Automatic Contour Detection Research Articles

Quantitative measurements in IVUS images.

IntraVascular UltraSound (IVUS) is a catheter-based technique which provides real-time high resolution tomographic images of both the lumen and arterial wall of a coronary segment, this in contrast to X-ray arteriography that provides a shadow image (luminogram) of the entire lumen. Nowadays the lumen and vessel parameters are measured manually, which is very time consuming and suffers from high inter- and intra-obser variability. With the continuing improvement in IVUS imaging, it is now feasible to develop and clinically apply automated methods of three-dimensional quantitative analysis of the coronary vessel morphology in an objective and reproducible way with automated contour detection techniques (QCU). Quantification, in 2D and 3D, as well as volume rendering for visualization of the IVUS images requires segmentation of the images (contour detection). The 3D contour detection system described in this article is based on the combination of contour detection in the transversal and sagital view. This article provides some of the basic principles of IVUS, the IVUS image quantification, the three-dimensional reconstruction and the contour detection and quantification in three-dimensional IVUS images.

10 Transoesophageal echocardiography: emerging and innovative technologies

Advances in transoesophageal echocardiography (TOE) have made this modality a powerful diagnostic tool in the critical care unit. TOE in combination with stress (atrial pacing, pharmacological stress) provides important information about myocardial ischaemia and myocardial viability. Recent progress in transducer technology and computerized image processing brings three-dimensional and contrast-enhanced TOE closer to the critical care unit. Finally, miniaturization of probes and automated contour detection techniques make TOE a sensitive and continuous monitor of global and segmental left ventricular function.

New developments in intravascular ultrasound imaging

IntraVascular Ultrasound Imaging (IVUS) has already been proposed in the early days of diagnostic ultrasound. Today, it has come under further full attention as a result of minimal invasive techniques. Not only excellent intravascular two-dimensional (2D) images are presently obtained, also three-dimensional (3D) reconstructed images show their diagnostic value. Based on 3D information, quantitative data such as plaque volume can be calculated. The procedure includes automatic contour detection based on image segmentation methods and greatly speeds up clinical evaluation. With the use of additional X-ray information, the true tortuous vessel geometry can be reconstructed in 3D. This allows, by numerical modelling techniques, to calculate endothelial shear stress values, which in turn may indicate sites prone to stenosis. With a decorrelation technique for radiofrequency (RF) echo information from sequential data in the same beam direction and integration method over the entire cross section, blood velocity can be shown colour-coded during the cardiac cycle, while even blood flow quantification seems to be possible. In vitro as well as in vivo experiments have shown the feasibility of the method. Intravascular imaging can be used to study the biomechanical properties of atheroma components. Local radial strain, used as a measure of local tissue hardness, can be estimated to identify hard or soft plaques independently of the echogenicity contrast between plaque and vessel wall.

Intravascular imaging

Based on three-dimensional (3D) information, quantitative data such as plaque volume can be calculated. The procedure includes automatic contour detection based on image segmentation methods and greatly speeds up clinical evaluation. With the use of additional X-ray information, the true tortuous vessel geometry can be reconstructed in 3D. This allows, by numerical modelling techniques, to calculate endothelial shear stress values which in turn may indicate sites prone to stenosis. With a decorrelation technique for radio frequency (RF) echo information from sequential data in the same beam direction and integration method over the entire cross section, blood velocity can be shown colour-coded during the cardiac cycle, while even blood flow quantification seems to be possible. In vitro as well as animal experiments have shown the feasibility of the method. Intravascular imaging can be used to study the biomechanical properties of atheroma components. Local radial strain as a measure of local tissue hardness can be estimated in principle. Hard or soft plaques can be identified from the strain images independently of the echogenic contrast between plaque and vessel wall.

