Stitching and Surface Reconstruction From Endoscopic Image Sequences: A Review of Applications and Methods

UC-NeRF: Uncertainty-Aware Conditional Neural Radiance Fields From Endoscopic Sparse Views

Visualizing surgical scenes is crucial for revealing internal anatomical structures during minimally invasive procedures. Novel View Synthesis is a vital technique that offers geometry and appearance reconstruction, enhancing understanding, planning, and decision-making in surgical scenes. Despite the impressive achievements of Neural Radiance Field (NeRF), its direct application to surgical scenes produces unsatisfying results due to two challenges: endoscopic sparse views and significant photometric inconsistencies. In this paper, we propose uncertainty-aware conditional NeRF for novel view synthesis to tackle the severe shape-radiance ambiguity from sparse surgical views. The core of UC-NeRF is to incorporate the multi-view uncertainty estimation to condition the neural radiance field for modeling the severe photometric inconsistencies adaptively. Specifically, our UC-NeRF first builds a consistency learner in the form of multi-view stereo network, to establish the geometric correspondence from sparse views and generate uncertainty estimation and feature priors. In neural rendering, we design a base-adaptive NeRF network to exploit the uncertainty estimation for explicitly handling the photometric inconsistencies. Furthermore, an uncertainty-guided geometry distillation is employed to enhance geometry learning. Experiments on the SCARED and Hamlyn datasets demonstrate our superior performance in rendering appearance and geometry, consistently outperforming the current state-of-the-art approaches. Our code will be released at https://github.com/wrld/UC-NeRF.

Unsupervised neural network-based image stitching method for bladder endoscopy

Bladder endoscopy enables the observation of intravesical lesion characteristics, making it an essential tool in urology. Image stitching techniques are commonly employed to expand the field of view of bladder endoscopy. Traditional image stitching methods rely on feature matching. In recent years, deep-learning techniques have garnered significant attention in the field of computer vision. However, the commonly employed supervised learning approaches often require a substantial amount of labeled data, which can be challenging to acquire, especially in the context of medical data. To address this limitation, this study proposes an unsupervised neural network-based image stitching method for bladder endoscopy, which eliminates the need for labeled datasets. The method comprises two modules: an unsupervised alignment network and an unsupervised fusion network. In the unsupervised alignment network, we employed feature convolution, regression networks, and linear transformations to align images. In the unsupervised fusion network, we achieved image fusion from features to pixel by simultaneously eliminating artifacts and enhancing the resolution. Experiments demonstrated our method's consistent stitching success rate of 98.11% and robust image stitching accuracy at various resolutions. Our method eliminates sutures and flocculent debris from cystoscopy images, presenting good image smoothness while preserving rich textural features. Moreover, our method could successfully stitch challenging scenes such as dim and blurry scenes. Our application of unsupervised deep learning methods in the field of cystoscopy image stitching was successfully validated, laying the foundation for real-time panoramic stitching of bladder endoscopic video images. This advancement provides opportunities for the future development of computer-vision-assisted diagnostic systems for bladder cavities.

A Serious Game to Study Reduced Field of View in Keyhole Surgery: Development and Experimental Study

BACKGROUND

During keyhole surgery, the surgeon is required to perform highly demanding tasks while only being able to see part of the patient's anatomy. This limited field of view is widely cited as a key limitation of the procedure, and many computational methods have been proposed to overcome it. However, the precise effects of a limited field of view on task performance remain unknown due to the lack of tools to study these effects effectively.

OBJECTIVE

This paper describes our work on developing a serious game with 2 objectives: (1) to create an engaging game that communicates some of the challenges of keyhole surgery, and (2) to test the effect of a limited field of view on task performance. The development of a serious game that can be played by a wide range of participants will enable us to gather quantitative data on the effects of the reduced field of view on task performance. These data can inform the future development of technologies to help surgeons reduce the impact of a limited field of view on clinical outcomes for patients. The game is open source and may be adapted and used by other researchers to study related problems.

METHODS

We implemented an open-source serious game in JavaScript, inspired by the surgical task of selectively cauterizing blood vessels during twin-to-twin transfusion surgery. During the game, the player is required to identify and cut the correct blood vessel under different fields of view and varying levels of vascular complexity. We conducted a quantitative analysis of task performance time under different conditions and a formative analysis of the game using participant questionnaires.

RESULTS

We recruited 25 players to test the game and recorded their task performance time, accuracy, and qualitative metrics. Reducing the field of view resulted in participants taking significantly longer (P<.001) to perform otherwise identical tasks (mean 6.4 seconds, 95% CI 5.0-7.8 seconds vs mean 13.6 seconds, 95% CI 10.3-16.9 seconds). Participants found the game engaging and agreed that it enhanced their understanding of the limited field of view during keyhole surgery.

CONCLUSIONS

We recruited 25 players to test the game and recorded their task performance time, accuracy, and qualitative metrics. Reducing the field of view resulted in participants taking statistically significantly longer (16.4 vs 9.8 seconds; P=.05) to perform otherwise identical tasks. Participants found the game engaging and agreed that it enhanced their understanding of the limited field of view during keyhole surgery.

Bladder image stitching algorithm for navigation and referencing using a standard cystoscope

To aid in the diagnosis, monitoring, and surveillance of bladder carcinoma, this study aimed to develop and test an algorithm that creates a referenceable bladder map rendered from standard cystoscopy videos without the need for specialized equipment. A vision-based algorithm was developed to generate 2D bladder maps from individual video frames, by sequentially stitching image frames based on matching surface features, and subsequently localize and track frames during reevaluation. The algorithm was developed and calibrated in a 2D model and 3D anthropomorphic bladder phantom. The performance was evaluated in vivo in swine and with retrospective clinical cystoscopy video. Results showed that the algorithm was capable of capturing and stitching intravesical images with different sweeping patterns. Between 93% and 99% of frames had sufficient features for bladder map generation. Upon reevaluation, the cystoscope accurately localized a frame within 4.5 s. In swine, a virtual mucosal surface map was generated that matched the explant anatomy. A surface map could be generated based on archived patient cystoscopy images. This tool could aid recording and referencing pathologic findings and biopsy or treatment locations for subsequent procedures and may have utility in patients with metachronous bladder cancer and in low-resource settings.

