Skip to main content
main-content

Über dieses Buch

This book constitutes the thoroughly refereed post-conference proceedings of the International Workshop on Computational Challenges and Clinical Opportunities in Virtual Colonoscopy and Abdominal Imaging, held in conjunction with MICCAI 2010, in Beijing, China, on September 20, 2010.

The 19 revised full papers presented were carefully reviewed and selected from 26 submissions. The papers are organized in topical sections on CT colonography CAD, abdominal imaging, and virtual colonoscopy.

Inhaltsverzeichnis

Frontmatter

CT Colonography CAD

Electronic Cleansing in CT Colonography: Past, Present, and Future

Fecal tagging is a means of ‘marking’ fecal residues (stool and fluid) in a colon by use of the oral administration of a positive contrast agent (barium or iodine) in CT Colonography (CTC). Electronic cleansing (EC) is an emerging technique for removal of tagged fecal materials in fecal tagging CTC images after the image acquisition, and thus to avoid the physical bowel cleansing prior to CT scanning. In this syllabus, we present a brief overview about the past, present, and the future developments of EC technology in fecal tagging CTC.
Wenli Cai, Hiroyuki Yoshida

Improved Curvature Estimation for Shape Analysis in Computer-Aided Detection of Colonic Polyps

In current methods of computer-aided detection (CAD) of colonic polyps, curvature-based shape measures, like the shape index, curvedness, sphericity ratio, Gaussian curvature, mean curvature, etc., are widely used to analyze the local shapes in the colon wall. Therefore, the curvature estimation is an essential step, which is often conducted through kernel methods. However, spurious calculations indicating high curvature are frequently observed when the kernel contains two surfaces (this happens for objects like a thin slab, sphere, etc.). In this study, we adapted the Knutsson mapping method to solve this problem, so that we can improve the curvature estimation for CAD of colonic polyps in virtual colonoscopy.
Hongbin Zhu, Yi Fan, Zhengrong Liang

Characterizing Colonic Detections in CT Colonography Using Curvature-Based Feature Descriptor and Bag-of-Words Model

We present a method based on the content-based image retrieval (CBIR) paradigm to enhance the performance of computer aided detection (CAD) in computed tomographic colonography (CTC). The method explores curvature-based feature descriptors in conjunction with bag-of-words (BoW) models to characterize colonic detections. The diffusion distance is adopted to improve feature matching and clustering. Word selection is also applied to remove non-informative words. A representative database is constructed to categorize different types of detections. Query detections are compared with the database for classification. We evaluated the performance of the system by using digital phantoms of common structures in the colon as well as real CAD detections. The results demonstrated the potential of our technique for distinguishing common structures within the colon as well as for classifying true and false-positive CAD detections.
Javed M. Aman, Ronald M. Summers, Jianhua Yao

Haustral Fold Segmentation of CT Colonography Using Ridge Line Detection

In computed tomographic colonography, colonic/haustral folds often serve as important anatomic landmarks for various tasks, such as virtual endoscopic navigation, tenia coli extraction, prone/supine registration, and polyp matching. In this paper, we present an automatic fold segmentation method based on the negative ridge line detection method. Because haustral folds meet the normal colon wall at negative ridge lines, automatic fold segmentation can be achieved by taking the negative ridge lines as fold boundaries. Preliminary results on patient data are very promising.
Hongbin Zhu, Lihong Li, Yi Fan, Zhengrong Liang

Recent Advances in Reduction of False Positives in Computerized Detection of Polyps in CT Colonography

One of the major challenges in computer-aided detection (CADe) of polyps in CT colonography (CTC) is the reduction of false-positive detections (FPs) without a concomitant reduction in sensitivity. Major sources of FPs generated by CADe schemes include haustral folds, residual stool, rectal tubes, the ileocecal valve, and extra-colonic structures such as the small bowel and stomach. A large number of FPs is likely to confound the radiologist’s task of image interpretation, lower the radiologist’s efficiency, and cause radiologists to lose their confidence in CADe as a useful tool. Therefore, it is important to reduce the number of FPs as much as possible while maintaining a high sensitivity. In this paper, FP reduction techniques used in CADe schemes for detection of polyps in CTC are reviewed.
Kenji Suzuki

