Skip to main content

Resultados relacionados

Resultados totales (Incluyendo duplicados): 3
Encontrada(s) 1 página(s)

RUC. Repositorio da Universidade da Coruña
Publication Contribución de congreso (ConferenceOutput)

Cálculo de flujo óptico denso en colonoscopia mediante aprendizaje no supervisado, Dense optical flow estimation in colonoscopy images using an unsupervised learning approach

  • Gonzalo, Iván
  • Morlana, Javier
  • Martínez Montiel, José María
[Resumen] La colonoscopia es la técnica de referencia en la detección del cáncer colorrectal. Sin embargo, los métodos asistidos por ordenador no se usan mucho en estos procedimientos. Este trabajo se enmarca dentro del proyecto EndoMapper, que tiene como objetivo crear reconstrucciones del colon que puedan ser utilizadas para ayudar a los médicos o para llevar a cabo cirugía robotizada. La asociación de datos es un elemento clave en estos sistemas, realizando la asociación entre píxeles de la imagen para permitir posteriores reconstrucciones 3D. En este trabajo, evaluamos un método de estimación de flujo óptico denso sobre imágenes de colonoscopia, adaptándolo al dominio del colon mediante aprendizaje no supervisado. Diseñamos un conjunto de datos para training, validation y test a partir de las secuencias reales de colonoscopia del conjunto de datos de Endomapper. El modelo se ha re-entrenado obteniendo una versión adaptada al dominio del colon. La validación experimental muestra cómo el modelo entrenado puede estimar el flujo de manera robusta bajo cambios de iluminación. También muestra una capacidad excepcional para estimar el flujo entre imágenes de colonoscopia muy separadas con grandes rotaciones., [Abstract] Colonoscopy is the gold standard in colorectal cancer screening. However, computer-assisted interventional methods are not widely used in these procedures. The Endomapper project, in which this work is embedded, aims to create reconstructions of the colon that can be used to assist doctors or for robotic surgery. Data association is a key element in these systems, performing the association between image pixels to enable subsequent 3D reconstructions. In this work, we evaluated a dense optical flow method on colonoscopy images, adapting it to the colon domain by using unsupervised learning. We built a dataset for training, validation and test from the real colonoscopy sequences of the Endomapper dataset. The model has been re-trained obtaining a version adapted to the colon domain. Experimental validation shows how the trained model is able to estimate flow robustly under illumination changes. It also shows an exceptional ability to estimate flow between widely separated colonoscopy images with large rotations., Ministerio de Educación y Formación Profesional; 22CO1/006926, Ministerio de Ciencia e Innovación; PID2021-127685NB-I00, Gobierno de Aragón; DGA_T45-17R
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publication Contribución de congreso (ConferenceOutput)

Photometric single-view dense 3D reconstruction in endoscopy

  • Martínez Batlle, V.
  • Martínez Montiel, J. M.
  • Tardos, J. D.
Visual SLAM inside the human body will open the way to computer-assisted navigation in endoscopy. However, due to space limitations, medical endoscopes only provide monocular images, leading to systems lacking true scale. In this paper, we exploit the controlled lighting in colonoscopy to achieve the first in-vivo 3D reconstruction of the human colon using photometric stereo on a calibrated monocular endoscope. Our method works in a real medical environment, providing both a suitable in-place calibration procedure and a depth estimation technique adapted to the colon's tubular geometry. We validate our method on simulated colonoscopies, obtaining a mean error of 7% on depth estimation, which is below 3 mm on average. Our qualitative results on the EndoMapper dataset show that the method is able to correctly estimate the colon shape in real human colonoscopies, paving the ground for truescale monocular SLAM in endoscopy.
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publication Contribución de congreso (ConferenceOutput)

LightDepth: Single-view depth self-supervision from illumination decline

  • Rodríguez-Puigvert, Javier
  • Batlle, Víctor M.
  • Martínez Montiel, J.M.
  • Martinez-Cantin, Ruben
  • Fua, Pascal
  • Tardós, Juan D.
  • Civera, Javier
Single-view depth estimation can be remarkably effective if there is enough ground-truth depth data for supervised training. However, there are scenarios, especially in medicine in the case of endoscopies, where such data cannot be obtained. In such cases, multi-view self-supervision and synthetic-to-real transfer serve as alternative approaches, however, with a considerable performance reduction in comparison to supervised case. Instead, we propose a single-view self-supervised method that achieves a performance similar to the supervised case. In some medical devices, such as endoscopes, the camera and light sources are co-located at a small distance from the target surfaces. Thus, we can exploit that, for any given albedo and surface orientation, pixel brightness is inversely proportional to the square of the distance to the surface, providing a strong single-view self-supervisory signal. In our experiments, our self-supervised models deliver accuracies comparable to those of fully supervised ones, while being applicable without depth ground-truth data.