Landmarks In Images Research Articles

Image Subset Selection Using Gabor Filters and Neural Networks

An automatic method for the selection of subsets of images, both modern and historic, out of a set of landmark large images collected from the Internet is presented in this paper. This selection depends on the extraction of dominant features using Gabor filtering. Features are selected carefully from a preliminary image set and fed into a neural network as a training data. The method collects a large set of raw landmark images containing modern and historic landmark images and non-landmark images. The method then processes these images to classify them as landmark and non-landmark images. The classification performance highly depends on the number of candidate features of the landmark.

Agreement of angle closure assessments between gonioscopy, anterior segment optical coherence tomography and spectral domain optical coherence tomography.

To determine angle closure agreements between gonioscopy and anterior segment optical coherence tomography (AS-OCT), as well as gonioscopy and spectral domain OCT (SD-OCT). A secondary objective was to quantify inter-observer agreements of AS-OCT and SD-OCT assessments. Seventeen consecutive subjects (33 eyes) were recruited from the study hospital's Glaucoma clinic. Gonioscopy was performed by a glaucomatologist masked to OCT results. OCT images were read independently by 2 other glaucomatologists masked to gonioscopy findings as well as each other's analyses of OCT images. Totally 84.8% and 45.5% of scleral spurs were visualized in AS-OCT and SD-OCT images respectively (P<0.01). The agreement for angle closure between AS-OCT and gonioscopy was fair at k=0.31 (95% confidence interval, CI: 0.03-0.59) and k=0.35 (95% CI: 0.07-0.63) for reader 1 and 2 respectively. The agreement for angle closure between SD-OCT and gonioscopy was fair at k=0.21 (95% CI: 0.07-0.49) and slight at k=0.17 (95% CI: 0.08-0.42) for reader 1 and 2 respectively. The inter-reader agreement for angle closure in AS-OCT images was moderate at 0.51 (95% CI: 0.13-0.88). The inter-reader agreement for angle closure in SD-OCT images was slight at 0.18 (95% CI: 0.08-0.45). Significant proportion of scleral spurs were not visualised with SD-OCT imaging resulting in weaker inter-reader agreements. Identifying other angle landmarks in SD-OCT images will allow more consistent angle closure assessments. Gonioscopy and OCT imaging do not always agree in angle closure assessments but have their own advantages, and should be used together and not exclusively.

Guest editorial: selected papers from ICIMCS 2012

These papers cover key issues in Internet multimedia computing and services, including large-scale visual search, mobile media processing, Internet image tagging, video transmission, as well as fundamental technologies for intelligent multimedia computing, such as image segmentation, logo recognition, and object tracking. The first article “Visual word expansion and BSIFT verification for large-scale image search” by Zhou et al. presents a visual word expansion method improving the recall of retrieval on large-scale image collections and a binary SIFT verification method boosting the retrieval precision. The former represents a query image as the aggregation of visual words of local features in the query and the neighboring words of these query words. The latter verifies feature matching between images based on the binarized SIFT feature. The work is experimented on two public image data sets. The second article “Accurate sensing of scene geo-context via mobile visual localization” by Liu et al. proposes a new mobile image geo-tagging approach toward accurate location tagging for images captured in urban areas. For a landmark image, the approach infers comprehensive and accurate geo-context of the image, including camera location, view direction, and scene location, by exploiting large-scale image retrieval and 3D reconstruction techniques. Experimental results on the San Francisco street view data set, which consists of approximately 150 k panoramic images in San Francisco at about 4-m intervals, have demonstrated the effectiveness of the approach. The third article entitled “Tag ranking based on salient region graph propagation” by Tang et al. presents a new tag ranking solution to rank user-contributed tags associated with Internet images. The solution exploits salient regions in images and ranks tags by integrating visual clues from entire images and salient regions of images. It constructs two sparse graphs over images and salient sub-images, International Conference on Internet Multimedia Computing and Services (ICIMCS) is an annual conference sponsored by ACM SIGMM China Chapter. The conference is especially interested in the latest technologies and applications that deal with the Web-scale processing and management of heterogeneous data from the Internet for multimedia computing and service. ICIMCS 2012 was held in Wuhan, China. The conference attracted around 90 participants, including researchers from academia and industries across ten countries/regions, for sharing their recent works on the topics ranging from visual feature representation and analysis to visual recognition and classification, and from social and mobile media analysis to multimedia services. This special issue comprises the extended versions of seven papers, including three best papers and four papers from the regular and special sessions of ICIMCS 2012.

