Krishna Subramanyan

Three-dimensional Bolton-Brush Growth Study landmark data : Ontogeny and sexual dimorphism of the Bolton standards cohort

OBJECTIVE The treatment of craniofacial reconstructive surgery patients may benefit from comparison to average referent three-dimensional landmark data. These data may be useful for diagnosis, treatment planning, prosthetic design, or outcomes assessment. With regard to subadult patients, we hypothesize that the pattern of ontogenetic shape change of same sex, same ethnicity, referent populations will show gross uniformity. We present a preliminary shape analysis of 50 three-dimensional landmarks derived from 317 Bolton-Brush Growth Study biorthogonal image pairs. We determine which landmarks can be collected from scanned radiographs reliably by four operators for the precisely locatable points, ontogenetic trends in landmark configuration shape change, and patterns of sexual dimorphism. PARTICIPANTS Participants were Bolton standards individuals (16 male and 16 female) who contributed biplane cephalograms seven or more times with annual or greater spacing between ages 3 and 18 years. DESIGN After removing outliers, we searched for ontogenetic heterogeneity, including sexual dimorphism and within sex-specific Procrustes coordinate shape spaces. RESULTS A cut-off of 4.3-mm interoperator error left 32 landmarks in our analysis. Three different approaches (principal component analysis, age-trend analysis, and principal components of age residuals) all found no patterns of individual variation around sex-specific average trends of shape change. Male shape change peaks at age 15, a correlate of the growth spurt. CONCLUSIONS Simultaneous frontal and lateral anatomic landmark identification improves three-dimensional localization reliability. Three-dimensional craniodental shape change from ages 8 to 18 within the Bolton standards presents little heterogeneity. Considerations of ethnicity aside, these may be initial grounds for use of these data as a normative referent.

Assessment of automatic vessel tracking techniques in preoperative planning of transluminal aortic stent graft implantation.

Objective: To evaluate automatic vessel tracking techniques in the course of preoperative planning prior to transluminal aortic endograft implantation by comparing accuracy, reproducibility, and postprocessing time with source image and volume-rendered analysis methods. Methods: Multislice computed tomography datasets of 5 patients with abdominal aortic aneurysms were preoperatively examined, performing volumetric analysis of diameter and position of renal artery orifices, aneurysmal neck, maximal aneurysmal extension, aortic bifurcation, and iliac arteries and bifurcation. Analysis was realized by utilizing transverse datasets, volume rendering, and automated vessel tracking strategies (MxView, Philips, Best, The Netherlands). Measurement techniques were evaluated by 2 independent readers 3 times for each patient and measurement modality. Statistical analysis evaluated accuracy of the measurements and intra- and interobserver reliability. Postprocessing time was documented. Results: Using transverse source datasets, intraobserver reliability ranged from 0.49 to 0.58. Intraobserver reliability improved to 0.7 to 0.98 when volume-rendered datasets were evaluated. Interobserver variability for transverse and volume-rendered datasets ranged from 0.49 to 0.76 and 0.70 to 0.96, respectively. Automated vessel tracking datasets did not demonstrate any intra- or interobserver variability. Based on transverse datasets, the length and diameter of iliac arteries and location and diameter of the aneurysmal neck were measured as statistically different in all cases in contrast to volume rendering and automated segmentation techniques. Postprocessing time consumption for measurements based on transverse, volume-rendered, and automated tracking segmentation datasets averaged 3.32 minutes, 25.43 minutes, and 2.24 minutes, respectively. Conclusions: Preoperative measurements improve significantly if datasets are evaluated based on volume-rendering techniques. This time-consuming procedure can be shortened, while further reducing observer variability, with automatic segmentation techniques.

Toward computer-aided emphysema quantification on ultralow-dose CT: reproducibility of ventrodorsal gravity effect measurement and correction

Computer aided quantification of emphysema in high resolution CT data is based on identifying low attenuation areas below clinically determined Hounsfield thresholds. However, the emphysema quantification is prone to error since a gravity effect can influence the mean attenuation of healthy lung parenchyma up to ± 50 HU between ventral and dorsal lung areas. Comparing ultra-low-dose (7 mAs) and standard-dose (70 mAs) CT scans of each patient we show that measurement of the ventrodorsal gravity effect is patient specific but reproducible. It can be measured and corrected in an unsupervised way using robust fitting of a linear function.