３次元映像情報メディア技術 ステレオ画像の中間視点画像生成のためのエッジ情報を用いた視差推定

In order to synthesize intermediate-view images of arbitrary viewpoints between a pair of left and right view images, we propose two methods for disparity estimation with discontinuity at object contours. Both methods have four steps : (1) Two initial disparity maps based on left and right images are estimated with block matching. (2) Ill-corresponding areas in both of the initial disparity maps are detected using 'confidence evaluation'. The object contours are, then, detected by (i) Snakes method or (ii) Automatic Contour Detection (ACD) method. (3) The disparity maps are reestimated with discontinuity at the object contours. (4) The reestimated disparity maps are used to synthesize two intermediate-view images based on both view images and these images are integrated into the final intermediate-view image. Our methods can estimate the disparity of occluded areas with discontinuity at the object contours. Experimental results show our methods yield better quality intermediate-view images near the object contours than do ones using the initial disparities.

Evaluation of an automatic intraluminal edge detection technique for intravascular ultrasound images.

Intravascular ultrasound (IVUS) imaging enables detailed analysis and precise measurements of vascular cross-sections. However, to achieve a reduction in the existing level of observer variability requires the development of quantitative IVUS. We have developed a fully automatic intraluminal edge detection technique, based on adaptive active contour models and called ADDER (adaptive damping dependent on echographic regions) that allows the quantitation of the intraluminal cross-sectional area (ICSA). Using a 30-MHz mechanically rotated transducer mounted at the tip of a 3.5-F catheter, 58 normal and pathologic arterial segments (from coronary, renal, splenic, iliac, and carotid arteries) were imaged in vitro. These images were analyzed by 2 experts, E1 and E2, who manually traced the intraluminal contour twice for each image, as well as with ADDER. Intra-observer variabilities for ICSAs were found to be excellent (-1.454 +/- 3.51% for E1, 0.96 +/- 5.4% for E2). The inter-observer variability was 2.1 +/- 4.3%. The success factor for ADDER was 89%. Its intra-observer variability was null, as the method always finds a unique contour. The correlation between the automatically detected ICSA and the manual ICSA was: r = 0.99 (y = 1.03x + 0.89 mm2). Morphometric variations between manually and automatically traced contours, analyzed by the centerline method, were 100 +/- 140 mm on average. In conclusion, the ADDER automatic contour detection applied to IVUS images is robust and characterized by small systematic and random errors; therefore, quantitative IVUS is a useful tool in clinical research trials.

ECG-Gated Three-dimensional Intravascular Ultrasound

Automated systems for the quantitative analysis of three-dimensional (3D) sets of intravascular ultrasound (IVUS) images have been developed to reduce the time required to perform volumetric analyses; however, 3D image reconstruction by these nongated systems is frequently hampered by cyclic artifacts. We used an ECG-gated 3D IVUS image acquisition workstation and a dedicated pullback device in atherosclerotic coronary segments of 30 patients to evaluate (1) the feasibility of this approach of image acquisition, (2) the reproducibility of an automated contour detection algorithm in measuring lumen, external elastic membrane, and plaque+media cross-sectional areas (CSAs) and volumes and the cross-sectional and volumetric plaque+media burden, and (3) the agreement between the automated area measurements and the results of manual tracing. The gated image acquisition took 3.9+/-1.5 minutes. The length of the segments analyzed was 9.6 to 40.0 mm, with 2.3+/-1.5 side branches per segment. The minimum lumen CSA measured 6.4+/-1.7 mm2, and the maximum and average CSA plaque+media burden measured 60.5+/-10.2% and 46.5+/-9.9%, respectively. The automated contour-detection required 34.3+/-7.3 minutes per segment. The differences between these measurements and manual tracing did not exceed 1.6% (SD<6.8%). Intraobserver and interobserver differences in area measurements (n=3421; r=.97 to.99) were <1.6% (SD<7.2%); intraobserver and interobserver differences in volumetric measurements (n=30; r=.99) were <0.4% (SD<3.2%). ECG-gated acquisition of 3D IVUS image sets is feasible and permits the application of automated contour detection to provide reproducible measurements of the lumen and atherosclerotic plaque CSA and volume in a relatively short analysis time.

