Identifying key mechanisms leading to visual recognition errors for missed colorectal polyps using eye-tracking technology.

BACKGROUND AND AIM: Lack of visual recognition of colorectal polyps may lead to interval cancers. The mechanisms contributing to perceptual variation, particularly for subtle and advanced colorectal neoplasia, have scarcely been investigated. We aimed to evaluate visual recognition errors and provide novel mechanistic insights. METHODS: Eleven participants (seven trainees and four medical students) evaluated images from the UCL polyp perception dataset, containing 25 polyps, using eye-tracking equipment. Gaze errors were defined as those where the lesion was not observed according to eye-tracking technology. Cognitive errors occurred when lesions were observed but not recognized as polyps by participants. A video study was also performed including 39 subtle polyps, where polyp recognition performance was compared with a convolutional neural network. RESULTS: Cognitive errors occurred more frequently than gaze errors overall (65.6%), with a significantly higher proportion in trainees (P = 0.0264). In the video validation, the convolutional neural network detected significantly more polyps than trainees and medical students, with per-polyp sensitivities of 79.5%, 30.0%, and 15.4%, respectively. CONCLUSIONS: Cognitive errors were the most common reason for visual recognition errors. The impact of interventions such as artificial intelligence, particularly on different types of perceptual errors, needs further investigation including potential effects on learning curves. To facilitate future research, a publicly accessible visual perception colonoscopy polyp database was created.

Artificial intelligence and automation in endoscopy and surgery.

Modern endoscopy relies on digital technology, from high-resolution imaging sensors and displays to electronics connecting configurable illumination and actuation systems for robotic articulation. In addition to enabling more effective diagnostic and therapeutic interventions, the digitization of the procedural toolset enables video data capture of the internal human anatomy at unprecedented levels. Interventional video data encapsulate functional and structural information about a patient's anatomy as well as events, activity and action logs about the surgical process. This detailed but difficult-to-interpret record from endoscopic procedures can be linked to preoperative and postoperative records or patient imaging information. Rapid advances in artificial intelligence, especially in supervised deep learning, can utilize data from endoscopic procedures to develop systems for assisting procedures leading to computer-assisted interventions that can enable better navigation during procedures, automation of image interpretation and robotically assisted tool manipulation. In this Perspective, we summarize state-of-the-art artificial intelligence for computer-assisted interventions in gastroenterology and surgery.

Refractive Two-View Reconstruction for Underwater 3D Vision.

Recovering 3D geometry from cameras in underwater applications involves the Refractive Structure-from-Motion problem where the non-linear distortion of light induced by a change of medium density invalidates the single viewpoint assumption. The pinhole-plus-distortion camera projection model suffers from a systematic geometric bias since refractive distortion depends on object distance. This leads to inaccurate camera pose and 3D shape estimation. To account for refraction, it is possible to use the axial camera model or to explicitly consider one or multiple parallel refractive interfaces whose orientations and positions with respect to the camera can be calibrated. Although it has been demonstrated that the refractive camera model is well-suited for underwater imaging, Refractive Structure-from-Motion remains particularly difficult to use in practice when considering the seldom studied case of a camera with a flat refractive interface. Our method applies to the case of underwater imaging systems whose entrance lens is in direct contact with the external medium. By adopting the refractive camera model, we provide a succinct derivation and expression for the refractive fundamental matrix and use this as the basis for a novel two-view reconstruction method for underwater imaging. For validation we use synthetic data to show the numerical properties of our method and we provide results on real data to demonstrate its practical application within laboratory settings and for medical applications in fluid-immersed endoscopy. We demonstrate our approach outperforms classic two-view Structure-from-Motion method relying on the pinhole-plus-distortion camera model.

Computer Vision in the Surgical Operating Room.

BACKGROUND: Multiple types of surgical cameras are used in modern surgical practice and provide a rich visual signal that is used by surgeons to visualize the clinical site and make clinical decisions. This signal can also be used by artificial intelligence (AI) methods to provide support in identifying instruments, structures, or activities both in real-time during procedures and postoperatively for analytics and understanding of surgical processes. SUMMARY: In this paper, we provide a succinct perspective on the use of AI and especially computer vision to power solutions for the surgical operating room (OR). The synergy between data availability and technical advances in computational power and AI methodology has led to rapid developments in the field and promising advances. KEY MESSAGES: With the increasing availability of surgical video sources and the convergence of technologies around video storage, processing, and understanding, we believe clinical solutions and products leveraging vision are going to become an important component of modern surgical capabilities. However, both technical and clinical challenges remain to be overcome to efficiently make use of vision-based approaches into the clinic.

Adjoint Transformation Algorithm for Hand-Eye Calibration with Applications in Robotic Assisted Surgery.