A radial adaptive filter for metal artifact reduction

High-density objects, such as metal prostheses or surgical clips, generate streak-like artifacts in CT images. We designed a radial adaptive filter, which directly operates on the corrupted reconstructed image, to effectively and efficiently reduce such artifacts. The filter adapts to the severity of local artifacts to preserve spatial resolution as much as possible. The widths and direction of the filter are derived from the local structure tensor. Visual inspection shows that this novel radial adaptive filter is superior with respect to existing methods in the case of mildly distorted images. In the presence of strong artifacts we propose a hybrid approach. An image corrected with a standard method, which performs well on images with regions of severe artifacts, is fused with an adaptively filtered clone to combine the strengths of both methods.

Semi-automatic procedure to extract Couinaud liver segments from multislice CT data

Liver resection and transplantation surgeries require careful planning and accurate knowledge of the vascular and gross anatomy of the liver. This study aims to create a semi-automatic method for segmenting the liver, along with its entire venous vessel tree from multi-detector computed tomograms. Using fast marching and region-growth techniques along with morphological operations, we have developed a software package which can isolate the liver and the hepatic venous network from a user-selected seed point. The user is then presented with volumetric analysis of the liver and a 3-Dimensional surface rendering. Software tools allow the user to then analyze the lobes of the liver based upon venous anatomy, as defined by Couinaud. The software package also has utilities for data management, key image specification, commenting, and reporting. Seven patients were scanned with contrast on the Mx8000 CT scanner (Philips Medical Systems), the data was analyzed using our method and compared with results found using a manual method. The results show that the semi-automated method utilizes less time than manual methods, with results that are consistent and similar. Also, display of the venous network along with the entire liver in three dimensions is a unique feature of this software.

Automatic vessel extraction and abdominal aortic stent planning in multislice CT

The abdominal aorta is the most common site for an aneurysm, which may lead to hemorrhage and death, to develop. The aim of this study was to develop a semi-automated method to de-lineate the vessels and detect the center-line of these vessels to make measurements necessary for stent design from multi-detector computed tomograms. We developed a robust method of tracking the aortic vessel tree with branches from a user selected seed point along the vessel path using scale space approaches, central transformation measures, vessel direction findings, iterative corrections and a priori information in determining the vessel branches. Fifteen patients were scanned with contrast on Mx8000 CT scanner (Philips Medical Systems), with a 3.2 mm thickness, 1.5 mm slice spacing, and a stack of 512x512x320 volume data sets were reconstructed. The algorithm required an initial user input to locate the vessel seen in axial CT slice. Next, the automated image processing took approximately two minutes to compute the centerline and borders of the aortic vessel tree. The results between the manually and automatically generated vessel diameters were compared and statistics were computed. We observed our algorithm was consistent (less than 0.01 S.D) and similar (less than 0.1 S.D) to manual results.

Evaluation of a noise-reduction contrast-enhancement algorithm for CT cardiac angiography

The objective of this study was to evaluate a new Cardiac Enhancement Filter (CEF) for noise reduction and edge enhancement of Computed Tomography Cardiac Angiography examinations. The filter is an adaptive noise reduction filter designed to achieve near real time functioning. Using a CT performance phantom, standard measurements of image quality, including spatial resolution, low contrast resolution, and image noise were assessed with and without the CEF. Quantitative assessment of the CEF showed slightly improved spatial resolution at 50% and 10% modulation, similar low contrast resolution and significantly lower image noise (up to 38%) characteristics. Two patient datasets were used for the clinical evaluation of the filter. The filter effectively reduced image noise by 13 to 22% in clinical datasets. These datasets exhibited a significant decrease in image noise without loss of vessel sharpness or introduction of new image artifacts. The results of the initial testing are encouraging, yet additional investigations are required to further assess the filters clinical utility.