Classification and indexing scheme of large-scale image repository for spatio-temporal landmark recognition

In this paper, we propose a classification and indexing scheme of large-scale image repository for spatio-temporal landmark recognition using the local features, GPS data and user tags of images. For spatio-temporal landmark image classification, we first divide Earth's entire surface into unit grid cells and collect pictures taken in each cell through Flickr. The collected images contain information such as location, titles and other user tags. Usually, the titles or user tags of landmark images include landmark names. Hence, by analyzing such tags, we can identify promising landmark names in the region and create a collection of images for each landmark using Flickr API. Even though each landmark class contains images of the same landmark, their spatio-temporal features could be different depending on shooting time, distance or angle. Therefore, we further divide the images in each landmark class into several subclasses according to their spatio-temporal characteristics using their color and local features. Especially, we detect the interest points of the images in the class, construct their feature descriptors using SURF and perform statistical analysis to select their representative points. Similar representative points are merged for fast comparison. Finally, we construct an index on the representative points using k-d tree. To identify the landmark in a user query image, we extract its SURF features and search for them in the index. Most similar matches are returned, along with descriptive text and GPS information. We implemented a prototype system based on a client-server architecture and performed various experiments to demonstrate that our scheme can achieve reasonable precision and scalability and provide a new browsing experience to the user.

Content-Based Visual Landmark Search via Multimodal Hypergraph Learning.

While content-based landmark image search has recently received a lot of attention and became a very active domain, it still remains a challenging problem. Among the various reasons, high diverse visual content is the most significant one. It is common that for the same landmark, images with a wide range of visual appearances can be found from different sources and different landmarks may share very similar sets of images. As a consequence, it is very hard to accurately estimate the similarities between the landmarks purely based on single type of visual feature. Moreover, the relationships between landmark images can be very complex and how to develop an effective modeling scheme to characterize the associations still remains an open question. Motivated by these concerns, we propose multimodal hypergraph (MMHG) to characterize the complex associations between landmark images. In MMHG, images are modeled as independent vertices and hyperedges contain several vertices corresponding to particular views. Multiple hypergraphs are firstly constructed independently based on different visual modalities to describe the hidden high-order relations from different aspects. Then, they are integrated together to involve discriminative information from heterogeneous sources. We also propose a novel content-based visual landmark search system based on MMHG to facilitate effective search. Distinguished from the existing approaches, we design a unified computational module to support query-specific combination weight learning. An extensive experiment study on a large-scale test collection demonstrates the effectiveness of our scheme over state-of-the-art approaches.

The influence of injecting an epidural contrast agent into the sacral canal on the fluoroscopic visibility of bony landmarks for sacroiliac screw fixation: a feasibility study.

In sacroiliac screw fixation of unstable pelvic injuries in geriatric patients, poor bone quality often obscures important bony landmarks in fluoroscopic images. The authors analyzed the feasibility of injecting a transhiatal contrast agent (CA) into the sacral canal to improve fluoroscopic visualization in the sacral epidural space. Eight fresh cadaveric whole-body specimens from human donors whose mean age at the time of death was 78 years (range 69-87 years) were used. First, to identify bony landmarks without CA enhancement, the authors acquired fluoroscopy images of the native sacral canal, using lateral, inlet, and outlet projections. Through puncture of the sacral hiatus, 8-10 ml of CA was injected into the epidural space. Fluoroscopy images were then acquired in the standard pelvic views to identify the bony landmarks. To assess the effect of the CA enhancement, visibility of the landmarks was assessed before and after CA injection. Each identified landmark was scored as 1, and summative landmark scores of up to 10 were determined for each specimen. The cadaveric specimens were representative of bone structures in the geriatric population. In all specimens, epidural CA injection enhanced the fluoroscopic visualization of the sacral canal and of the S-1 foramina. The enhancement increased the total bony landmark score from 5.9 (range 4-8) without CA injection to 8.1 (range 6-10) after CA injection. Considering only intrasacral landmarks, the score was increased from 1.5 to 3. Injection of a transhiatal epidural CA improves fluoroscopic imaging of the sacral canal and of the neural foramina. Hence, this technique could be applied to help the surgeon identify anatomical landmarks during sacroiliac screw fixation in geriatric patients.