Image-Based 3D Reconstruction in Laparoscopy: A Review Focusing on the Quantitative Evaluation by Applying the Reconstruction Error

Image-based 3D reconstruction enables laparoscopic applications as image-guided navigation and (autonomous) robot-assisted interventions, which require a high accuracy. The review's purpose is to present the accuracy of different techniques to label the most promising. A systematic literature search with PubMed and google scholar from 2015 to 2023 was applied by following the framework of "Review articles: purpose, process, and structure". Articles were considered when presenting a quantitative evaluation (root mean squared error and mean absolute error) of the reconstruction error (Euclidean distance between real and reconstructed surface). The search provides 995 articles, which were reduced to 48 articles after applying exclusion criteria. From these, a reconstruction error data set could be generated for the techniques of stereo vision, Shape-from-Motion, Simultaneous Localization and Mapping, deep-learning, and structured light. The reconstruction error varies from below one millimeter to higher than ten millimeters-with deep-learning and Simultaneous Localization and Mapping delivering the best results under intraoperative conditions. The high variance emerges from different experimental conditions. In conclusion, submillimeter accuracy is challenging, but promising image-based 3D reconstruction techniques could be identified. For future research, we recommend computing the reconstruction error for comparison purposes and use ex/in vivo organs as reference objects for realistic experiments.

Surgical Tattoos in Infrared: A Dataset for Quantifying Tissue Tracking and Mapping

Quantifying performance of methods for tracking and mapping tissue in endoscopic environments is essential for enabling image guidance and automation of medical interventions and surgery. Datasets developed so far either use rigid environments, visible markers, or require annotators to label salient points in videos after collection. These are respectively: not general, visible to algorithms, or costly and error-prone. We introduce a novel labeling methodology along with a dataset that uses said methodology, Surgical Tattoos in Infrared (STIR). STIR has labels that are persistent but invisible to visible spectrum algorithms. This is done by labelling tissue points with IR-fluorescent dye, indocyanine green (ICG), and then collecting visible light video clips. STIR comprises hundreds of stereo video clips in both in vivo and ex vivo scenes with start and end points labelled in the IR spectrum. With over 3,000 labelled points, STIR will help to quantify and enable better analysis of tracking and mapping methods. After introducing STIR, we analyze multiple different frame-based tracking methods on STIR using both 3D and 2D endpoint error and accuracy metrics. STIR is available at https://dx.doi.org/10.21227/w8g4-g548.

3D reconstruction from endoscopy images: A survey

Three-dimensional reconstruction of images acquired through endoscopes is playing a vital role in an increasing number of medical applications. Endoscopes used in the clinic are commonly classified as monocular endoscopes and binocular endoscopes. We have reviewed the classification of methods for depth estimation according to the type of endoscope. Basically, depth estimation relies on feature matching of images and multi-view geometry theory. However, these traditional techniques have many problems in the endoscopic environment. With the increasing development of deep learning techniques, there is a growing number of works based on learning methods to address challenges such as inconsistent illumination and texture sparsity. We have reviewed over 170 papers published in the 10 years from 2013 to 2023. The commonly used public datasets and performance metrics are summarized. We also give a taxonomy of methods and analyze the advantages and drawbacks of algorithms. Summary tables and result atlas are listed to facilitate the comparison of qualitative and quantitative performance of different methods in each category. In addition, we summarize commonly used scene representation methods in endoscopy and speculate on the prospects of deep estimation research in medical applications. We also compare the robustness performance, processing time, and scene representation of the methods to facilitate doctors and researchers in selecting appropriate methods based on surgical applications.

Tracking and mapping in medical computer vision: A review

As computer vision algorithms increase in capability, their applications in clinical systems will become more pervasive. These applications include: diagnostics, such as colonoscopy and bronchoscopy; guiding biopsies, minimally invasive interventions, and surgery; automating instrument motion; and providing image guidance using pre-operative scans. Many of these applications depend on the specific visual nature of medical scenes and require designing algorithms to perform in this environment. In this review, we provide an update to the field of camera-based tracking and scene mapping in surgery and diagnostics in medical computer vision. We begin with describing our review process, which results in a final list of 515 papers that we cover. We then give a high-level summary of the state of the art and provide relevant background for those who need tracking and mapping for their clinical applications. After which, we review datasets provided in the field and the clinical needs that motivate their design. Then, we delve into the algorithmic side, and summarize recent developments. This summary should be especially useful for algorithm designers and to those looking to understand the capability of off-the-shelf methods. We maintain focus on algorithms for deformable environments while also reviewing the essential building blocks in rigid tracking and mapping since there is a large amount of crossover in methods. With the field summarized, we discuss the current state of the tracking and mapping methods along with needs for future algorithms, needs for quantification, and the viability of clinical applications. We then provide some research directions and questions. We conclude that new methods need to be designed or combined to support clinical applications in deformable environments, and more focus needs to be put into collecting datasets for training and evaluation.

