Interactive CT-Video Registration for the Continuous Guidance of Bronchoscopy

Abstract

Bronchoscopy is a major step in lung cancer staging. To perform bronchoscopy, the physician uses a procedure plan, derived from a patient’s 3D computed-tomography (CT) chest scan, to navigate the bronchoscope through the lung airways. Unfortunately, physicians vary greatly in their ability to perform bronchoscopy. As a result, image-guided bronchoscopy systems, drawing upon the concept of CT-based virtual bronchoscopy (VB), have been proposed. These systems attempt to register the bronchoscope’s live position within the chest to a CT-based virtual chest space. Recent methods, which register the bronchoscopic video to CT-based endoluminal airway renderings, show promise but do not enable continuous real-time guidance. We present a CT-video registration method inspired by computer-vision innovations in the fields of image alignment and image-based rendering. In particular, motivated by the Lucas–Kanade algorithm, we propose an inverse-compositional framework built around a gradient-based optimization procedure. We next propose an implementation of the framework suitable for image-guided bronchoscopy. Laboratory tests, involving both single frames and continuous video sequences, demonstrate the robustness and accuracy of the method. Benchmark timing tests indicate that the method can run continuously at 300 frames/s, well beyond the real-time bronchoscopic video rate of 30 frames/s. This compares extremely favorably to the ≥1 s/frame speeds of other methods and indicates the method’s potential for real-time continuous registration. A human phantom study confirms the method’s efficacy for real-time guidance in a controlled setting, and, hence, points the way toward the first interactive CT-video registration approach for image-guided bronchoscopy. Along this line, we demonstrate the method’s efficacy in a complete guidance system by presenting a clinical study involving lung cancer patients.

Construction of a multimodal CT-video chest model

“…It is important to note that all bronchoscopic video frames are corrected for barrel distortion prior to any video processing and to also facilitate proper CT-video registration later. 8,31,32 Video analysis begins with the selection of a manually marked video frame within an input bronchoscopicvideo stream and with initialization of the first sub-map. Next, video parsing, the main processing block, processes the entire video sequence to arrive at a series of representative sub-maps.…”

“…First, Bronchoscopy Path-History Construction employs the processing flow below: (a) Interactively locate an MDCT-based endoluminal rendering I Θ CT that appears to be at approximately the same airway-tree pose as I K (i), where Θ is the standard 6-parameter pose vector indicating the virtual camera's field of view within the MDCT-based airway tree. 8,30 (b) Apply CT-video registration to the image pair {I Θ CT , I K (i)} to arrive at a pose Θ i that aligns the MDCT-based virtual space and real space (as imaged by the video); i.e., after registration, the following relationship holds:…”

“…30 CT-video registration (step 1. (b)) uses the heavily validated method of Merritt et al 8 Finally, in CT-Video Model Construction, we use the previously developed methodology of Rai and Higgins for video-to-CT texture mapping. 18 …”

“…The system builds upon our previous CT-video efforts described by Higgins et al 7,8,27,30 In particular, we augment this system to allow for the storage and interaction of all data elements in the model. Next, we have made initial prototypes of the following tools: 1.…”

Multi-view Stereo and Advanced Navigation for Transanal Endoscopic Microsurgery

“…Moreover, contrary to information augmentation in bronchosopy [8] or sinus skull-base surgery [9], two procedures that also involve intraluminal interventions, in TEM the endoscope does not exhibit a large travel range and does not allow the use of bifurcations as landmarks between the endoscopic view and the preoperative volume. It lends itself, however, to two important patientspecific priors: a) the endoscope is inserted close to the centreline of the rectum, and b) the tumour to be excised is identifiable in the endoscopic view and in the MRI/CT.…”