Optic disc segmentation using the sliding band filter

Background: The optic disc (OD) centre and boundary are important landmarks in retinal images and are essential for automating the calculation of health biomarkers related with some prevalent systemic disorders, such as diabetes, hypertension, cerebrovascular and cardiovascular diseases.Methods: This paper presents an automatic approach for OD segmentation using a multiresolution sliding band filter (SBF). After the preprocessing phase, a low-resolution SBF is applied on a downsampled retinal image and the locations of maximal filter response are used for focusing the analysis on a reduced region of interest (ROI). A high-resolution SBF is applied to obtain a set of pixels associated with the maximum response of the SBF, giving a coarse estimation of the OD boundary, which is regularized using a smoothing algorithm.Results: Our results are compared with manually extracted boundaries from public databases (ONHSD, MESSIDOR and INSPIRE-AVR datasets) outperforming recent approaches for OD segmentation. For the ONHSD, 44% of the results are classified as Excellent, while the remaining images are distributed between the Good (47%) and Fair (9%) categories. An average overlapping area of 83%, 89% and 85% is achieved for the images in ONHSD, MESSIDOR and INSPIR-AVR datasets, respectively, when comparing with the manually delineated OD regions.Discussion: The evaluation results on the images of three datasets demonstrate the better performance of the proposed method compared to recently published OD segmentation approaches and prove the independence of this method when from changes in image characteristics such as size, quality and camera field of view.

Pose Estimation of Web-Shared Landscape Pictures

Abstract. We propose a robust method for registering high oblique images of landscapes. Typically, an input image can be registered by matching it against a set of registered images of the same location. While this has been shown to work very well for images of popular urban landmarks, registering landscape images remains a very challenging task: For a given place, only a very small amount of registered images is generally already available on photo-sharing platforms. Moreover, the appearance of landscapes can vary drastically depending on the season and the weather conditions. For these two reasons, matching the input images with registered images in a reliable way remains a challenging task. Our contribution is two-fold: first, we show how to estimate the camera orientation for images with GPS data using a novel algorithm for horizon matching based on Dynamic Time Warping. The proposed algorithm exploits an elevation model. Each image is processed independently from the others, there is therefore no need neither for image matching or for a large set of images. This step provides a set of reliable, fully registered images. Second, and in odrer to register new images with no GPS data available, we first ask the user to provide an approximate image localization on a 2D map. Then, we exploit this prior on the camera location to efficiently and robustly constrain and guide the matching process used to register the query image. We apply our method to a case study from the Zermatt area in Southern Switzerland, and show that the method provides registrations, which are accurate enough to map each pixel to an aerial map.

Bony Landmarks of the Anterior Cruciate Ligament Tibial Footprint

Background: Although the importance of tibial tunnel position for achieving stability after anterior cruciate ligament (ACL) reconstruction was recently recognized, there are fewer detailed reports of the anatomy of the tibial topographic footprint compared with the femoral side. Hypothesis: The ACL tibial footprint has a relationship to bony prominences and surrounding bony landmarks. Study Design: Descriptive laboratory study. Methods: This study consisted of 2 anatomic procedures for the identification of bony prominences that correspond to the ACL tibial footprint and 3 surrounding landmarks: the anterior ridge, lateral groove, and intertubercular fossa. In the first procedure, after computed tomography (CT) was performed on 12 paired, embalmed cadaveric knees, 12 knees were visually observed, while their contralateral knees were histologically observed. Comparisons were made between macroscopic and microscopic findings and 3-dimensional (3D) CT images of these bony landmarks. In the second procedure, the shape of the bony prominence and incidence of their bony landmarks were evaluated from the preoperative CT data of 60 knee joints. Results: In the first procedure, we were able to confirm a bony prominence and all 3 surrounding landmarks by CT in all cases. Visual evaluation confirmed a small bony eminence at the anterior boundary of the ACL. The lateral groove was not confirmed macroscopically. The ACL was not attached to the lateral intercondylar tubercle, ACL tibial ridge, and intertubercular space at the posterior boundary. Histological evaluation confirmed that the anterior ridge and lateral groove were positioned at the anterior and lateral boundaries, respectively. There was no ligament tissue on the intercondylar space corresponding to the intercondylar fossa. In the second investigation, the bony prominence showed 2 morphological patterns: an oval type (58.3%) and a triangular type (41.6%). The 3 bony landmarks, including the anterior ridge, lateral groove, and intertubercular fossa, existed in 96.6%, 100.0%, and 96.6% of the cases, respectively. Conclusion: There is a bony prominence corresponding to the ACL footprint and bony landmarks on the anterior, posterior, and lateral boundaries. Clinical Relevance: The study results may help create an accurate and reproducible tunnel, which is essential for successful ACL reconstruction surgery.