Comparative Analysis of Color Space and Channel, Detector, and Descriptor for Feature-Based Image Registration

The current study aimed to quantify the value of color spaces and channels as a potential superior replacement for standard grayscale images, as well as the relative performance of open-source detectors and descriptors for general feature-based image registration purposes, based on a large benchmark dataset. The public dataset UDIS-D, with 1106 diverse image pairs, was selected. In total, 21 color spaces or channels including RGB, XYZ, Y'CrCb, HLS, L*a*b* and their corresponding channels in addition to grayscale, nine feature detectors including AKAZE, BRISK, CSE, FAST, HL, KAZE, ORB, SIFT, and TBMR, and 11 feature descriptors including AKAZE, BB, BRIEF, BRISK, DAISY, FREAK, KAZE, LATCH, ORB, SIFT, and VGG were evaluated according to reprojection error (RE), root mean square error (RMSE), structural similarity index measure (SSIM), registration failure rate, and feature number, based on 1,950,984 image registrations. No meaningful benefits from color space or channel were observed, although XYZ, RGB color space and L* color channel were able to outperform grayscale by a very minor margin. Per the dataset, the best-performing color space or channel, detector, and descriptor were XYZ/RGB, SIFT/FAST, and AKAZE. The most robust color space or channel, detector, and descriptor were L*a*b*, TBMR, and VGG. The color channel, detector, and descriptor with the most initial detector features and final homography features were Z/L*, FAST, and KAZE. In terms of the best overall unfailing combinations, XYZ/RGB+SIFT/FAST+VGG/SIFT seemed to provide the highest image registration quality, while Z+FAST+VGG provided the most image features.

Surface Reconstruction of the Pediatric Larynx via Structure from Motion Photogrammetry: A Pilot Study

Endoscopy is the gold standard for characterizing pediatric airway disorders, however, it is limited for quantitative analysis due to lack of three-dimensional (3D) vision and poor stereotactic depth perception. We utilize structure from motion (SfM) photogrammetry, to reconstruct 3D surfaces of pathologic and healthy pediatric larynges from monocular two-dimensional (2D) endoscopy. Models of pediatric subglottic stenosis were 3D printed and airway endoscopies were simulated. 3D surfaces were successfully reconstructed from endoscopic videos of all models using an SfM analysis toolkit. Average subglottic surface error between SfM reconstructed surfaces and 3D printed models was 0.65 mm as measured by Modified Hausdorff Distance. Average volumetric similarity between SfM surfaces and printed models was 0.82 as measured by Jaccard Index. SfM can be used to accurately reconstruct 3D surface renderings of the larynx from 2D endoscopy video. This technique has immense potential for use in quantitative analysis of airway geometry and virtual surgical planning.

A half jaw panoramic stitching method of intraoral endoscopy images based on dental arch arrangement

To address the challenges of repetitive and low-texture features in intraoral endoscopic images, a novel methodology for stitching panoramic half jaw images of the oral cavity is proposed. Initially, an enhanced self-attention mechanism guided by Time-Weighting concepts is employed to augment the clustering potential of feature points, thereby increasing the number of matched features. Subsequently, a combination of the Sinkhorn algorithm and Random Sample Consensus (RANSAC) is utilized to maximize the count of matched feature pairs, accurately remove outliers and minimize error. Last, to address the unique spatial alignment among intraoral endoscopic images, a wavelet transform and weighted fusion algorithm based on dental arch arrangement in intraoral endoscopic images have been developed, specifically for use in the fusion stage of intraoral endoscopic images. This enables the local oral images to be precisely positioned along the dental arch, and seamless stitching is achieved through wavelet transformation and a gradual weighted fusion technique. Experimental results demonstrate that this method yields promising outcomes in panoramic stitching tasks for intraoral endoscopic images, achieving a matching accuracy of 84.6% and a recall rate of 78.4% in a dataset with an average overlap of 35%. A novel solution for panoramic stitching of intraoral endoscopic images is provided by this method.

3-D and 2-D reconstruction of bladders for the assessment of inter-session detection of tissue changes: a proof of concept

PURPOSE

Abnormalities in the bladder wall require careful investigation regarding type, spatial position and invasiveness. Construction of a 3-D model of the bladder is helpful to ensure adequate coverage of the scanning procedure, quantitative comparison of bladder wall textures between successive sessions and finding back previously discovered abnormalities.

METHODS

Videos of both an in vivo bladder and a textured bladder phantom were acquired. Structure-from-motion and bundle adjustment algorithms were used to construct a 3-D point cloud, approximate it by a surface mesh, texture it with the back-projected camera frames and draw the corresponding 2-D atlas. Reconstructions of successive sessions were compared; those of the bladder phantom were co-registered, transformed using 3-D thin plate splines and post-processed to highlight significant changes in texture.

RESULTS

The reconstruction algorithms of the presented workflow were able to construct 3-D models and corresponding 2-D atlas of both the in vivo bladder and the bladder phantom. For the in vivo bladder the portion of the reconstructed surface area was 58% and 79% for the pre- and post-operative scan, respectively. For the bladder phantom the full surface was reconstructed and the mean reprojection error was 0.081 mm (range 0-0.79 mm). In inter-session comparison the changes in texture were correctly indicated for all six locations.

CONCLUSION

The proposed proof of concept was able to perform 3-D and 2-D reconstruction of an in vivo bladder wall based on a set of monocular images. In a phantom study the computer vision algorithms were also effective in co-registering reconstructions of successive sessions and highlighting texture changes between sessions. These techniques may be useful for detecting, monitoring and revisiting suspicious lesions.

Texture-less surface reconstruction using shape-based image augmentation

The development of intelligent Robot-Assisted Minimally Invasive Surgery demands geometric reconstruction from endoscopic images. However, images of human tissue surfaces are commonly texture-less. Obtaining the dense depth map of a texture-less scene is very difficult because traditional feature-based 3D reconstruction methods cannot detect enough features to build dense correspondences for depth computation. Given this problem, this study proposes a novel reconstruction method based on our shape-based image augmentation method. The main contribution of this method is the provision of a novel means to resolve the texture-less problem mainly on the input data level. In our method, we first calculate two shape gradient maps using Shape-From-Shading (SFS) method and we build Fast Point Feature Histogram (FPFH) 3D descriptor map according to the shape. Second, a series of augmented images can be computed by combining shape gradient maps, FPFH map, and the original image with different weights. Finally, we detect features on the new augmented images. Based on feature calculated sparse depth information and SFS calculated dense shape information, we further integrate a rectified dense depth map. Experiments show that our method can reconstruct texture-less surfaces with good accuracy.

