Recognition of organs in CT-image sequences: a model guided approach

Tissue segmentation of computed tomography images using a Random Forest algorithm: a feasibility study

There is a need for robust, fully automated whole body organ segmentation for diagnostic CT. This study investigates and optimizes a Random Forest algorithm for automated organ segmentation; explores the limitations of a Random Forest algorithm applied to the CT environment; and demonstrates segmentation accuracy in a feasibility study of pediatric and adult patients. To the best of our knowledge, this is the first study to investigate a trainable Weka segmentation (TWS) implementation using Random Forest machine-learning as a means to develop a fully automated tissue segmentation tool developed specifically for pediatric and adult examinations in a diagnostic CT environment. Current innovation in computed tomography (CT) is focused on radiomics, patient-specific radiation dose calculation, and image quality improvement using iterative reconstruction, all of which require specific knowledge of tissue and organ systems within a CT image. The purpose of this study was to develop a fully automated Random Forest classifier algorithm for segmentation of neck-chest-abdomen-pelvis CT examinations based on pediatric and adult CT protocols. Seven materials were classified: background, lung/internal air or gas, fat, muscle, solid organ parenchyma, blood/contrast enhanced fluid, and bone tissue using Matlab and the TWS plugin of FIJI. The following classifier feature filters of TWS were investigated: minimum, maximum, mean, and variance evaluated over a voxel radius of 2 (n) , (n from 0 to 4), along with noise reduction and edge preserving filters: Gaussian, bilateral, Kuwahara, and anisotropic diffusion. The Random Forest algorithm used 200 trees with 2 features randomly selected per node. The optimized auto-segmentation algorithm resulted in 16 image features including features derived from maximum, mean, variance Gaussian and Kuwahara filters. Dice similarity coefficient (DSC) calculations between manually segmented and Random Forest algorithm segmented images from 21 patient image sections, were analyzed. The automated algorithm produced segmentation of seven material classes with a median DSC of 0.86  ±  0.03 for pediatric patient protocols, and 0.85  ±  0.04 for adult patient protocols. Additionally, 100 randomly selected patient examinations were segmented and analyzed, and a mean sensitivity of 0.91 (range: 0.82-0.98), specificity of 0.89 (range: 0.70-0.98), and accuracy of 0.90 (range: 0.76-0.98) were demonstrated. In this study, we demonstrate that this fully automated segmentation tool was able to produce fast and accurate segmentation of the neck and trunk of the body over a wide range of patient habitus and scan parameters.

Segmentation of abdominal organs from CT using a multi-level, hierarchical neural network strategy

Precise measurements on abdominal organs are vital prior to the important clinical procedures. Such measurements require accurate segmentation of these organs, which is a very challenging task due to countless anatomical variations and technical difficulties. Although, several features with various classifiers have been designed to overcome these challenges, abdominal organ segmentation via classification is still an emerging field in order to reach desired precision. Recent studies on multiple feature-classifier combinations show that hierarchical systems outperform composite feature-single classifier models. In this study, how hierarchical formations can translate to improved accuracy, when large size feature spaces are involved, is explored for the problem of abdominal organ segmentation. As a result, a semi-automatic, slice-by-slice segmentation method is developed using a novel multi-level and hierarchical neural network (MHNN). MHNN is designed to collect complementary information about organs at each level of the hierarchy via different feature-classifier combinations. Moreover, each level of MHNN receives residual data from the previous level. The residual data is constructed to preserve zero false positive error until the last level of the hierarchy, where only most challenging samples remain. The algorithm mimics analysis behaviour of a radiologist by using the slice-by-slice iteration, which is supported with adjacent slice similarity features. This enables adaptive determination of system parameters and turns into the advantage of online training, which is done in parallel to the segmentation process. Proposed design can perform robust and accurate segmentation of abdominal organs as validated by using diverse data sets with various challenges.

Landmarking and segmentation of computed tomographic images of pediatric patients with neuroblastoma

OBJECTIVES

Segmentation and landmarking of computed tomographic (CT) images of pediatric patients are important and useful in computer-aided diagnosis, treatment planning, and objective analysis of normal as well as pathological regions. Identification and segmentation of organs and tissues in the presence of tumors is difficult. Automatic segmentation of the primary tumor mass in neuroblastoma could facilitate reproducible and objective analysis of the tumor's tissue composition, shape, and volume. However, due to the heterogeneous tissue composition of the neuroblastic tumor, ranging from low-attenuation necrosis to high-attenuation calcification, segmentation of the tumor mass is a challenging problem. In this context, we explore methods for identification and segmentation of several abdominal and thoracic landmarks to assist in the segmentation of neuroblastic tumors in pediatric CT images.