Mobile Landmark Search with 3D Models

Landmark search is crucial to improve the quality of travel experience. Smart phones make it possible to search landmarks anytime and anywhere. Most of the existing work computes image features on smart phones locally after taking a landmark image. Compared with sending original image to the remote server, sending computed features saves network bandwidth and consequently makes sending process fast. However, this scheme would be restricted by the limitations of phone battery power and computational ability. In this paper, we propose to send compressed (low resolution) images to remote server instead of computing image features locally for landmark recognition and search. To this end, a robust 3D model based method is proposed to recognize query images with corresponding landmarks. Using the proposed method, images with low resolution can be recognized accurately, even though images only contain a small part of the landmark or are taken under various conditions of lighting, zoom, occlusions and different viewpoints. In order to provide an attractive landmark search result, a 3D texture model is generated to respond to a landmark query. The proposed search approach, which opens up a new direction, starts from a 2D compressed image query input and ends with a 3D model search result.

Left atrial appendage: anatomy and imaging landmarks pertinent to percutaneous transcatheter occlusion

Percutaneous left atrial appendage (LAA) closure represents a complementary option and effective treatment for patients at risk of thromboembolism, especially in patients for whom it may be difficult to achieve...

Image-guided microbeam irradiation to brain tumour bearing mice using a carbon nanotube x-ray source array

Microbeam radiation therapy (MRT) is a promising experimental and preclinical radiotherapy method for cancer treatment. Synchrotron based MRT experiments have shown that spatially fractionated microbeam radiation has the unique capability of preferentially eradicating tumour cells while sparing normal tissue in brain tumour bearing animal models. We recently demonstrated the feasibility of generating orthovoltage microbeam radiation with an adjustable microbeam width using a carbon nanotube based x-ray source array. Here we report the preliminary results from our efforts in developing an image guidance procedure for the targeted delivery of the narrow microbeams to the small tumour region in the mouse brain. Magnetic resonance imaging was used for tumour identification, and on-board x-ray radiography was used for imaging of landmarks without contrast agents. The two images were aligned using 2D rigid body image registration to determine the relative position of the tumour with respect to a landmark. The targeting accuracy and consistency were evaluated by first irradiating a group of mice inoculated with U87 human glioma brain tumours using the present protocol and then determining the locations of the microbeam radiation tracks using γ-H2AX immunofluorescence staining. The histology results showed that among 14 mice irradiated, 11 received the prescribed number of microbeams on the targeted tumour, with an average localization accuracy of 454 µm measured directly from the histology (537 µm if measured from the registered histological images). Two mice received one of the three prescribed microbeams on the tumour site. One mouse was excluded from the analysis due to tissue staining errors.

Age Estimation Based on CLM, Tree Mixture With Adaptive Neuron Fuzzy, Fuzzy Sum

As you know, age diagnosis based on the image is one of the most attractive topics in computer .In this paper, we present a intelligent model to estimate the age of face image. We use shape and texture feature extraction from FG-NET landmark image data set using AAM(Active Appearance Model), CLM (Constrained Local Model), tree Mixture algorithms. Finally, the obtained features were given as the training data to the ANFIS (adaptive neuro fuzzy influence system), FSVM(Fuzzy Support Vector Machine). Our experimental results show that In our proposed system, fuzzy svm has less errors and system worked more accurate and appropriative than prior methods. Our system is able to identify age of face image from different directions as is.

Robust anatomical landmark detection for MR brain image registration.