A Bayesian Approach for False Positive Reduction in CTC CAD

This paper presents an automated detection method for identifying colonic polyps and reducing false positives (FPs) in CT images. It formulates the problem of polyp detection as a probability calculation through a unified Bayesian statistical model. The polyp likelihood is modeled with a combination of shape and intensity features. A second principal curvature PDE provides a shape model; and the partial volume effect is considered in modeling of the polyp intensity distribution. The performance of the method was evaluated on a large multi-center dataset of colonic CT scans. Both qualitative and quantitative experimental results demonstrate the potential of the proposed method.
Xujiong Ye, Gareth Beddoe, Greg Slabaugh

False-Positive Reduction in Computer-Aided Detection of Polyps in CT Colonography: A Massive-Training Support Vector Regression Approach

A massive-training artificial neural network (MTANN) has been investigated for reduction of false positives (FPs) in computer-aided detection (CADe) of polyps in CT colonography (CTC). A major limitation of the MTANN is a long training time. To address this issue, we investigated the feasibility of a support vector regression (SVR) in the massive-training framework and developed a massive-training SVR (MTSVR). To test the proposed MTSVR, we compared it with the original MTANN in FP reduction in CADe of polyps in CTC. With MTSVR, we reduced the training time by a factor of 190, while achieving a performance (by-polyp sensitivity of 94.7% with 2.5 FPs/patient) comparable to that of the original MTANN (which has the same sensitivity with 2.6 FPs/patient).
Jian-Wu Xu, Kenji Suzuki

Learning to Detect 3D Rectal Tubes in CT Colonography Using a Global Shape Model

The rectal tube (RT) is a common source of false positives (FPs) in computer-aided detection (CAD) systems for CT colonography. In this paper, we present a novel and robust bottom-up approach to detect the RT. Probabilistic models, trained using kernel density estimation (KDE) on simple low-level features, are employed to rank and select the most likely RT tube candidate on each axial slice. Then, a shape model, robustly estimated using Random Sample Consensus (RANSAC), infers the global RT path from the selected local detections. Our method is validated using a diverse database, including data from five hospitals. The experiments demonstrate a high detection rate of the RT path, and when tested in a CAD system, reduce 20.3% of the FPs with no loss of CAD sensitivity.
Xiaoyun Yang, Gareth Beddoe, Greg Slabaugh

Abdominal Imaging

Estimation of Necrosis Volumes in Focal Liver Lesions Based on Multi-phase Hepatic CT Images

This study presents a computer-aided volumety (CAV) scheme that estimates the necrosis volumes in the focal liver lesions based on the multi-phase hepatic CT images for estimation of liver tumor burden. We developed a CAV scheme that consisted of the following three major steps: registration of multi-phase series based upon the portal-venous phase images, modeling of the concentration-time curve and thus estimattion of the arterial and portal-venous blood flow, and segmentation of the necrotic and tumorous tissues. Sixteen hepatocellular carcinoma cases were used for the evaluation of the CAV scheme. The total blood volume distribution of the liver tissue, tumor, and necrosis was computed in these datasets. The blood volumes for the liver tissue, tumor, and necrosis had 0.250±0.129, 0.171±0.073, and 0.054±0.045 ml/sec, respectively. The CAV scheme was shown to be potentially useful for efficient and accurate longitudinal measurement of liver tumor burdens in hepatic CT images.
June-Goo Lee, Wenli Cai, Anand Singh, Hiroyuki Yoshida

Detection of the Invasion of Bladder Tumor into Adjacent Wall Based on Textural Features Extracted from MRI Images

The invasion depth of a bladder tumor is of great importance for tumor staging and treatment planning. Considering that MRI bladder images could provide natural contrast between the urine and bladder wall, some texture features have been extracted from MRI images in our previous study, demonstrating a statistically significant difference between tumor tissues and wall tissues. In this study, a classification and labeling scheme has been proposed for the detection of the invasion depth of bladder tumors, based on these selected features, such as mean, standard deviation, uniformity, covariance, and contrast. Experimental results using patients’ MRI datasets show the feasibility of the proposed scheme for labeling of bladder tumors, indicating its potential for noninvasive detection of bladder tumors and their stage.
Zhide Wu, Zhengxing Shi, Guopeng Zhang, Hongbing Lu

Detecting Bladder Abnormalities Based on Inter-layer Intensity Curve for Virtual Cystoscopy