Optical biopsy mapping on endoscopic image mosaics with a marker-free probe

Gastric cancer is the second leading cause of cancer-related deaths worldwide. Early diagnosis significantly increases the chances of survival; therefore, improved assisted exploration and screening techniques are necessary. Previously, we made use of an augmented multi-spectral endoscope by inserting an optical probe into the instrumentation channel. However, the limited field of view and the lack of markings left by optical biopsies on the tissue complicate the navigation and revisit of the suspect areas probed in-vivo. In this contribution two innovative tools are introduced to significantly increase the traceability and monitoring of patients in clinical practice: (i) video mosaicing to build a more comprehensive and panoramic view of large gastric areas; (ii) optical biopsy targeting and registration with the endoscopic images. The proposed optical flow-based mosaicing technique selects images that minimize texture discontinuities and is robust despite the lack of texture and illumination variations. The optical biopsy targeting is based on automatic tracking of a free-marker probe in the endoscopic view using deep learning to dynamically estimate its pose during exploration. The accuracy of pose estimation is sufficient to ensure a precise overlapping of the standard white-light color image and the hyperspectral probe image, assuming that the small target area of the organ is almost flat. This allows the mapping of all spatio-temporally tracked biopsy sites onto the panoramic mosaic. Experimental validations are carried out from videos acquired on patients in hospital. The proposed technique is purely software-based and therefore easily integrable into clinical practice. It is also generic and compatible to any imaging modality that connects to a fiberscope.

Study on Reconstruction and Feature Tracking of Silicone Heart 3D Surface

At present, feature-based 3D reconstruction and tracking technology is widely applied in the medical field. In minimally invasive surgery, the surgeon can achieve three-dimensional reconstruction through the images obtained by the endoscope in the human body, restore the three-dimensional scene of the area to be operated on, and track the motion of the soft tissue surface. This enables doctors to have a clearer understanding of the location depth of the surgical area, greatly reducing the negative impact of 2D image defects and ensuring smooth operation. In this study, firstly, the 3D coordinates of each feature point are calculated by using the parameters of the parallel binocular endoscope and the spatial geometric constraints. At the same time, the discrete feature points are divided into multiple triangles using the Delaunay triangulation method. Then, the 3D coordinates of feature points and the division results of each triangle are combined to complete the 3D surface reconstruction. Combined with the feature matching method based on convolutional neural network, feature tracking is realized by calculating the three-dimensional coordinate changes of the same feature point in different frames. Finally, experiments are carried out on the endoscope image to complete the 3D surface reconstruction and feature tracking.

Intensity-Mosaic: automatic panorama mosaicking of disordered images with insufficient features

Purpose: Handling low-quality and few-feature medical images is a challenging task in automatic panorama mosaicking. Current mosaicking methods for disordered input images are based on feature point matching, whereas in this case intensity-based registration achieves better performance than feature-point registration methods. We propose a mosaicking method that enables the use of mutual information (MI) registration for mosaicking randomly ordered input images with insufficient features. Approach: Dimensionality reduction is used to map disordered input images into a low dimensional space. Based on the low dimensional representation, the image global correspondence can be recognized efficiently. For adjacent image pairs, we optimize the MI metric for registration. The panorama is then created after image blending. We demonstrate our method on relatively lower-cost handheld devices that acquire images from the retina in vivo, kidney ex vivo, and bladder phantom, all of which contain sparse features. Results: Our method is compared with three baselines: AutoStitch, "dimension reduction + SIFT," and "MI-Only." Our method compared to the first two feature-point based methods exhibits 1.25 (ex vivo microscope dataset) to two times (in vivo retina dataset) rate of mosaic completion, and MI-Only has the lowest complete rate among three datasets. When comparing the subsequent complete mosaics, our target registration errors can be 2.2 and 3.8 times reduced when using the microscopy and bladder phantom datasets. Conclusions: Using dimensional reduction increases the success rate of detecting adjacent images, which makes MI-based registration feasible and narrows the search range of MI optimization. To the best of our knowledge, this is the first mosaicking method that allows automatic stitching of disordered images with intensity-based alignment, which provides more robust and accurate results when there are insufficient features for classic mosaicking methods.

Real-Time Nonrigid Mosaicking of Laparoscopy Images

The ability to extend the field of view of laparoscopy images can help the surgeons to obtain a better understanding of the anatomical context. However, due to tissue deformation, complex camera motion and significant three-dimensional (3D) anatomical surface, image pixels may have non-rigid deformation and traditional mosaicking methods cannot work robustly for laparoscopy images in real-time. To solve this problem, a novel two-dimensional (2D) non-rigid simultaneous localization and mapping (SLAM) system is proposed in this paper, which is able to compensate for the deformation of pixels and perform image mosaicking in real-time. The key algorithm of this 2D non-rigid SLAM system is the expectation maximization and dual quaternion (EMDQ) algorithm, which can generate smooth and dense deformation field from sparse and noisy image feature matches in real-time. An uncertainty-based loop closing method has been proposed to reduce the accumulative errors. To achieve real-time performance, both CPU and GPU parallel computation technologies are used for dense mosaicking of all pixels. Experimental results on in vivo and synthetic data demonstrate the feasibility and accuracy of our non-rigid mosaicking method.

Real-Time Expanded Field-of-View for Minimally Invasive Surgery Using Multi-Camera Visual Simultaneous Localization and Mapping

Minimally invasive surgery is widely used because of its tremendous benefits to the patient. However, there are some challenges that surgeons face in this type of surgery, the most important of which is the narrow field of view. Therefore, we propose an approach to expand the field of view for minimally invasive surgery to enhance surgeons’ experience. It combines multiple views in real-time to produce a dynamic expanded view. The proposed approach extends the monocular Oriented features from an accelerated segment test and Rotated Binary robust independent elementary features—Simultaneous Localization And Mapping (ORB-SLAM) to work with a multi-camera setup. The ORB-SLAM’s three parallel threads, namely tracking, mapping and loop closing, are performed for each camera and new threads are added to calculate the relative cameras’ pose and to construct the expanded view. A new algorithm for estimating the optimal inter-camera correspondence matrix from a set of corresponding 3D map points is presented. This optimal transformation is then used to produce the final view. The proposed approach was evaluated using both human models and in vivo data. The evaluation results of the proposed correspondence matrix estimation algorithm prove its ability to reduce the error and to produce an accurate transformation. The results also show that when other approaches fail, the proposed approach can produce an expanded view. In this work, a real-time dynamic field-of-view expansion approach that can work in all situations regardless of images’ overlap is proposed. It outperforms the previous approaches and can also work at 21 fps.