MATERIALS AND METHODS

Methods are proposed to identify and segment automatically peripheral artifacts and tissues, the rib structure, the vertebral column, the spinal canal, the diaphragm, and the pelvic surface. The results of segmentation of the vertebral column, the spinal canal, the diaphragm and the pelvic girdle are quantitatively evaluated by comparing with the results of independent manual segmentation performed by a radiologist.

RESULTS AND CONCLUSION

The use of the landmarks and removal of several tissues and organs assisted in limiting the scope of the tumor segmentation process to the abdomen, and resulted in the reduction of the false-positive error rates by 22.4%, on the average, over ten CT exams of four patients, and improved the result of segmentation of neuroblastic tumors.

A deformable model for automatic CT liver extraction

RATIONALE AND OBJECTIVES

This study was performed to design an automatic liver region extraction system to facilitate clinical liver size estimation and further serve as a prestage for liver reconstruction and volume estimation.

MATERIALS AND METHODS

We present a modification of the well-known snakes algorithm for extracting liver regions in noisy CT images. Our modification addresses the issues of selection of the control points on an estimate of the contour and the determination of the weighting coefficients. The weighting coefficients are determined dynamically on the basis of the distance between the control points and the local curvature of the contour.

RESULTS

The proposed method was used in extracting liver regions from 98 cross-sectional abdominal images. The overall performance was estimated by comparisons with original liver regions.

CONCLUSION

The deformable model method enables an efficient and effective automatic liver region extraction in noisy environments. This approach eliminates human-in-the loop, which is the common practice for the majority of current methods.

Specially adapted interactive tools for an improved 3D-segmentation of the spine

For imaging purposes of the spine, segmented image data provides the basis for a variety of modern clinical applications. However, the anatomical complex structure of the spine as well as the extensive degenerative bony deformations apparent in the clinical situation, generally complicate the application of a fully automated segmentation. To serve the special needs for image segmentation of the spine anatomy a newly developed software system is presented, that implements specially adapted interactive tools, taking its 'axis'-skeletal structure into account. A standardized protocol combines the newly developed interactive tools (rotation transformation, warped dissection plane) with standard segmentation tools to provide both a fast and accurate segmentation procedure. The introduced software environment has been valuable for the segmentation of cervical, thoracic and lumbar spines segments based on clinical routine and research images.

Determination of height, weight, body mass index, and body surface area with a single abdominal CT image

Techniques for estimation of an individual's height, weight, body mass index (BMI), and body surface area (BSA) with a single abdominal computed tomographic (CT) image were developed. Eighty-seven abdominal CT examinations performed in adult humans were analyzed. Anatomic structures were outlined on the CT section that included L1. Multiple linear regression analysis was used to derive sex-specific predictive equations. Correlation for height was good (r > 0.65). Relationship between predicted weight and actual weight was good (r > 0.93). For BMI and BSA, r was greater than 0.893 and greater than 0.895, respectively. In this study, predictive equations for height, weight, BMI, and BSA were generated.

Knowledge-based segmentation of pediatric kidneys in CT for measurement of parenchymal volume

PURPOSE

The purpose of this work was to develop an automated method for segmenting pediatric kidneys in helical CT images and measuring their volume.

METHOD

An automated system was developed to segment the kidneys. Parametric features of anatomic structures were used to guide segmentation and labeling of image regions. Kidney volumes were calculated by summing included voxels. For validation, the kidney volumes of four swine were calculated using our approach and compared with the "true" volumes measured after harvesting the kidneys. Automated volume calculations were also performed in a cohort of nine children.

RESULTS

The mean difference between the calculated and measured values in the swine kidneys was 1.38 ml. For the pediatric cases, calculated volumes ranged from 41.7 to 252.1 ml/kidney, and the mean ratio of right to left kidney volume was 0.96.

CONCLUSION

These results demonstrate the accuracy of a volumetric technique that may in the future provide an objective assessment of renal damage.

Automated measurement of single and total lung volume from CT

PURPOSE

The goal of this work was to develop an automated method for calculating single (SLV) and total (TLV) lung volumes from CT images.