Ultrasound contrast agent in intravascular echography: An in vitro study

The intravascular ultrasound image of the intraluminal contour depends on the difference between acoustic impedances of the media which create the endoluminal interface. There are several limitations to the visualization and detection of this interface. These limitations are due to artifacts encountered during image formation and to anatomical complexity. The purpose of this study is to obtain intraluminal contour enhancement using ultrasound contrast agent (UCA). Therefore, our objective was to address the feasibility of this technique by documenting the following: (i) the acoustic properties of UCA at 30 MHz; (ii) in vitro experimentation with tube or postnecrotic artery; and (iii) suitable digital processing. The images obtained with UCA (enhanced image quality) and subtracted from those without UCA provided, after simple digital processing, accurate visualization of the arterial lumen. The image obtained exhibits an even, high-contrast intraluminal edge. Such characteristics facilitate contour extraction by the automated contour detection procedures.

Accuracy and precision of angiographic volumetry methods for left and right ventricle

We imaged and quantified 60 ventricle casts (30 LV, 30 RV) to evaluate the accuracy and reliability of angiographic ventricle volumetry. We analyzed the seven biplane methods most frequently used in clinical routine: Arcilla, Arvidsson, Dodge, Ferlinz, Simpson (LV + RV) and Wynne. The ventricle contours were defined by (1) manual drawing on the computer screen, (2) manual drawing using a graphical tablet and (3) automatic contour detection. A high inter-class variation in volume accuracy between the different methods was observed (S.D. = 12.7 ml). The volume methods for the LV (mean differences MD LV: [−2.2, +8.5] ml, average MD LV = 1.8 ml) are more accurate than for the RV (MD RV: [−11.4, +33.1] ml, average MD RV = 12.1 ml). The intrinsic error is about the same for all approaches and is very high: average S.D. = 20 ml, RMS = 185 ml. Manual contour definition results in a volume over-estimation (average MD man = +32.8 ml, r = 0.731) compared with automatic contour detection (average MD auto = +6.2 ml, r = 0.810). LV hypertrophy results in a volume under-estimation of the LV (MD LV = −7 ml) and an over-estimation of the RV (MD RV = +6 ml). RV hypertrophy leads to the opposite effect. It was shown that ventricle volumetry and the calculation of derived parameters (ejection fraction) is extremely case dependent and can only be an estimate of the actual value.

Directional Coronary Atherectomyによる切除標本の病理組織学的検討

This study characterized histopathologic findings using specimens excised during directional coronary atherectomy (DCA) in patients with different ischemic syndromes.We reviewed the histopathologic findings in 61 coronary atherectomy specimens from 61 consecutive patients. Clinical subgroups were classified into three groups: primary lesions of stable angina (Group P, n=7), 29 restenosis lesions (Group R, n=29) and acute coronary syndromes (Group A, n=25). These patients were diagnosed according to AHA criteria on admission. Fragments retrieved were screened on light microscopy after Hematoxylin-eosin staining to evaluate for presence of calcification, thrombus, angiogenesis, stellate cell proliferation, cholesterin, hemosiderin, foam cells and inflammation cells. All specimens were reviewed by two independent observers, using the light microscope, who were unaware of the clinical data. Angiograms were suitable for quantitative analysis by a cardiac image analysis system (Cardio 500, Kontron Elektronik GmbH). Automated contour detection was performed by a geometric edge differentiation technique. Restenosis was defined as presence of stenosis≥50% at follow-up angiography.There were no significant differences among the three groups in terms of reference diameter, or minimal lumen diameter (MLD) after DCA. MLD and severity of stenosis before DCA tended to be less in the primary group than in the other two groups, although these differences were not significant. There were no significant differences among the three groups in the presence of calcification, cholesterin deposit, foam cells or inflammation cells. There were less thrombus in the group P than that in the other two groups. In group A, angiogenesis was detected more frequently (60%) than in the other two groups (group P=14%, group R=35%) and hemosiderin deposit also tended to be more frequent. Stellate cell proliferation tended to be more frequent in group R (72%) than in the other two groups (P=43%, A=52%).The presence of angiogenesis correlated with acute coronary syndromes and the presence of activated smooth-muscle cells correlated with restenosis after angioplasty.

Inaccuracy of quantitative coronary arteriography when analyzed from S-VHS videotape.