Stomach 3D Reconstruction Using Virtual Chromoendoscopic Images

Gastric endoscopy is a golden standard in the clinical process that enables medical practitioners to diagnose various lesions inside a patient’s stomach. If a lesion is found, a success in identifying the location of the found lesion relative to the global view of the stomach will lead to better decision making for the next clinical treatment. Our previous research showed that the lesion localization could be achieved by reconstructing the whole stomach shape from chromoendoscopic indigo carmine (IC) dye-sprayed images using a structure-from-motion (SfM) pipeline. However, spraying the IC dye to the whole stomach requires additional time, which is not desirable for both patients and practitioners. Our objective is to propose an alternative way to achieve whole stomach 3D reconstruction without the need of the IC dye. We generate virtual IC-sprayed (VIC) images based on image-to-image style translation trained on unpaired real no-IC and IC-sprayed images, where we have investigated the effect of input and output color channel selection for generating the VIC images. We validate our reconstruction results by comparing them with the results using real IC-sprayed images and confirm that the obtained stomach 3D structures are comparable to each other. We also propose a local reconstruction technique to obtain a more detailed surface and texture around an interesting region. The proposed method achieves the whole stomach reconstruction without the need of real IC dye using SfM. We have found that translating no-IC green-channel images to IC-sprayed red-channel images gives the best SfM reconstruction result. Clinical impact We offer a method of the frame localization and local 3D reconstruction of a found gastric lesion using standard endoscopy images, leading to better clinical decision.

Robust Mosaicing of Endomicroscopic Videos via Context-Weighted Correlation Ratio

Probe-based confocal laser endomicroscopy (pCLE) is a promising imaging tool that provides in situ and in vivo optical imaging to perform real-time pathological assessments. However, due to limited field of view, it is difficult for clinicians to get a full understanding of the scanned tissues. In this paper, we develop a novel mosaicing framework to assemble all frame sequences into a full view image. First, a hybrid rigid registration that combines feature matching and template matching is presented to achieve a global alignment of all frames. Then, the parametric free-form deformation (FFD) model with a multiresolution architecture is implemented to accommodate non-rigid tissue distortions. More importantly, we devise a robust similarity metric called context-weighted correlation ratio (CWCR) to promote registration accuracy, where spatial and geometric contexts are incorporated into the estimation of functional intensity dependence. Experiments on both robotic setup and manual manipulation have demonstrated that the proposed scheme significantly precedes some state-of-the-art mosaicing schemes in the presence of intensity fluctuations, insufficient overlap and tissue distortions. Moreover, the comparisons of the proposed CWCR metric and two other metrics have validated the effectiveness of the context-weighted strategy in quantifying the differences between two frames. Benefiting from more rational and delicate mosaics, the proposed scheme is more suitable to instruct diagnosis and treatment during optical biopsies.

Deep learning-based anatomical site classification for upper gastrointestinal endoscopy

Purpose

Upper gastrointestinal (GI) endoscopic image documentation has provided an efficient, low-cost solution to address quality control for endoscopic reporting. The problem is, however, challenging for computer-assisted techniques, because different sites have similar appearances. Additionally, across different patients, site appearance variation may be large and inconsistent. Therefore, according to the British and modified Japanese guidelines, we propose a set of oesophagogastroduodenoscopy (EGD) images to be routinely captured and evaluate its efficiency for deep learning-based classification methods.

Methods

A novel EGD image dataset standardising upper GI endoscopy to several steps is established following landmarks proposed in guidelines and annotated by an expert clinician. To demonstrate the discrimination of proposed landmarks that enable the generation of an automated endoscopic report, we train several deep learning-based classification models utilising the well-annotated images.

Results

We report results for a clinical dataset composed of 211 patients (comprising a total of 3704 EGD images) acquired during routine upper GI endoscopic examinations. We find close agreement between predicted labels using our method and the ground truth labelled by human experts. We observe the limitation of current static image classification scheme for EGD image classification.

Conclusion

Our study presents a framework for developing automated EGD reports using deep learning. We demonstrate that our method is feasible to address EGD image classification and can lead towards improved performance and additionally qualitatively demonstrate its performance on our dataset.

Image computing for fibre-bundle endomicroscopy: A review

Endomicroscopy is an emerging imaging modality, that facilitates the acquisition of in vivo, in situ optical biopsies, assisting diagnostic and potentially therapeutic interventions. While there is a diverse and constantly expanding range of commercial and experimental optical biopsy platforms available, fibre-bundle endomicroscopy is currently the most widely used platform and is approved for clinical use in a range of clinical indications. Miniaturised, flexible fibre-bundles, guided through the working channel of endoscopes, needles and catheters, enable high-resolution imaging across a variety of organ systems. Yet, the nature of image acquisition though a fibre-bundle gives rise to several inherent characteristics and limitations necessitating novel and effective image pre- and post-processing algorithms, ranging from image formation, enhancement and mosaicing to pathology detection and quantification. This paper introduces the underlying technology and most prevalent clinical applications of fibre-bundle endomicroscopy, and provides a comprehensive, up-to-date, review of relevant image reconstruction, analysis and understanding/inference methodologies. Furthermore, current limitations as well as future challenges and opportunities in fibre-bundle endomicroscopy computing are identified and discussed.