This paper presents a level set based method for bladder abnormality detection on T1-weighted MR images. First, the bladder wall is segmented by using a coupled level set framework, in which the inner and outer borders of the bladder wall are extracted by two level set functions. Then, the middle layer of the bladder wall is founded and represented by a new level set function. Finally, the new level set function divides the bladder wall into several layers. The inter-layer intensity of all voxels in each layer is sorted in ascending order to generate the inter-layer intensity curve. The results prove the effectiveness of inter-layer intensity curve in indicating the emerging of the bladder abnormalities.
Fanghua Liu, Chaijie Duan, Kehong Yuan, Zhengrong Liang, Shanglian Bao

Computer-Assisted Diagnosis for Quantitative Image-Based Analysis of Crohn’s Disease in CT Enterography

Crohn’s disease is an inflammatory disease that can cause a wide variety of symptoms and is increasing in prevalence. We developed a computer-assisted diagnosis (CADx) scheme for quantitative image-based analysis of Crohn’s disease in CT enterography (CTE). The CADx scheme extracts regions of interest automatically from CTE data, analyzes the small bowel automatically by use of mural features, and uses a support vector machine to predict the presence of active Crohn’s disease. For pilot evaluation, two radiologists diagnosed the CTE data of 54 patients with known or suspected Crohn’s disease. An unblinded gastroenterologist established the truth about the patients. The CADx scheme was then trained with the CTE data of 46 patients where the radiologists agreed on their diagnosis, and it was tested with the 8 difficult cases where the radiologists disagreed on their diagnosis. A bootstrapping analysis of the per-patient performance of the CADx scheme in predicting the presence of active Crohn’s disease yielded an area under receiver-operating characteristic (ROC) curve of 0.92±0.05. The result indicates that the CADx scheme could provide a useful decision-making tool for CTE.
Janne Näppi, June-Goo Lee, Joel G. Fletcher, Hiroyuki Yoshida

Computer-Aided Detection of Small Bowel Strictures for Emergency Radiology in CT Enterography

Computer-aided detection (CAD) of small bowel strictures can have significant impact in improving the workflow of CT enterography in an emergency setting where even non-expert radiologists could use it to rapidly detect sites of obstruction. A CAD scheme was developed to detect strictures from abdominal CT enterography data by use of multi-scale template matching and a blob detector. A pilot study was performed on 15 patients with 22 surgically confirmed strictures to study the effect of the CAD scheme on observer performance. The 77% sensitivity of an inexperienced radiologist assisted by the CAD scheme was comparable with the 81% sensitivity of an unaided expert radiologist (p=0.07). The use of CAD significantly reduced the reading time to identify strictures (p<0.0001). Most of the false-positive CAD detections were caused by collapsed bowel loops, approximated bowel wall, muscles or vessels, and they were easy to dismiss. The results indicate that CAD can provide radiologists with rapid and accurate interpretations of strictures to improve workflow in an emergency setting.
Nisha I. Sainani, Janne Näppi, Dushyant V. Sahani, Hiroyuki Yoshida

Virtual Colonoscopy

Teniae Coli Extraction in Human Colon for Computed Tomographic Colonography Images

Teniae coli are three bands of longitudinal smooth muscle on the surface of the colon, serving as anatomically meaningful landmarks for guiding virtual colonoscopic navigation and registration. This paper presents a novel method for teniae coli extraction for CT colonography. Because teniae coli are muscles running between haustral folds, they can be extracted by analysis of fold information. In our method, the 3D colon surface is first preprocessed into a 2D flattened colon. Then a 2D Gabor filter is employed to extract the feature of haustral folds, following by a Sobel operator to enhance the fold edge. The fold center is then detected by thresholding. A path of the fold can be obtained by connecting the fold center. Teniae coli are then extracted as lines in the middle of a pair of fold paths. Experiments were carried out on 5 cases, and the normalized RMSE was 5.01% with a 4.13% standard deviation.
Zhuoshi Wei, Jianhua Yao, Shijun Wang, Ronald M. Summers

Extraction of Landmarks and Features from Virtual Colon Models

The colon is a very complicated structure with a large number of distortions and bends. Landmarks and features serve as tools to guide colon flattening and to assist in the study of the colon surface segment by segment. Identification of feature points and landmarks is also useful for the registration of colon surfaces. In this paper, we present methods for identifying the locations of the taeniae coli and the four major flexures which form the prominent anatomic landmarks on the colon surface. The colon surface is cut open along these landmarks, and the segments obtained can be used for study of the surface of the colon. We define new feature points on the flattened colon surfaces and use well-established graph-based algorithms for their detection. We demonstrate the results showing the extracted landmarks and the detected features.
Krishna Chaitanya Gurijala, Arie Kaufman, Wei Zeng, Xianfeng Gu