In the transition period between 35-mm cinefilm as the medium for coronary arteriographic data and digital media such as CD-R, S-VHS videotape has been used both as an exchange and store medium, and for quantitative coronary arteriographic (QCA) studies. To determine the extent to which S-VHS video tape affects QCA measurements, an X-ray phantom study was completed. A plexiglass phantom with 12 straight circular tubes (0.51-5.00 mm in diameter) filled with contrast medium was recorded under clinical conditions using both the 5" and 7" modes of the image intensifier with the phantom tubes positioned horizontally as well as vertically in the field of view. The digitally acquired images were recorded on S-VHS tape without any image enhancement (raw data) and with default image enhancement. Video frames were then selected on a professional VCR such that individual tubes were positioned in the center of the field of view and digitized (512(2) x 8 bits) with a high-quality frame grabber onto a QCA workstation. The contours along the individual tubes were defined using previously validated automated contour detection techniques. For each tube, an average diameter (mm) and a standard deviation (mm) were calculated. Calibration was based on a cm-grid acquired at the same geometry as the phantom. Due to the poor signal-to-noise ratio and the limited bandwidth of the S-VHS video tape, the following objective observations were made: 1) large overestimations (up to 0.87 mm) occur for tube sizes below 1 mm for vertically positioned tubes; 2) random errors in measurements are much larger for vertically positioned tubes (0.36 mm, 7" II) than for horizontally positioned tubes (0.17 mm, 7" II); and 3) little differences in results between enhanced and nonenhanced images were found due to these deteriorating factors. In conclusion, S-VHS video tape is unacceptable for QCA and should be excluded from quantitative angiographic clinical trials.

Automated contour detection and acoustic quantification.

Echocardiographic left and right ventricular sequences are usually interpreted visually, but current quantitation techniques have been found to be tedious, time-consuming and associated with significant inter- and intra-observer variabilities. In an attempt to eliminate these problems, we have developed the Echocardiographic Analysis System (EAS) which uses robust automated border detection techniques both for single frames as well as for sequences. Comparison of LV cross-sectional area with the semi-automated border detection (AUTO) and those assessed from manual tracings (MAN) yielded a systematic difference (MAN-AUTO) of -6.6% (p < 0.001), and a random difference (standard deviation of paired signed differences) of 11.8%. Current developments are directed towards real-time automated border detection with an Accelerator board, integration of Acoustic Quantification (AQ) data as edge information into EAS, and intravascular echocardiographic applications.

Quantitative Analysis of Computed Tomography Scans of the Lungs for the Diagnosis of Pulmonary Emphysema

To develop an analytic software package based on automated contour detection for the objective and reproducible assessment of emphysema from computed tomography (CT) scans. A semiautomated technique was developed for the definition of lung contours in CT cross-sections followed by the assessment of pulmonary CT parameters describing the disease state. For 78 images, the semiautomated contour detection was performed and compared with contours drawn by an experienced radiologist by calculating the systematic area difference (bias) and differences in pulmonary CT parameters such as the mean lung density (MLD). In addition, intraobserver and interobserver variabilities were determined in a subset of 15 images. The areas enclosed by the semiautomatically detected contours were slightly larger than the manual ones (bias < 2.1%). The biases in the observer studies were smaller in the semiautomated versus the manual case (0.3% vs. 1.3%). The standard deviation of the MLD differences with a manual analysis was larger by a factor of five than in the semiautomated case. On average, manual analysis required 2 minutes, 18 seconds per lung; this time was reduced to 11.5 to 29 seconds with the semiautomated approach, depending on the respiration state. The semiautomated approach is preferred over the manual approach because of its higher consistency and its shorter analysis time.