Confidence-based dynamic optimization model for biomedical image mosaicking

Biomedical image mosaicking is a trending topic. It consists of computing a single large image from multiple observations and becomes a challenging task when said observations barely overlap or are subject to illumination changes, poor resolution, blur, or either highly textured or predominantly homogeneous content. Because such challenges are common in biomedical images, classical keypoint/feature-based methods perform poorly. In this paper, we propose a new framework based on pairwise template matching coupled with a constrained, confidence-driven global optimization strategy to solve the issue of microscopic biomedical image mosaicking. First we provide experimental results that show significant improvement on a subjective level. Then we describe a new validation strategy for objective assessment, which shows improvement as well.

2D reconstruction of small intestine's interior wall

Examining and interpreting of a large number of wireless endoscopic images from the gastrointestinal tract is a tiresome task for physicians. A practical solution is to automatically construct a two dimensional representation of the gastrointestinal tract for easy inspection. However, little has been done on wireless endoscopic image stitching, let alone systematic investigation. The proposed new wireless endoscopic image stitching method consists of two main steps to improve the accuracy and efficiency of image registration. First, the keypoints are extracted by Principle Component Analysis and Scale Invariant Feature Transform (PCA-SIFT) algorithm and refined with Maximum Likelihood Estimation SAmple Consensus (MLESAC) outlier removal to find the most reliable keypoints. Second, the optimal transformation parameters obtained from first step are fed to the Normalised Mutual Information (NMI) algorithm as an initial solution. With modified Marquardt-Levenberg search strategy in a multiscale framework, the NMI can find the optimal transformation parameters in the shortest time. The proposed methodology has been tested on two different datasets - one with real wireless endoscopic images and another with images obtained from Micro-Ball (a new wireless cubic endoscopy system with six image sensors). The results have demonstrated the accuracy and robustness of the proposed methodology both visually and quantitatively - registration residual error of 0.93±0.33 pixels on 2500 real endoscopy image pairs and residual error accumulation of 16.59 pixels and without affecting the visual registration quality on stitching 152 images of Micro-Ball.

Adaptive Image Enhancement Based on Guide Image and Fraction-Power Transformation for Wireless Capsule Endoscopy

Good image quality of the wireless capsule endoscopy (WCE) is the key for doctors to diagnose gastrointestinal (GI) tract diseases. However, the poor illumination, limited performance of the camera in WCE, and complex environment in the GI tract usually result in low-quality endoscopic images. Existing image enhancement methods only use the information of the image itself or multiple images of the same scene to accomplish the enhancement. In this paper, we propose an adaptive image enhancement method based on guide image and fraction-power transformation. First, intensities of endoscopic images are analyzed to assess the illumination conditions. Second, images captured under poor illumination conditions are enhanced by a brand-new image enhancement method called adaptive guide image based enhancement (AGIE). AGIE enhances low-quality images by using the information of a good quality image of the similar scene. Otherwise, images are enhanced by the proposed adaptive fraction-power transformation. Experimental results show that the proposed method improves the average intensity of endoscopic images by 64.20% and the average local entropy by 31.25%, which outperforms the state-of-art methods.

An enhanced hybrid tracking-mosaicking approach for surgical view expansion

The aim of this work is to overcome the narrow surgical field of view problem in minimally invasive surgery. We achieve this by combining multiple views of the camera-retractable trocar which can obtain surgical viewpoints different from laparoscopic view. However, the accuracy and time are essential factors in this process. Therefore, we tend to improve the accuracy of a hybrid tracking-mosaicking approach which can combine several views at high speed. Two improvements are presented and analyzed here. The first improvement utilizes two sharping methodologies to enhance the image quality. This enhancement, in turn, improves the interest point extraction process and increases the number of extracted points. In the second enhancement, the tracking accuracy is improved by applying a filtering methodology to select the set of valid flow vectors only. This process reduces the tracking error which may accumulate during tracking. The experimental evaluation, shows that these improvements enhance the final mosaicking accuracy and allows us to construct a more accurate expanded view.

Image-based 3D surface approximation of the bladder using structure-from-motion for enhanced cystoscopy based on phantom data

Bladder cancer is likely to recur after resection. For this reason, bladder cancer survivors often undergo follow-up cystoscopy for years after treatment to look for bladder cancer recurrence. 3D modeling of the bladder could provide more reliable cystoscopic documentation by giving an overall picture of the organ and tumor positions. However, 3D reconstruction of the urinary bladder based on endoscopic images is challenging. This is due to the small field of view of the endoscope, considerable image distortion, and occlusion by urea, blood or particles. In this paper, we will demonstrate a method for the conversion of uncalibrated, monocular, endoscopic videos of the bladder into a 3D model using structure-from-motion (SfM). First of all, frames are extracted from video sequences. Distortions are then corrected in a calibration procedure. Finally, the 3D reconstruction algorithm generates a sparse surface approximation of the bladder lining based on the corrected frames. This method was tested using an endoscopic video of a phantom that mimics the rich structure of the bladder. The reconstructed 3D model covered a large part of the object, with an average reprojection error of 1.15 pixels and a relative accuracy of 99.4%.

Efficient Bronchoscopic Video Summarization

Bronchoscopy enables many minimally invasive chest procedures for diseases such as lung cancer and asthma. Guided by the bronchoscope's video stream, a physician can navigate the complex three-dimensional (3-D) airway tree to collect tissue samples or administer a disease treatment. Unfortunately, physicians currently discard procedural video because of the overwhelming amount of data generated. Hence, they must rely on memory and anecdotal snapshots to document a procedure. We propose a robust automatic method for summarizing an endobronchial video stream. Inspired by the multimedia concept of the video summary and by research in other endoscopy domains, our method consists of three main steps: 1) shot segmentation, 2) motion analysis, and 3) keyframe selection. Overall, the method derives a true hierarchical decomposition, consisting of a shot set and constituent keyframe set, for a given procedural video. No other method to our knowledge gives such a structured summary for the raw, unscripted, unedited videos arising in endoscopy. Results show that our method more efficiently covers the observed endobronchial regions than other keyframe-selection approaches and is robust to parameter variations. Over a wide range of video sequences, our method required on average only 6.5% of available video frames to achieve a video coverage = 92.7%. We also demonstrate how the derived video summary facilitates direct fusion with a patient's 3-D chest computed-tomography scan in a system under development, thereby enabling efficient video browsing and retrieval through the complex airway tree.