Conformal Geometry Based Supine and Prone Colon Registration

In virtual colonoscopy, CT scans are typically acquired with the patient in both supine and prone positions. The registration of these two scans is desirable so that the physician can clarify situations or confirm polyp findings at a location in one scan with the same location in the other, thereby improving polyp detection rates and reducing false positives. However, this supine-prone registration is challenging because of the substantial distortions in the colon shape due to the patient’s position shifting. We present an efficient algorithm and framework for performing this registration through the use of conformal geometry to guarantee the registration is a diffeomorphism. The colon surface is conformally flattened to a rectangle using holomorphic differentials. The flattened domains of supine and prone are aligned by the harmonic map with feature correspondence constraints. We demonstrate the efficiency and efficacy of our method by measuring the distance between features on the registered colons.
Wei Zeng, Joseph Marino, Xianfeng Gu, Arie Kaufman

Colon Visualization Using Shape Preserving Flattening

Virtual colonoscopy is a well-known screening modality for colon cancer, and virtual colon flattening techniques have been proposed to map the 3D surface to the 2D domain. Performing this flattening with conformal geometry allows for the local shapes to be preserved. We explore here how these shape preserving flattened maps can be used for visualization of the colon and to enhance the VC environment. Our discussion focuses on two types of use of the flattened colon, using the flattened mesh alone and using it with integration into the 3D endoluminal view. For the mesh alone, flattened views can be generated using volume rendering to attain the same image quality present in a typical view. When integrated with the 3D endoluminal view, the flattened mesh can be used to assist in navigation through a colon, or in acquiring corresponding view points in two scans if there is a one-to-one and onto mapping between two flattened meshes.
Joseph Marino, Arie Kaufman

Synchronized Display of Virtual Colonoscopic Views in Supine and Prone CT Images

Colon diagnosis using CT images of supine and prone positions is time-consuming task because a radiologist has to control viewing fields of virtual colonoscopic (VC) views individually. We propose a method to display VC views of corresponding areas of the colons in the two positions. Viewpoints of VC views in the two positions are obtained by a point interpolation based on the result of the supine-prone correspondence finding. Up-directions of VC views defined based on body orientations of a patient. We generated synchronized VC views of the two positions using the proposed method. Positional difference of the viewpoints between the two positions was 5.03 [mm]. The method can synchronize viewpoints of the supine and prone positions without significant error.
Masahiro Oda, Eiichiro Fukano, Takayuki Kitasaka, Hirotsugu Takabatake, Masaki Mori, Hiroshi Natori, Shigeru Nawano, Kensaku Mori

Colorectal Polyp Segmentation Based on Geodesic Active Contours with a Shape-Prior Model

Automated polyp segmentation is important both in measuring polyp size and in improving polyp detection performance in CTC. We present a polyp segmentation method that is based on the combination of geodesic active contours and a shape-prior model of polyps. To train the shape model, polyps identified by radiologists are grouped by morphologic characteristics. Each group of polyps is used for building a shape-prior model. Then the geodesic active contours method is employed to segment polyps constrained by this shape-prior model. This method can reliably segment polyp boundaries even where the image contrast is not sufficient to define a boundary between a polyp and its surrounding colon tissue. As a pilot study, we developed one polyp shape-prior model for sessile polyps that are located on a relatively flat colon wall. We use the model to segment similar polyps, and the results are evaluated visually.
Haiyong Xu, H. Donald Gage, Pete Santago, Yaorong Ge

Backmatter

Weitere Informationen

Premium Partner

BranchenIndex Online

Die B2B-Firmensuche für Industrie und Wirtschaft: Kostenfrei in Firmenprofilen nach Lieferanten, Herstellern, Dienstleistern und Händlern recherchieren.

Whitepaper

- ANZEIGE -

Best Practices für die Mitarbeiter-Partizipation in der Produktentwicklung

Unternehmen haben das Innovationspotenzial der eigenen Mitarbeiter auch außerhalb der F&E-Abteilung erkannt. Viele Initiativen zur Partizipation scheitern in der Praxis jedoch häufig. Lesen Sie hier  - basierend auf einer qualitativ-explorativen Expertenstudie - mehr über die wesentlichen Problemfelder der mitarbeiterzentrierten Produktentwicklung und profitieren Sie von konkreten Handlungsempfehlungen aus der Praxis.
Jetzt gratis downloaden!

Bildnachweise