Correspondence matching between MR brain images is often challenging due to large inter-subject structural variability. In this paper, we propose a novel landmark detection method for robust establishment of correspondences between subjects. Specifically, we first annotate distinctive landmarks in the training images. Then, we use regression forest to simultaneously learn (1) the optimal set of features to best characterize each landmark and (2) the non-linear mappings from local patch appearances of image points to their displacements towards each landmark. The learned regression forests are used as landmark detectors to predict the locations of these landmarks in new images. Since landmark detection is performed in the entire image domain, our method can cope with large anatomical variations among subjects. We evaluated our method by applying it to MR brain image registration. Experimental results indicate that by combining our method with existing registration method, obvious improvement in registration accuracy can be achieved.

Post-treatment appearances, pitfalls, and patterns of failure in head and neck cancer on FDG PET/CT imaging.

Majority of patients with head and neck cancer are treated with combined treatment regimes such as surgery, radiation therapy, and chemotherapy. The loss of structural symmetry and imaging landmarks as a result of therapy makes post-treatment imaging a daunting task on conventional modalities like computed tomography (CT) scan and magnetic resonance imaging (MRI) as well as on 18 Fluorine-Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography (18F FDG PET/CT). Combined multimodality treatment approach causes various tissue changes that give rise to a spectrum of findings on FDG PET/CT imaging, which are depicted in this atlas along with a few commonly encountered imaging pitfalls. The incremental value of FDG PET/CT in detecting locoregional recurrences in the neck as well as distant failures has also been demonstrated.

Distribution of Abdominal Aortic Calcium by Computed Tomography: Impact of Analysis Method on Quantitative Calcium Score

Abdominal aortic calcification (AAC) can be quantified using computed tomography (CT), but imaging planesare prescribed based on bony landmarks, so that individual variation between the landmark and the aortoiliac junction can result in variable aortic coverage. In the Framingham CT substudy, we scanned a 15-cm (Z-direction) abdominal segment cranial to the S1 vertebral body. We sought to determine the range and distribution of length of aorta scanned and the distribution of AAC within the abdominal aorta and to compare burden of AAC measured from fixed-length segments versus AAC from all slices cranial to the aortoiliac bifurcation. AAC was quantified by modified Agatston score (AS) in 100 Framingham Heart Study participants (60±13years old, 51 men). We compared the AS measured from 5-cm and 8-cm segments with the ASALL (total visualized aorta). Of 100, 73 participants had AAC >0. The total length of aorta imaged was ≥8cm in 84% of participants. Qualitatively, 5-cm and 8-cm segments correctly identified 96% and 99%, respectively, of participants as having or not having AAC. Quantitatively, AS8cm was within 20% of ASALL in four-fifths and within 30% of ASALL in nine-tenths of participants. AS5cm more severely underestimated ASALL. The use of S1 as the caudal imaging landmark in a 15-cm slab yields ≥8cm aortic coverage in most adults. Both 5-cm and 8-cm analysis strategies are comparable to analyzing the total visualized abdominal aorta for prevalent AAC, but only 8-cm segment analysis yields quantitatively similar measures of AAC.

Context-aware Discriminative Vocabulary Tree Learning for mobile landmark recognition

Recently, mobile landmark recognition has become one of the emerging applications in mobile media, offering landmark information and e-commerce opportunities to both mobile users and business owners. Existing mobile landmark recognition techniques mainly use GPS (Global Positioning System) location information to obtain a shortlist of database landmark images nearby the query image, followed by visual content analysis within the shortlist. This is insufficient since (i) GPS data often has large errors in dense build-up areas, and (ii) direction data that can be acquired from mobile devices is underutilized to further improve recognition. In this paper, we propose to integrate content and context in an effective and efficient vocabulary tree framework. Specifically, visual content and two types of mobile context: location and direction, can be integrated by the proposed Context-aware Discriminative Vocabulary Tree Learning (CDVTL) algorithm. The experimental results show that the proposed mobile landmark recognition method outperforms the state-of-the-art methods by about 6%, 21% and 13% on NTU Landmark-50, PKU Landmark-198 and the large-scale San Francisco landmark dataset, respectively.