METHOD

Patients underwent volumetric CT scanning through the entire chest in a single breath-hold, as well as pulmonary function tests. An automated, knowledge-based system was developed to segment the lungs in the CT images. Image-processing routines were used to extract sets of voxels from the image data that were identified by matching them to anatomical objects defined in a model. SLV and TLV were calculated by summing included voxels.

RESULTS

For 43 patients analyzed, TLV from CT and total lung capacity from body plethysmography were strongly correlated (r = 0.90). On average, the CT-derived volume of the left lung accounted for 47.2% of the total.

CONCLUSION

A knowledge-based approach to segmentation of the lungs in CT can be used to automatically estimate SLV and TLV.

Knowledge-based method for segmentation and analysis of lung boundaries in chest X-ray images

We present a knowledge-based approach to segmentation and analysis of the lung boundaries in chest X-rays. Image edges are matched to an anatomical model of the lung boundary using parametric features. A modular system architecture was developed which incorporates the model, image processing routines, an inference engine and a blackboard. Edges associated with the lung boundary are automatically identified and abnormal features are reported. In preliminary testing on 14 images for a set of 18 detectable abnormalities, the system showed a sensitivity of 88% and a specificity of 95% when compared with assessment by an experienced radiologist.

Method for segmenting chest CT image data using an anatomical model: preliminary results

We present an automated, knowledge-based method for segmenting chest computed tomography (CT) datasets. Anatomical knowledge including expected volume, shape, relative position, and X-ray attenuation of organs provides feature constraints that guide the segmentation process. Knowledge is represented at a high level using an explicit anatomical model. The model is stored in a frame-based semantic network and anatomical variability is incorporated using fuzzy sets. A blackboard architecture permits the data representation and processing algorithms in the model domain to be independent of those in the image domain. Knowledge-constrained segmentation routines extract contiguous three-dimensional (3-D) sets of voxels, and their feature-space representations are posted on the blackboard. An inference engine uses fuzzy logic to match image to model objects based on the feature constraints. Strict separation of model and image domains allows for systematic extension of the knowledge base. In preliminary experiments, the method has been applied to a small number of thoracic CT datasets. Based on subjective visual assessment by experienced thoracic radiologists, basic anatomic structures such as the lungs, central tracheobronchial tree, chest wall, and mediastinum were successfully segmented. To demonstrate the extensibility of the system, knowledge was added to represent the more complex anatomy of lung lesions in contact with vessels or the chest wall. Visual inspection of these segmented lesions was also favorable. These preliminary results suggest that use of expert knowledge provides an increased level of automation compared with low-level segmentation techniques. Moreover, the knowledge-based approach may better discriminate between structures of similar attenuation and anatomic contiguity. Further validation is required.

Invited review: computer aids for decision-making in diagnostic radiology--a literature review

This review looks at a variety of different ways in which computers can be used to assist in the interpretation of radiological images and in radiological decision-making. The issues involved in the design of computerized decision aids are introduced and four criteria proposed for evaluating such aids: need, practicality, veracity and relevance. These criteria are used to assess research into decision aids based on: image databases, numerical methods, expert systems, image processing and image understanding systems. Possible directions for research leading to aids of practical value are discussed in the conclusion.

Segmentation of small structures in MR images: semiautomated tissue hydration measurement

Segmentation of small anatomic structures in noisy magnetic resonance (MR) images is inherently challenging because the edge information is contained in the same high-frequency image component as the noise. The authors overcame this obstacle in the analysis of the sural nerve in the ankle by processing images to reduce noise and extracting edges with an edge detection algorithm less sensitive to noise. Anatomic accuracy of the segmentation was confirmed by a neuroradiologist. A nerve hydration coefficient was determined from the signal intensity of the nerve in these segmented images. These semiautomated measurements of hydration agreed closely with those obtained with a previously described manual method (n = 44, P = .76). Each image in the study was analyzed identically, with no modification of the computer algorithm parameters. The data suggest that this robust method may be useful in a multicenter evaluation of diabetes treatment protocols.