A study investigating automated quantitative analyses of coronary multidetector computed tomography images

With the recent, rapid development of multidetector computed tomography (MDCT), excitement has built around the possibility of noninvasively imaging the coronary arteries. While the development of hardware and reconstruction technologies have advanced significantly, current image analysis techniques are dominated by manual interpretation using maximum intensity projections and volume rendering. If MDCT is to become the tool that it aims to be, objective, quantitative methods of image analysis will be necessary - not only to facilitate the study of atherosclerosis and coronary heart disease, but also for the accurate and timely interpretation of clinical data. This study focuses on the interobserver variability associated with the analysis of coronary MDCT images and a method for automatic segmentation of the same images. In the study of interobserver variability, six independent experts manually traced the luminal border in 60 randomly selected vascular cross sections (5 cross section each from: 4 LAD, 4 LCX, and 4 RCA). The images were acquired with an Mx8000 IDT 16-slice MDCT scanner. The mean unsigned difference for all observers was 0.38 ± 0.26 mm, with an average maximum difference of 1.32 mm. Using the expertly identified luminal borders, an independent standard was created by averaging the six sets of contours. This standard was then used to validate a prototypical automated segmentation system that uses dynamic programming and a knowledge-based cost function to optimally segment the luminal border. The resulting border positioning error was 0.17 ± 0.12 mm.

Quantitative analysis of vascular dimension and plaque composition in coronary multidetector computed tomography images

The noninvasive assessment of coronary atherosclerosis holds great promise for the future of cardiovascular medicine, and multidetector computed tomography (MDCT) has recently taken the lead in this area. Earlier studies have shown the ability of MDCT to visualize the coronary lumen and various types of atherosclerotic plaque. The aims of this project are to design, implement, and validate a complete system for the automated, quantitative analysis of coronary MDCT images. The developed system uses graph algorithms and knowledge-based cost functions to automatically segment the lumen and wall, and then uses pattern classification techniques to identify and quantify the tissue types found within the detected vascular wall. The system has been validated in comparison with expert tracings and labels, as well as in comparison with intravascular ultrasound (IVUS). In the former, the radial position of the lumen and adventitia were compared at 360 corresponding angular locations in 299 vascular cross sections (from 13 vessels in 5 patients: 5 RCA, 4 LAD, 4 LCX). Results show a border positioning error of 0.150 ± 0.090 mm unsigned / 0.007 ± 0.001 mm signed for the lumen, and 0.210 ± 0.120 mm unsigned / 0.020 ± 0.030 mm signed for the vessel wall. In the comparison with IVUS, the luminal and vascular cross sectional areas were compared in 7 vessels; good correlation was shown for both the lumen (R=0.83) and the vessel wall (R=0.76). The plaque characterization algorithm correctly classified 92% of calcified plaques and 87% of non-calcified plaques.

Vessel addition using fuzzy technique in CT angiography

CT angiography (CTA) is increasingly used for vascular disease assessment because of its non-invasive characteristics. In order to get a comprehensive overview of the vascular anatomy, the bone has to be removed since it occludes the cranial vessels. One of the commonly used algorithms is bone subtraction, which obtains vessel images by subtracting pre-contrast images from post-contrast images. The current problems are that it removes too much vessel and sometimes pieces of bone still exist near vessel. The purpose of this study is to provide radiologist with a fuzzy technique tool to add back parts of missing vessel. A seed point is put on part of the vessel that was not well preserved and an area is selected to restrict the vessel growing. Vessel extraction is based on fuzzy-connectedness technique proposed by Udupa in 1996. Incorporating intensity information from both pre-contrast and post-contrast images creates the membership images. The value of each voxel in the membership images represents strength of fuzzy connectedness. The bigger the strength value, the more likely the voxel belongs to the classified vessel. After choosing a threshold for the strength, the vessel is extracted and added back. This method may also apply to the whole images to segment out the bone and the vessel. The study will improve the current vessel extraction and bone removal algorithms and provide a good tool for aiding radiologist to diagnose vascular diseases.