1000–33 On-line Automated Lumen Volume Measurement With 3-D Intracoronary Ultrasound During Coronary Interventions

The aim of the study was the clinical validation of an on-line fully automated lumen detection technique with intravascular ultrasound (IVUS). Assessment of lumen dimensions with IVUS was performed in 10 pts undergoing transcatheter therapy (5 PTCA, 2 DCA, 3 Stents). The images of 10 coronary segments 20 mm long in the treated area were acquired during motorized pull-back (1 mm/s) and processed using an automated contour detection algorithm based on acoustic quantification of blood backscatter and coronary wall. For each segment, 171 ultrasound images (8.5/mm) were processed in the laboratory, immediately after acquisition (processing time: 1.35 min.) and subsequently corrected off-line by an experienced analyst. Segment lumen volume was calculated by multiplying mean lumen area by segment length. Of the 1,710 on-line IVUS cross-sections obtained, 232 (13.6%) could not be analyzed due to failure of the algorithm used for lumen detection induced by presence of side branches, lack of sharpness of the lumen-wall interface or shadowing from calcium interpreted as lumen. On the remaining frames (1,478; 86.4%), on-line vs off-line comparison for mean lumen area and segment volume was performed. On-line measurement of lumen area and volume were highly correlated with the off-line measurements (r = 0.85 for area; see fig. for lumen), with a mean difference of 7.4 ± 14.1% and 5.9 ± 5.4% for lumen area and volume, respectively. In selected optimal sequences of images acquired during coronary interventions, on-line automated quantitative 3-D provides reliable measurements of lumen dimension, requiring limited manual interaction.Download : Download high-res image (57KB)Download : Download full-size image

901-20 Usefulness of On-line 3D Reconstruction for Stent Implantation

The additional information provided by 3-D reconstruction of intracoronary ultrasound images (ICUS) during stent implantation, was assessed in 8 pts. In total 14 stents were used: 6 Palmaz-Schatz, 3 AVE and 5 Wallstent. The images of all 14 coronary segments encompassing the full stent length and the adjacent reference segments were acquired by using a motorized pullback at a constant speed (1 mm/s) and immediately processed in the interventional suite using an automated contour detection algorithm based on acoustic quantification (Figure). Residual percent area stenosis (%AS) of the stented segment, automatically measured on-line after 3-D reconstruction, was compared with ICUS measurements obtained by direct review of the video-tape. In three pts studied before stent implantation on-line 3-D reconstruction facilitated the selection of the stent length by measuring length of dissection and distance between stenosis and left main or diagonal branch. After stent implantation, the absence of stent overlapping in three patients receiving multiple stents and the relationship with side branches could be assessed. The visual information from the 3-D longitudinal views and the on-line measurements triggered additional balloon inflations in 5 pts. The selection of frames and the measurement from tape required a longer time than 3-D reconstruction and measurement (4.4 vs 2.35 min. p &lt; 0.0011 and resulted in a slight underestimation of the residual %AS with 3 pts showing a &gt;20% %AS with 3-D not recognized from tape analysis. Online 3-D reconstruction of ICUS for intracoronary stenting facilitates stent selection, evaluates adequate stent deployment by rapidly and accurately measuring residual intrastent stenosis and detecting its relationship with anatomic landmarks.Download : Download high-res image (146KB)Download : Download full-size image

1004-58 Quantitative Intravascular Ultrasound Imaging: Evaluation of an Automatic Approach

Intravascular ultrasound (IVUS) imaging allows a detailed interpretation and a precise quantitation of vascular cross-sections. To reduce the existing observer variabilities, quantitative intravascular ultrasound (OIVUS) needs to be made available. We have developed a fully automatic edge detection technique, based on adaptive active contour models and called ADDER (Adaptive Damping Dependent on Echographic Regions), that allows the quantitation of the vascular luminal area (LA). Using a 20 and 30 MHz mechanically rotated transducer mounted at the tip of a 4.8 F or 3.5F catheter, 53 normal and pathological arterial segments (from coronary, renal, splenic, iliac, and carotid arteries) were imaged in vitro. These images were analyzed by two experts, Eland E2, who manually traced the intra-luminal contour twice for each image, as well as with ADDER. Intra-observer variabilities for LA's were found to be excellent(-1.7 ± 3.3% for E1, 1.3 ± 6.5% for E2). The inter-observer variability was: 2.1 ± 4.3%. The success factor for ADDER was 98.1%. Its intra-observer variability was null, since the method always finds a unique contour. The correlation between the automatically found LA and the manual LA was: r = 0.99 (y = 1.03x + 0.89 mm 2 ). Morphometric variations between manually and automatically traced contours, analyzed by the centerline method, were 98 ± 139 μ m on average (N = 52 images). In conclusion, the ADDER automatic contour detection applied to IVUS images is robust and characterized by small systematic and random errors; therefore, OIVUS is a useful tool in clinical research trials.

