Perceptual Intelligence

Visual Search in Minimal Access Surgery and Surgical Skills Assessment 

The advent and accelerated adoption of laparoscopic surgery requires an objective assessment of both operative performance and perceptual events that lead to clinical decisions. In this work,  we present a framework to extract the underlying strategy through the analysis of saccadic eye-movements that lead to visual attention, and identification of intrinsic features central to the execution of basic laparoscopic tasks. 

- see paper presented at MICCAI 2004


ViTAL - Visual Tracking for Active Learning

The use of the spatio-temporal characteristics of human visual search captured by eye tracking, together with the intrinsic visual features of the fixation points, as the basis for domain knowledge representation. The use of eye tracking facilitates the systematic definition of diagnostic features and extraction of relevant knowledge from a visual domain. 

- see the ViTAL project web site



Gaze Contingent Robotic Control

The use of real-time binocular eye tracking for empowering the robots with human vision using knowledge acquired in situ, thus simplifying, as well as enhancing, robotic control in surgery. By utilizing the close relationship between the horizontal disparity and the depth perception, varying with the viewing distance, we demonstrate how vergence can be effectively used for recovering 3D depth at the fixation points and further be used for adaptive motion stabilization during surgery. 

- see paper presented at MIAR 2004

Saliency Driven 2D/3D Registration

The  use of eye tracking for extracting visual saliency, and its application for 2D to 3D image registration of endoscopy videos with computed tomography data. It is shown that by only processing selective regions of interest as determined by the saliency map, rendering overhead can be greatly reduced. Significant improvements in pose estimation efficiency can be achieved without apparent reduction in registration accuracy when compared to that of using a non-saliency based similarity measure. 


- see paper presented at ETRA 2004


Detection of Visual Saliency in Visual Search & Transient Fixation Moments

The development of a new method for identifying salient visual features based on scan path analysis. The key difference of the proposed method compared to existing approaches is that it is based on the modeling of fixations in feature space, and therefore identifies the intrinsic link between overt eye movements and visual attention. Similar to the two-stage visual search theory, the proposed method first projects scan path into the feature space, which is then followed by identifying “hot spots” where visual attention has been focussed upon. The results are subsequently projected back into the spatial space, allowing the identification of salient regions of interest.

In the News