SLAM-based dense surface reconstruction in monocular Minimally Invasive Surgery and its application to Augmented Reality

BACKGROUND AND OBJECTIVE

While Minimally Invasive Surgery (MIS) offers considerable benefits to patients, it also imposes big challenges on a surgeon's performance due to well-known issues and restrictions associated with the field of view (FOV), hand-eye misalignment and disorientation, as well as the lack of stereoscopic depth perception in monocular endoscopy. Augmented Reality (AR) technology can help to overcome these limitations by augmenting the real scene with annotations, labels, tumour measurements or even a 3D reconstruction of anatomy structures at the target surgical locations. However, previous research attempts of using AR technology in monocular MIS surgical scenes have been mainly focused on the information overlay without addressing correct spatial calibrations, which could lead to incorrect localization of annotations and labels, and inaccurate depth cues and tumour measurements. In this paper, we present a novel intra-operative dense surface reconstruction framework that is capable of providing geometry information from only monocular MIS videos for geometry-aware AR applications such as site measurements and depth cues. We address a number of compelling issues in augmenting a scene for a monocular MIS environment, such as drifting and inaccurate planar mapping.

METHODS

A state-of-the-art Simultaneous Localization And Mapping (SLAM) algorithm used in robotics has been extended to deal with monocular MIS surgical scenes for reliable endoscopic camera tracking and salient point mapping. A robust global 3D surface reconstruction framework has been developed for building a dense surface using only unorganized sparse point clouds extracted from the SLAM. The 3D surface reconstruction framework employs the Moving Least Squares (MLS) smoothing algorithm and the Poisson surface reconstruction framework for real time processing of the point clouds data set. Finally, the 3D geometric information of the surgical scene allows better understanding and accurate placement AR augmentations based on a robust 3D calibration.

RESULTS

We demonstrate the clinical relevance of our proposed system through two examples: (a) measurement of the surface; (b) depth cues in monocular endoscopy. The performance and accuracy evaluations of the proposed framework consist of two steps. First, we have created a computer-generated endoscopy simulation video to quantify the accuracy of the camera tracking by comparing the results of the video camera tracking with the recorded ground-truth camera trajectories. The accuracy of the surface reconstruction is assessed by evaluating the Root Mean Square Distance (RMSD) of surface vertices of the reconstructed mesh with that of the ground truth 3D models. An error of 1.24 mm for the camera trajectories has been obtained and the RMSD for surface reconstruction is 2.54 mm, which compare favourably with previous approaches. Second, in vivo laparoscopic videos are used to examine the quality of accurate AR based annotation and measurement, and the creation of depth cues. These results show the potential promise of our geometry-aware AR technology to be used in MIS surgical scenes.

CONCLUSIONS

The results show that the new framework is robust and accurate in dealing with challenging situations such as the rapid endoscopy camera movements in monocular MIS scenes. Both camera tracking and surface reconstruction based on a sparse point cloud are effective and operated in real-time. This demonstrates the potential of our algorithm for accurate AR localization and depth augmentation with geometric cues and correct surface measurements in MIS with monocular endoscopes.

Learning-based classification of informative laryngoscopic frames

BACKGROUND AND OBJECTIVE

Early-stage diagnosis of laryngeal cancer is of primary importance to reduce patient morbidity. Narrow-band imaging (NBI) endoscopy is commonly used for screening purposes, reducing the risks linked to a biopsy but at the cost of some drawbacks, such as large amount of data to review to make the diagnosis. The purpose of this paper is to present a strategy to perform automatic selection of informative endoscopic video frames, which can reduce the amount of data to process and potentially increase diagnosis performance.

METHODS

A new method to classify NBI endoscopic frames based on intensity, keypoint and image spatial content features is proposed. Support vector machines with the radial basis function and the one-versus-one scheme are used to classify frames as informative, blurred, with saliva or specular reflections, or underexposed.

RESULTS

When tested on a balanced set of 720 images from 18 different laryngoscopic videos, a classification recall of 91% was achieved for informative frames, significantly overcoming three state of the art methods (Wilcoxon rank-signed test, significance level = 0.05).

CONCLUSIONS

Due to the high performance in identifying informative frames, the approach is a valuable tool to perform informative frame selection, which can be potentially applied in different fields, such us computer-assisted diagnosis and endoscopic view expansion.

Retrieval and registration of long-range overlapping frames for scalable mosaicking of in vivo fetoscopy

Purpose

The standard clinical treatment of Twin-to-Twin transfusion syndrome consists in the photo-coagulation of undesired anastomoses located on the placenta which are responsible to a blood transfer between the two twins. While being the standard of care procedure, fetoscopy suffers from a limited field-of-view of the placenta resulting in missed anastomoses. To facilitate the task of the clinician, building a global map of the placenta providing a larger overview of the vascular network is highly desired.

Methods

To overcome the challenging visual conditions inherent to in vivo sequences (low contrast, obstructions or presence of artifacts, among others), we propose the following contributions: (1) robust pairwise registration is achieved by aligning the orientation of the image gradients, and (2) difficulties regarding long-range consistency (e.g. due to the presence of outliers) is tackled via a bag-of-word strategy, which identifies overlapping frames of the sequence to be registered regardless of their respective location in time.

Results

In addition to visual difficulties, in vivo sequences are characterised by the intrinsic absence of gold standard. We present mosaics motivating qualitatively our methodological choices and demonstrating their promising aspect. We also demonstrate semi-quantitatively, via visual inspection of registration results, the efficacy of our registration approach in comparison with two standard baselines.

