Transbronchial Biopsy with Virtual CT Bronchoscopy and Nodal Highlighting

Abstract

Transbronchial biopsy to sample lymph nodes and tumors that are not visible at endoscopy has a poor (<50%) success rate. These nodes can be highlighted easily at virtual computed tomographic (CT) bronchoscopy to provide a guide. This study was performed to evaluate if the addition of this information to the bronchoscopist improved the success rate of transbronchial biopsy of subcarinal and aortopulmonary lymph nodes. The addition of virtual CT bronchoscopy with lymph node highlighting significantly (P < .5) increased biopsy success rates for pretracheal, hilar, and high pretracheal adenopathy.

Computer-based system for the virtual-endoscopic guidance of bronchoscopy

“…It accounts for nearly 30% of all cancer deaths and has a five-year survival rate under 15%. The diagnosis of lung cancer occurs in two stages: (1) three-dimensional (3D) computed-tomography (CT) image assessment; and (2) bronchoscopy [2][3][4].…”

3D CT-Video Fusion for Image-Guided Bronchoscopy

“…Trans-bronchial needle aspiration (TBNA) is often used to sample lymph nodes for pathology [2][3][4][5]. To perform TBNA, the physician first visually assesses a patient's computed tomography (CT) scan to identify suspect lymph-node sites.…”

“…These proposals, summarized below, show promise, but can be improved upon: (1) McAdams et al and Hopper et al applied 3D CT-based analysis (virtual bronchoscopy) to lymph-node biopsy, but these efforts involved no live guidance [3,4]; (2) Bricault et al did rudimentary registration of 3D CT chest data to videobronchoscopy, but their results were limited [17]; (3) other researchers have proposed a fundamentally different approach whereby an extra electromagnetic sensor is attached to the bronchoscope tip | this approach also requires considerable extra hardware [5,18,19]; (4) Mori et al registered videobronchoscopy images directly to rendered 3D CT images and also proposed including an electromagnetic sensor, but their efforts do not work interactively [20,21].…”

“…where h(V ) and h(CT) are entropy-like quantities for I V and I CT , and h(V, CT ) is a joint entropy between I V and I CT ; h(V ) is given by (4) where p V (k) is the probability density function (pdf) for the pixel gray-level values contained in I V , p V,CT (k, l) is the joint gray-scale histogram between I V and I CT , and the w k (and w l ) are weights. The pdf p V (k) is estimated as the normalized histogram of I V , per:…”

“…while the quantity p V,CT (k, l) is the 2D joint intensity pdf between the two images I CT and I V defined as (6) In ( (4). Finally, the weights are set to favor the more informative darker pixels:…”