Automatic registration of multi-modal microscopy images for integrative analysis of prostate tissue sections

BackgroundProstate cancer is one of the leading causes of cancer related deaths. For diagnosis, predicting the outcome of the disease, and for assessing potential new biomarkers, pathologists and researchers routinely analyze histological samples. Morphological and molecular information may be integrated by aligning microscopic histological images in a multiplex fashion. This process is usually time-consuming and results in intra- and inter-user variability. The aim of this study is to investigate the feasibility of using modern image analysis methods for automated alignment of microscopic images from differently stained adjacent paraffin sections from prostatic tissue specimens.MethodsTissue samples, obtained from biopsy or radical prostatectomy, were sectioned and stained with either hematoxylin & eosin (H&E), immunohistochemistry for p63 and AMACR or Time Resolved Fluorescence (TRF) for androgen receptor (AR).Image pairs were aligned allowing for translation, rotation and scaling. The registration was performed automatically by first detecting landmarks in both images, using the scale invariant image transform (SIFT), followed by the well-known RANSAC protocol for finding point correspondences and finally aligned by Procrustes fit. The Registration results were evaluated using both visual and quantitative criteria as defined in the text.ResultsThree experiments were carried out. First, images of consecutive tissue sections stained with H&E and p63/AMACR were successfully aligned in 85 of 88 cases (96.6%). The failures occurred in 3 out of 13 cores with highly aggressive cancer (Gleason score ≥ 8). Second, TRF and H&E image pairs were aligned correctly in 103 out of 106 cases (97%).The third experiment considered the alignment of image pairs with the same staining (H&E) coming from a stack of 4 sections. The success rate for alignment dropped from 93.8% in adjacent sections to 22% for sections furthest away.ConclusionsThe proposed method is both reliable and fast and therefore well suited for automatic segmentation and analysis of specific areas of interest, combining morphological information with protein expression data from three consecutive tissue sections. Finally, the performance of the algorithm seems to be largely unaffected by the Gleason grade of the prostate tissue samples examined, at least up to Gleason score 7.

Radiothérapie guidée par l’image et partage des tâches avec les manipulateurs en électroradiologie : expérience de Clermont-Ferrand

The various image-guided radiotherapy techniques raise the question of how to achieve the control of patient positioning before irradiation session and sharing of tasks between radiation oncologists and radiotherapy technicians. We have put in place procedures and operating methods to make a partial delegation of tasks to radiotherapy technicians and secure the process in three situations: control by orthogonal kV imaging (kV-kV) of bony landmarks, control by kV-kV imaging of intraprostatic fiducial goldmarkers and control by cone beam CT (CBCT) imaging for prostate cancer. Significant medical overtime is required to control these three IGRT techniques. Because of their competence in imaging, these daily controls can be delegated to radiotherapy technicians. However, to secure the process, initial training and regular evaluation are essential. The analysis of the comparison of the use of kV/kV on bone structures allowed us to achieve a partial delegation of control to radiotherapy technicians. Controlling the positioning of the prostate through the use and automatic registration of fiducial goldmarkers allows better tracking of the prostate and can be easily delegated to radiotherapy technicians. The analysis of the use of daily cone beam CT for patients treated with intensity modulated irradiation is underway, and a comparison of practices between radiotherapy technicians and radiation oncologists is ongoing to know if a partial delegation of this control is possible.

Landmark Image Super-Resolution by Retrieving Web Images

This paper proposes a new super-resolution (SR) scheme for landmark images by retrieving correlated web images. Using correlated web images significantly improves the exemplar-based SR. Given a low-resolution (LR) image, we extract local descriptors from its up-sampled version and bundle the descriptors according to their spatial relationship to retrieve correlated high-resolution (HR) images from the web. Though similar in content, the retrieved images are usually taken with different illumination, focal lengths, and shot perspectives, resulting in uncertainty for the HR detail approximation. To solve this problem, we first propose aligning these images to the up-sampled LR image through a global registration, which identifies the corresponding regions in these images and reduces the mismatching. Second, we propose a structure-aware matching criterion and adaptive block sizes to improve the mapping accuracy between LR and HR patches. Finally, these matched HR patches are blended together by solving an energy minimization problem to recover the desired HR image. Experimental results demonstrate that our SR scheme achieves significant improvement compared with four state-of-the-art schemes in terms of both subjective and objective qualities.