Conclusion

This paper proposes the first approach for the construction of mosaics of placenta in in vivo fetoscopy sequences. Robustness to visual challenges during registration and long-range temporal consistency are proposed, offering first positive results on in vivo data for which standard mosaicking techniques are not applicable.

Electronic supplementary material

The online version of this article (10.1007/s11548-018-1728-4) contains supplementary material, which is available to authorized users.

Digital Mapping of the Urinary Bladder: Potential for Standardized Cystoscopy Reports

OBJECTIVE

To develop a standardized digital reporting tool for cystoscopy of the urinary bladder using panoramic imaging.

MATERIALS AND METHODS

An image processing and stitching software (Endorama) was developed to generate panoramic images from cystoscopy data. In a processing phase, algorithms were modulated and refined by reference to cystoscopy sequences (n = 30). Subsequently, standard systematic cystoscopies (n = 12) were recorded in patients undergoing transurethral resection of a bladder tumor to create panoramic images.

RESULTS

All sequences were applicable for the development and refinements of the software. Processing increasingly allowed the creation of images illustrating large parts of the bladder and relevant anatomic landmarks in different locations. The pathway covered by the endoscope during the intervention was illustrated as a route in the respective digital image. During the application phase, panoramic images were successfully created in 10 out of 12 cases. The resolution of the images was 4096 × 2048 pixels and the images required a median digital memory of 3.9 MB (3.4-5.7). The panoramic images illustrated 22 relevant findings of which 7 were papillary tumors.

CONCLUSION

High-quality digital panoramic maps of the urinary bladder were created using specifically processed data of videocystoscopy. In this preliminary series, relevant findings were illustrated in the respective image. Our tool may help improve standardization of cystoscopy reports and reduce interobserver variability.

Automated visibility map of the internal colon surface from colonoscopy video

PURPOSE

Optical colonoscopy is a prominent procedure by which clinicians examine the surface of the colon for cancerous polyps using a flexible colonoscope. One of the main concerns regarding the quality of the colonoscopy is to ensure that the whole colonic surface has been inspected for abnormalities. In this paper, we aim at estimating areas that have not been covered thoroughly by providing a map from the internal colon surface.

METHODS

Camera parameters were estimated using optical flow between consecutive colonoscopy frames. A cylinder model was fitted to the colon structure using 3D pseudo stereo vision and projected into each frame. A circumferential band from the cylinder was extracted to unroll the internal colon surface (band image). By registering these band images, drift in estimating camera motion could be reduced, and a visibility map of the colon surface could be generated, revealing uncovered areas by the colonoscope. Hidden areas behind haustral folds were ignored in this study. The method was validated on simulated and actual colonoscopy videos. The realistic simulated videos were generated using a colonoscopy simulator with known ground truth, and the actual colonoscopy videos were manually assessed by a clinical expert.

RESULTS

The proposed method obtained a sensitivity and precision of 98 and 96 % for detecting the number of uncovered areas on simulated data, whereas validation on real videos showed a sensitivity and precision of 96 and 78 %, respectively. Error in camera motion drift could be reduced by almost 50 % using results from band image registration.

CONCLUSION

Using a simple cylindrical model for the colon and reducing drift by registering band images allows for the generation of visibility maps. The current results also suggest that the provided feedback through the visibility map could enhance clinicians' awareness of uncovered areas, which in return could reduce the probability of missing polyps.

Automatic workflow for narrow-band laryngeal video stitching

In narrow band (NB) laryngeal endoscopy, the clinician usually positions the endoscope near the tissue for a correct inspection of possible vascular pattern alterations, indicative of laryngeal malignancies. The video is usually reviewed many times to refine the diagnosis, resulting in loss of time since the salient frames of the video are mixed with blurred, noisy, and redundant frames caused by the endoscope movements. The aim of this work is to provide to the clinician a unique larynx panorama, obtained through an automatic frame selection strategy to discard non-informative frames. Anisotropic diffusion filtering was exploited to lower the noise level while encouraging the selection of meaningful image features, and a feature-based stitching approach was carried out to generate the panorama. The frame selection strategy, tested on on six pathological NB endoscopic videos, was compared with standard strategies, as uniform and random sampling, showing higher performance of the subsequent stitching procedure, both visually, in terms of vascular structure preservation, and numerically, through a blur estimation metric.

Comparative study on surface reconstruction accuracy of stereo imaging devices for microsurgery

PURPOSE

Processing stereoscopic image data is an emerging field. Especially in microsurgery that requires sub-millimeter accuracy, application of stereo-based methods on endoscopic and microscopic scenarios is of major interest. In this context, direct comparison of stereo-based surface reconstruction applied to several camera settings is presented.

METHODS

A method for stereo matching is proposed and validated on in-vitro data. Demonstrating suitability for surgical scenarios, this method is applied to two custom-made stereo cameras, a miniaturized, bendable surgical endoscope and an operating microscope. Reconstruction accuracy is assessed on a custom-made reference sample. Subsequent to its fabrication, a coordinate measuring arm is used to acquire ground truth. Next, the sample is positioned by a robot at varying distances to each camera. Surface estimation is performed, while the specimen is localized based on. markers. Finally, the error between estimated surface and ground truth is computed.

RESULTS

Sample measurement with the coordinate measuring arm yields reliable ground truth data with a root-mean-square error of 11.2 μm. Overall surface reconstruction with analyzed cameras is quantified by a root-mean-square error of less than 0.18 mm. Microscope setting with the highest magnification yields the most accurate measurement, while the maximum deviation does not exceed 0.5 mm. Custom-made stereo cameras perform similar but with outliers of increased magnitude. Miniaturized, bendable surgical endoscope produces the maximum error of approximately 1.2 mm.

CONCLUSIONS

Reconstruction results reveal that microscopic imaging outperforms investigated chip-on-the-tip solutions, i.e., at higher magnification. Nonetheless, custom-made cameras are suitable for application in microsurgery. Although reconstruction with the miniaturized endoscope is more inaccurate, it provides a good trade-off between accuracy, outer dimensions and accessibility to hard-to-reach surgical sites.