Browsing by Author "Medina, Rubén"

Now showing 1 - 4 of 4

Accuracy of connected confidence left ventricle segmentation in 3-D multi-slice computerized tomography images
(Institute of Electrical and Electronics Engineers Inc., 2018) Medina, Rubén; Bautista Llivisaca, Mateo Sebastian; Morocho Zurita, Carlos Villie; Bautista Llivisaca, Mateo Sebastian
Cardiovascular diseases are the main cause of death in the World. This fact has motivated different actions for prevention, diagnosis and monitoring of cardiovascular diseases. In this work, the accuracy of a connected confidence left ventricle segmentation method is performed. This task is accomplished using a software platform for left ventricle segmentation of 3-D cardiac Multi-Slice Computerized Tomography (MSCT) images that is also described. The software platform has as a goal performing research about efficient methods for cardiac image segmentation and quantification. The accuracy assessment of the segmentation method is performed by comparing the estimated segmentation with respect to segmentations manually traced by cardiologists. Results show that the segmentation method provides Dice Similarity coefficients higher than 0.90 with low computational cost. The obtained segmentation is able to include within the left ventricular lumen the papillary trabeculae muscles, enabling further accurate estimation of the left ventricular mass.
Gearbox fault classification using dictionary sparse based representations of vibration signals
(2018) Medina, Rubén; Jadán Avilés, Diana Carolina; Álvarez Palomeque, Lourdes Ximena; Macancela Poveda, Jean Carlo; Sánchez Loja, René Vinicio; Cerrada Lozada, Mariela
Fault detection in rotating machinery is important for optimizing maintenance chores and avoiding severe damages to other parts. Signal processing based fault detection is usually performed by considering classical techniques for alternative representation of significant signals in time domain, frequency domain or time-frequency domain. An approach based on dictionary learning for sparse representations of vibration signals aiming at gearbox fault detection and classification is proposed. A gearbox signal dataset with 900 records considering the normal case and nine fault classes is analyzed. A dictionary is learned by using a training set of signals from the normal case. This dictionary is used for obtaining the sparse representation of signals in the test set and the norm metric is used to measure the residual from the sparse representation. The extracted features are useful for machine learning based fault detection. The analysis is performed considering different load conditions. ANOVA statistical analysis shows that there are significant differences between features in the normal case and each of the faulty classes, and best ranked features form well separated clusters. An experiment of fault classification is developed using a support vector machine for multi-class classification of faults. The accuracy obtained is 95.1% in the cross-validation testing.
Level-set segmentation of footprint images aimed at insole design
(Institute of Electrical and Electronics Engineers Inc., 2018) Medina, Rubén; Zeas Puga, Ana Lucía; Morocho, Villie; Bautista Llivisaca, Mateo Sebastian; Medina, Rubén
Chronic foot pain is a disease that progresses with age and has a high prevalence. Therapeutic procedures include the utilization of orthoses or insoles that are placed inside the footwear. Design of personalized insoles is a process that includes several stages. An important stage is the acquisition and analysis of footprint images. Their segmentation enables quantification of the footprint shape by estimating several indices that allow classification and diagnosis of foot morphology abnormalities. A segmentation method for footprint images using Level-Set algorithms is reported. Two area based Level-Set segmentation algorithms were applied. The first is the Chan-Vese algorithm using a global minimizer. The second is the Lankton algorithm that implements the Chan-Vese energy function using a localized minimizer and the Sparse Field Method for reducing the computational cost. Algorithms tested are accurate for segmenting the footprint images, providing an average Dice coefficient higher than 0.93. The Lankton algorithm is robust with respect to spatial variation in intensities within the footprint shape. It is also fast as the average time for segmenting one image is only 6.4 seconds
Validación de un algoritmo robusto para la estimación del movimiento en secuencias de imágenes cardiacas
(Universidad de Cuenca, 2015) Medina, Rubén; Ibarra, Emiro; Morocho, Villie; Vanegas, Pablo; Universidad de Cuenca; Dirección de Investigación de la Universidad de Cuenca; DIUC
This paper describes the validation of a sparse based optical flow algorithm using two sequences extracted from the Sintel database and also images extracted from a tagged magnetic resonance image sequence representing a short–axis slice located at the mid-wall of the left ventricle. Results are promising as the average magnitude error for the Sintel sequences is lower than 3 pixels and lower than 1 mm for the tagged MRI.