Hand-eye calibration aims at determining the unknown rigid transformation between the coordinate systems of a robot arm and a camera. Existing hand-eye algorithms using closed-form solutions followed by iterative non-linear refinement provide accurate calibration results within a broad range of robotic applications. However, in the context of surgical robotics hand-eye calibration is still a challenging problem due to the required accuracy within the millimetre range, coupled with a large displacement between endoscopic cameras and the robot end-effector. This paper presents a new method for hand-eye calibration based on the adjoint transformation of twist motions that solves the problem iteratively through alternating estimations of rotation and translation. We show that this approach converges to a solution with a higher accuracy than closed form initializations within a broad range of synthetic and real experiments. We also propose a stereo hand-eye formulation that can be used in the context of both our proposed method and previous state-of-the-art closed form solutions. Experiments with real data are conducted with a stereo laparoscope, the KUKA robot arm manipulator, and the da Vinci surgical robot, showing that both our new alternating solution and the explicit representation of stereo camera hand-eye relations contribute to a higher calibration accuracy.

A Continuum Robot and Control Interface for Surgical Assist in Fetoscopic Interventions.

Twin-twin transfusion syndrome requires interventional treatment using a fetoscopically introduced laser to sever the shared blood supply between the fetuses. This is a delicate procedure relying on small instrumentation with limited articulation to guide the laser tip and a narrow field of view to visualize all relevant vascular connections. In this letter, we report on a mechatronic design for a comanipulated instrument that combines concentric tube actuation to a larger manipulator constrained by a remote centre of motion. A stereoscopic camera is mounted at the distal tip and used for imaging. Our mechanism provides enhanced dexterity and stability of the imaging device. We demonstrate that the imaging system can be used for computing geometry and enhancing the view at the operating site. Results using electromagnetic sensors for verification and comparison to visual odometry from the distal sensor show that our system is promising and can be developed further for multiple clinical needs in fetoscopic procedures.

Medical-grade Sterilizable Target for Fluid-immersed Fetoscope Optical Distortion Calibration.

We have developed a calibration target for use with fluid-immersed endoscopes within the context of the GIFT-Surg (Guided Instrumentation for Fetal Therapy and Surgery) project. One of the aims of this project is to engineer novel, real-time image processing methods for intra-operative use in the treatment of congenital birth defects, such as spina bifida and the twin-to-twin transfusion syndrome. The developed target allows for the sterility-preserving optical distortion calibration of endoscopes within a few minutes. Good optical distortion calibration and compensation are important for mitigating undesirable effects like radial distortions, which not only hamper accurate imaging using existing endoscopic technology during fetal surgery, but also make acquired images less suitable for potentially very useful image computing applications, like real-time mosaicing. In this paper proposes a novel fabrication method to create an affordable, sterilizable calibration target suitable for use in a clinical setup. This method involves etching a calibration pattern by laser cutting a sandblasted stainless steel sheet. This target was validated using the camera calibration module provided by OpenCV, a state-of-the-art software library popular in the computer vision community.

Shape-from-Template.

We study a problem that we call Shape-from-Template, which is the problem of reconstructing the shape of a deformable surface from a single image and a 3D template. Current methods in the literature address the case of isometric deformations, and relax the isometry constraint to the convex inextensibility constraint, solved using the so-called maximum depth heuristic. We call these methods zeroth-order since they use image point locations (the zeroth-order differential structure) to solve the shape inference problem from a perspective image. We propose a novel class of methods that we call first-order. The key idea is to use both image point locations and their first-order differential structure. The latter can be easily extracted from a warp between the template and the input image. We give a unified problem formulation as a system of PDEs for isometric and conformal surfaces that we solve analytically. This has important consequences. First, it gives the first analytical algorithms to solve this type of reconstruction problems. Second, it gives the first algorithms to solve for the exact constraints. Third, it allows us to study the well-posedness of this type of reconstruction: we establish that isometric surfaces can be reconstructed unambiguously and that conformal surfaces can be reconstructed up to a few discrete ambiguities and a global scale. In the latter case, the candidate solution surfaces are obtained analytically. Experimental results on simulated and real data show that our isometric methods generally perform as well as or outperform state of the art approaches in terms of reconstruction accuracy, while our conformal methods largely outperform all isometric methods for extensible deformations.

Enhanced imaging colonoscopy facilitates dense motion-based 3D reconstruction.

We propose a novel approach for estimating a dense 3D model of neoplasia in colonoscopy using enhanced imaging endoscopy modalities. Estimating a dense 3D model of neoplasia is important to make 3D measurements and to classify the superficial lesions in standard frameworks such as the Paris classification. However, it is challenging to obtain decent dense 3D models using computer vision techniques such as Structure-from-Motion due to the lack of texture in conventional (white light) colonoscopy. Therefore, we propose to use enhanced imaging endoscopy modalities such as Narrow Band Imaging and chromoendoscopy to facilitate the 3D reconstruction process. Thanks to the use of these enhanced endoscopy techniques, visualization is improved, resulting in more reliable feature tracks and 3D reconstruction results. We first build a sparse 3D model of neoplasia using Structure-from-Motion from enhanced endoscopy imagery. Then, the sparse reconstruction is densified using a Multi-View Stereo approach, and finally the dense 3D point cloud is transformed into a mesh by means of Poisson surface reconstruction. The obtained dense 3D models facilitate classification of neoplasia in the Paris classification, in which the 3D size and the shape of the neoplasia play a major role in the diagnosis.