The use of machine vision to inspect the outcome of surgical tasks is investigated, with the aim of incorporating this approach in robotic surgery systems. Machine vision is a non-contact form of inspection i.e. no part of the vision system is in direct contact with the patient, and is therefore well suited for surgery where sterility is an important consideration,. As a proof-of-concept, three primary surgical tasks for a common neurosurgical procedure were inspected using machine vision. Experiments were performed on cadaveric pig heads to simulate the two possible outcomes i.e. satisfactory or unsatisfactory, for tasks involved in making a burr hole, namely incision, retraction, and drilling. We identify low level image features to distinguish the two outcomes, as well as report on results that validate our proposed approach. The potential of using machine vision in a surgical environment, and the challenges that must be addressed, are identified and discussed.

In this paper, algorithms for the automatic localisation of two anatomical soft tissue landmarks of the head the medial canthus (inner corner of the eye) and the tragus (a small, pointed, cartilaginous flap of the ear), in CT images are describet. These landmarks are to be used as a basis for an automated image-to-patient registration system we are developing. The landmarks are localised on a surface model extracted from CT images, based on surface curvature and a rule based system that incorporates prior knowledge of the landmark characteristics. The approach was tested on a dataset of near isotropic CT images of 95 patients. The position of the automatically localised landmarks was compared to the position of the manually localised landmarks. The average difference was 1.5 mm and 0.8 mm for the medial canthus and tragus, with a maximum difference of 4.5 mm and 2.6 mm respectively.The medial canthus and tragus can be automatically localised in CT images, with performance comparable to manual localisation

This paper presents the automated methods employed for extracting craniofacial landmarks in white light images as part of a registration framework designed to support three neurosurgical procedures. The intraoperative space is characterised by white light stereo imaging while the preoperative plan is performed on CT scans. The registration aims at aligning these two modalities to provide a calibrated environment to enable image-guided solutions. The neurosurgical procedures can then be carried out by mapping the entry and target points from CT space onto the patient-s space. The registration basis adopted consists of natural landmarks (eye corner and ear tragus). A 5mm accuracy is deemed sufficient for these three procedures and the validity of the selected registration basis in achieving this accuracy has been assessed by simulation studies. The registration protocol is briefly described, followed by a presentation of the automated techniques developed for the extraction of the craniofacial features and results obtained from tests on the AR and FERET databases. Since the three targeted neurosurgical procedures are routinely used for head injury management, the effect of bruised/swollen faces on the automated algorithms is assessed. A user-interactive method is proposed to deal with such unpredictable circumstances.

A registration framework for image-guided robotic surgery is proposed for three emergency neurosurgical procedures, namely Intracranial Pressure (ICP) Monitoring, External Ventricular Drainage (EVD) and evacuation of a Chronic Subdural Haematoma (CSDH). The registration paradigm uses CT and white light as modalities. This paper presents two simulation studies for a preliminary evaluation of the registration protocol: (1) The loci of the Target Registration Error (TRE) in the patient-s axial, coronal and sagittal views were simulated based on a Fiducial Localisation Error (FLE) of 5 mm and (2) Simulation of the actual framework using projected views from a surface rendered CT model to represent white light images of the patient. Craniofacial features were employed as the registration basis to map the CT space onto the simulated intraoperative space. Photogrammetry experiments on an artificial skull were also performed to benchmark the results obtained from the second simulation. The results of both simulations show that the proposed protocol can provide a 5mm accuracy for these neurosurgical procedures.

In this paper, the feasibility of using machine vision to assess task completion in a surgical intervention is investigated, with the aim of incorporating vision based inspection in robotic surgery systems. The visually rich operative field presents a good environment for the development of automated visual inspection techniques in these systems, for a more comprehensive approach when performing a surgical task. As a proof of concept, machine vision techniques were used to distinguish the two possible outcomes i.e. satisfactory or unsatisfactory, of three primary surgical tasks involved in creating a burr hole in the skull, namely incision, retraction, and drilling. Encouraging results were obtained for the three tasks under consideration, which has been demonstrated by experiments on cadaveric pig heads. These findings are suggestive for the potential use of machine vision to validate successful task completion in robotic surgery systems. Finally, the potential of using machine vision in the operating theatre, and the challenges that must be addressed, are identified and discussed.

The use of anatomical landmarks as a basis for image to patient registration is appealing because the registration may be performed retrospectively. We have previously proposed the use of two anatomical soft tissue landmarks of the head, the canthus (corner of the eye) and the tragus (a small, pointed, cartilaginous flap of the ear), as a registration basis for an automated CT image to patient registration system, and described their localization in patient space using close range photogrammetry. In this paper, the automatic localization of these landmarks in CT images, based on their curvature saliency and using a rule based system that incorporates prior knowledge of their characteristics, is described. Existing approaches to landmark localization in CT images are predominantly semi-automatic and primarily for localizing internal landmarks. To validate our approach, the positions of the landmarks localized automatically and manually in near isotropic CT images of 102 patients were compared. The average difference was 1.2mm (std = 0.9mm, max = 4.5mm) for the medial canthus and 0.8mm (std = 0.6mm, max = 2.6mm) for the tragus. The medial canthus and tragus can be automatically localized in CT images, with performance comparable to manual localization, based on the approach presented.