3-D image analysis of intra-cerebral brain hemorrhage from digitized CT films

A new 3-D technique for the segmentation and quantification of human spontaneous intra-cerebral brain hemorrhage (ICH) is presented in this paper. The algorithm for ICH primary region segmentation uses the spatially weighted K-means histogram-based clustering algorithm. The ICH edema region segmentation algorithm employs an iterative morphological processing of the ICH brain data. A volume rendering technique is used for the effective 3-D visualization of ICH segmented regions. A computer program is developed for use in the human spontaneous ICH study involving a large number of patients. Experimental measurements and visualization results are presented which were computed on real ICH patient brain data.

A neural network architecture for understanding discrete three-dimensional scenes in medical imaging

Magnetic resonance and computed tomography produce sets of tomograms which are termed discrete 3D scenes. Usually, discrete 3D scenes are analyzed in two dimensions by observing each tomogram on a screen so that the three-dimensional information contained in the scene can be recovered only partially and qualitatively. The three-dimensional reconstruction of the shape of biological structures from discrete 3D scenes would allow a complete and quantitative recovery of the available information, but this task has proved hard for conventional processing techniques. In this paper we present a system architecture based on neural networks for the fully automated segmentation and recognition of structures of interest in discrete 3D scenes. The system includes a retina and two main processing modules, an Attention-Focuser System and a Region-Finder System, which have been implemented by using feed-forward nets trained with the back-propagation algorithm. This architecture has been tested on computer-simulated structures and has been applied to the reconstruction of the spinal cord and the brain from sets of tomograms.

A view from the Nijmegen PACS project

A view on PACS is given from a radiology department involved in PACS as part of its medical research environment. Special attention is payed to historical developments in medical imaging to study the context of actual PACS developments. Some directions of future diagnostic developments are indicated. Both image data-acquisition and presentation techniques are of interest to medical as well as industrial applications. It is pointed out how PACS is thought to depend on contextual factors.

Automated localization and identification of lower spinal anatomy in magnetic resonance images

Clinical interpretation of the subtle changes present in MR images in the setting of disease currently relies on subjective image analysis. Image evaluation could potentially be improved by computerized segmentation and precise quantification of the image anatomy. However, this cannot be automated unless reliable navigation within an image is established, capable of compensating for unpredictable factors such as anatomical variability, positioning of an image plane in the body, and variable image characteristics. Focusing on the lower spinal region, this paper explores the presence of image- and anatomy-invariant features which facilitate automated, unconstrained identification, and localization of basic lower spine anatomy.

Computed tomography in the assessment of the postenucleation socket syndrome

To gain a deeper insight into the cause of the postenucleation socket syndrome, high-resolution computed tomography (CT) was performed in 22 anophthalmic patients before insertion of an intraorbital implant. The anatomy of the normal and the anophthalmic orbits was compared. Computed tomographic scans were made either in the sagittal and the coronal plane or in the sagittal and transverse plane. The authors discovered a sagging and retraction of the superior muscle complex and a downward and forward redistribution of orbital fat. Finally, an upward displacement of the distal end together with a retraction of the inferior rectus muscle was found. These phenomena were measured and appear to cause a rotatory displacement of orbital contents from superior to posterior and from posterior to inferior which is best demonstrated in the sagittal plane. This redistribution of orbital contents can explain the sequelae of the anophthalmic orbit. No signs of orbital fat atrophy could be demonstrated. With this knowledge, the proper treatment of patients with a postenucleation socket syndrome is ascertained.

PACS in practice: the status of the PACS project at the St. Radboud University Hospital. Part A: Introduction and the picture system

A so-called bottom-up approach of a Picture Archiving and Communication System (PACS) is initiated by existing clinical questions. It is investigated how far existing Data-Acquisition (DA) modalities combined with the existing computer infrastructure of the hospital can provide soil to a, at first local, PACS system. Examples of current research projects in Digital Subtraction Angiography (DSA), Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) demonstrate how image and raw data processing evolve from applications based on a general matrix manipulation software package to projects within the field of PACS. Starting from the existing facilities a stepwise increase of connections and expansions of required features is going to be brought about by separately considering the picture system, the communication system and the archive system. A three step phasing is proposed: (1) Software linking; (2) Hardware linking: small scale; and (3) Hardware linking: local area network. Examples are given from the first phase, i.e. the development and expansion of software on existing DA-modalities or processing hardware to receive the data on floppy disk, hard disk or tape. Data are converted and transported for further processing: (a) within the department; (b) between hospital departments; and (c) between research centres. With regard to the picture system special attention has to be given to the requirements for digitizing analogue film images and the reading of images from monitor screens instead of films on lightboxes.