New Approach to Automatic Contour Detection from Image Sequences: An Application to Ventriculographic Images

A procedure for detecting contours automatically is particularly needed when processing sequences of cardiographic images. In fact, in this field the number of images to be processed is, in general, large enough to make a manual tracing phase unacceptable. In this paper a new approach to automatic contour detection from image sequences is presented. If a rough contour of the desired structure is available on the first frame, every contour of the sequence is automatically outlined. The entire process is based on two main steps carried out in turn. First, an image is convolved with a mask of suitable size in order to transform any discontinuity, such as the boundary of the structure one is looking for, into a furrow of suitable width. Second, every point of a given starting contour, which is entirely contained inside the furrow, is made to fall to the bottom of the furrow by means of a localization algorithm.

Automated contour detection of the left ventricle in short axis view in 2D echocardiograms

This paper describes a semi-automatic system for the extraction of the endocardial border on a sequence of two-dimensional short axis echocardiograms. The delineation in the images is based on a dynamic programming technique which takes into consideration the radial and tangent gradient, the intensity, the smoothness of the extracted contour and a rough estimation of the location of the contour, derived from the previous frame of the time sequence. For the first frame, the user is asked to specify a region of interest. Within this region of interest, the developed system defines a circle as a rough estimation of the location of the endocardium. The developed system allows iteration of the delineation process to obtain a more detailed outline of the endocardium. The outlined contours on successive echo-frames are verified as the system compares their areas. This verification is important since the delineation process is purely sequential and errors occurring in an intermediate frame of the time sequence may influence the results in the subsequent frames.

Size of Cortical Bone and Relationship to Bone Mineral Density Assessed by Quantitative Computed Tomography Image Segmentation

The accuracy of the measurement of the size of cortical bone on computed tomography (CT) images of human vertebrae was evaluated using an automated contour detection and segmentation procedure. Forty human lumbar vertebrae were scanned using 8-mm slices and an automated detection for definition of trabecular and cortical region of interest. The vertebrae were embedded in a polyester resin and 8-mm-thick midvertebral specimens were excised using a diamond circular saw. Contact radiographs of these specimens were performed and, after photograph magnification, the cortical area was measured using computerized planimetry. Cortical area measured on CT images was highly correlated with the area measured by planimetry on the specimens (r = .91; P < .001) with, however, a systematic over-estimation. A significant relationship was found between density and width of the cortex (r = .56; P < .001). Computed tomography is able to assess the size of cortical bone in human vertebrae, but a threshold detection algorithm, as used in the current study, is not adequate to obtain the precise anatomic dimensions.

Size of Cortical Bone and Relationship to Bone Mineral Density Assessed by Quantitative Computed Tomography Image Segmentation

Louis O, van den Winkel P, Covens P, Schoutens A, Osteaux M. Size of cortical bone and relationship to bone density assessed by quantitative computed tomography image segmentation. Invest Radiol 1993;28:802-805. rationale and objectives. The accuracy of the measurement of the size of cortical bone on computed tomography (CT) images of human vertebrae was evaluated using an automated contour detection and segmentation procedure. methods. Forty human lumbar vertebrae were scanned using 8-mm slices and an automated detection for definition of trabecular and cortical region of interest. The vertebrae were embedded in a polyester resin and 8-mm-thick midvertebral specimens were excised using a diamond circular saw. Contact radiographs of these specimens were performed and, after photograph magnification, the cortical area was measured using computerized planimetry. results. Cortical area measured on CT images was highly correlated with the area measured by planimetry on the specimens (r = .91;P< .001) with, however, a systematic overestimation. A significant relationship was found between density and width of the cortex (r = .56;P< .001). conclusions. Computed tomography is able to assess the size of cortical bone in human vertebrae, but a threshold detection algorithm, as used in the current study, is not adequate to obtain the precise anatomic dimensions.