedding (t-SNE) is applied to effectively represent the solution space in 2D. Then, k-Means clustering divides this representation space having in account the relative distance between neighbors. The algorithm is tested in the CIFAR-10 image dataset. The obtained solution improves the CNN validation loss by over 15% and the respective accuracy by 5%. Moreover, the network shows higher predictive power and robustness, validating our method for the optimization of CNN design. © Springer International Publishing AG 2017.

2018

Retinal image quality assessment by mean-subtracted contrast-normalized coefficients

Autores
Galdran, A; Araujo, T; Mendonca, AM; Campilho, A;

Publicação
Lecture Notes in Computational Vision and Biomechanics

Abstract
The automatic assessment of visual quality on images of the eye fundus is an important task in retinal image analysis. A novel quality assessment technique is proposed in this paper. We propose to compute Mean-Subtracted Contrast-Normalized (MSCN) coefficients on local spatial neighborhoods of a given image and analyze their distribution. It is known that for natural images, such distribution behaves normally, while distortions of different kinds perturb this regularity. The combination of MSCN coefficients with a simple measure of local contrast allows us to design a simple but effective retinal image quality assessment algorithm that successfully discriminates between good and low-quality images, while delivering a meaningful quality score. The proposed technique is validated on a recent database of quality-labeled retinal images, obtaining results aligned with state-of-the-art approaches at a low computational cost. © 2018, Springer International Publishing AG.

2017

Estimation of retinal vessel caliber using model fitting and random forests

Autores
Araujo, T; Mendonca, AM; Campilho, A;

Publicação
Medical Imaging 2017: Computer-Aided Diagnosis, Orlando, Florida, United States, 11-16 February 2017

Abstract

2017

Segmentation and three-dimensional reconstruction of lesions using the automated breast volume scanner (ABVS)

Autores
Araujo, T; Abayazid, M; Rutten, MJCM; Misra, S;

Publicação
INTERNATIONAL JOURNAL OF MEDICAL ROBOTICS AND COMPUTER ASSISTED SURGERY

Abstract
BackgroundUltrasound is an effective tool for breast cancer diagnosis. However, its relatively low image quality makes small lesion analysis challenging. This promotes the development of tools to help clinicians in the diagnosis. MethodsWe propose a method for segmentation and three-dimensional (3D) reconstruction of lesions from ultrasound images acquired using the automated breast volume scanner (ABVS). Segmentation and reconstruction algorithms are applied to obtain the lesion's 3D geometry. A total of 140 artificial lesions with different sizes and shapes are reconstructed in gelatin-based phantoms and biological tissue. Dice similarity coefficient (DSC) is used to evaluate the reconstructed shapes. The algorithm is tested using a human breast phantom and clinical data from six patients. ResultsDSC values are 0.860.06 and 0.86 +/- 0.05 for gelatin-based phantoms and biological tissue, respectively. The results are validated by a specialized clinician. ConclusionsEvaluation metrics show that the algorithm accurately segments and reconstructs various lesions. Copyright (c) 2016 John Wiley & Sons, Ltd.

  • 1
  • 3