Pathological Site Retargeting under Tissue Deformation Using Geometrical Association and Tracking

Abstract

Abstract. Recent advances in microscopic detection techniques include fluorescence spectroscopy, fibred confocal microscopy and optical coherence tomography. These methods can be integrated with miniaturised probes to assist endoscopy, thus enabling diseases to be detected at an early and pre-invasive stage, forgoing the need for histopathological samples and off-line analysis. Since optical-based biopsy does not leave visible marks after sampling, it is important to track the biopsy sites to enable accurate retargeting and subsequent serial examination. In this paper, a novel approach is proposed for pathological site retargeting in gastroscopic examinations. The proposed method is based on affine deformation modelling with geometrical association combined with cascaded online learning and tracking. It provides online in vivo retargeting, and is able to track pathological sites in the presence of tissue deformation. It is also robust to partial occlusions and can be applied to a range of imaging probes including confocal laser endomicroscopy.

Online Scene Association for Endoscopic Navigation

“…The ground truth data of landmark detection and scene association were obtained from expert observers. For landmark detection, we compare the results of the proposed online scene association (OSA) with the orignal [1] (without pairwise geometry learning). For scene association, our approach (OSA) is compared with two standard image matching methods [11]: homography (Hom) and fundamental matrix (Fund).…”

“…Candidates passing this stage are then classified by calculating Normalised Cross Correlation (NCC) and finding the NN to all previous features for this landmark. The initialisation and online updating process of RF and the NN classifier is similar to [1,9].…”

“…However, surveillance endoscopy can be complicated, as the postprocessing stage often involves manual assessment and labelling of abnormalities on a large number of images, which is both time consuming and error prone. In cases when online retargeting is required for optical biopsy, the procedure becomes technically even more challenging [1].…”

Hybrid Retargeting for High-Speed Targeted Optical Biopsies

“…3). Similarly to [9], the associated sites are defined as the vertices of a set of n concentric, fixed radius pentagons. The set is constructed by incremental rotation of an initial pentagon by {0, θ, 2θ, .…”

“…Atasoy et al [2] address tissue deformation using a geometrically-constrained MRF model, while Allain et al [1] propose retargeting based on epipolar lines, derived from multiple views of a given biopsy site. Ye et al [9] model tissue deformation as locally affine and perform indirect retargeting by geometric association and online-learning. Partial occlusion is addressed by reinitialising the associated sites when the optical probe exits the field of view (determined by simple blob detection).…”