Simina Vasilache

Unified wavelet and gaussian filtering for segmentation of CT images; application in segmentation of bone in pelvic CT images

BackgroundThe analysis of pelvic CT scans is a crucial step for detecting and assessing the severity of Traumatic Pelvic Injuries. Automating the processing of pelvic CT scans could impact decision accuracy, decrease the time for decision making, and reduce health care cost. This paper discusses a method to automate the segmentation of bone from pelvic CT images. Accurate segmentation of bone is very important for developing an automated assisted-decision support system for Traumatic Pelvic Injury diagnosis and treatment.MethodsThe automated method for pelvic CT bone segmentation is a hierarchical approach that combines filtering and histogram equalization, for image enhancement, wavelet analysis and automated seeded region growing. Initial results of segmentation are used to identify the region where bone is present and to target histogram equalization towards the specific area. Speckle Reducing Anisotropic Didffusion (SRAD) filter is applied to accentuate the desired features in the region. Automated seeded region growing is performed to refine the initial bone segmentation results.ResultsThe proposed method automatically processes pelvic CT images and produces accurate segmentation. Bone connectivity is achieved and the contours and sizes of bones are true to the actual contour and size displayed in the original image. Results are promising and show great potential for fracture detection and assessing hemorrhage presence and severity.ConclusionPreliminary experimental results of the automated method show accurate bone segmentation. The novelty of the method lies in the unique hierarchical combination of image enhancement and segmentation methods that aims at maximizing the advantages of the combined algorithms. The proposed method has the following advantages: it produces accurate bone segmentation with maintaining bone contour and size true to the original image and is suitable for automated bone segmentation from pelvic CT images.

Automated bone segmentation from Pelvic CT images

Segmentation of bone tissue from pelvic CT images is a crucial step in developing an automated system for assisting experts with diagnostic decisions for traumatic pelvic injuries. The method proposed in this paper combines wavelet processing, Laplacian filtering, morphology operations, a series of region growing techniques and gradient based segmentation methods to create an automated segmentation system. The method, tested against a database of pelvic injury CT images, provides promising results. This computationally efficient method sets the grounds for creating an automated decision making system that will be able to provide physicians with reliable recommendations for the treatment of traumatic pelvic injuries.

A unified method based on wavelet filtering and Active Contour Models for segmentation of Pelvic CT images

Accurate segmentation of bone tissue from Pelvic CT images is an important step in the process of developing an automated computer aided decision making system that would provide physicians with recommendations for the diagnosis and treatment of traumatic pelvic injuries. The proposed algorithm is an automated, unsupervised, and hierarchical method for the segmentation of bone tissue. The method incorporates, as key components, wavelet processing, automated seed growing and Active Contour Models (ACMs). A wavelet based method is applied for filtering and enhancing of noisy CT images that are the target of segmentation. The main task of the proposed seed growing is to automatically find a suitable set of points for ACM initialization. Another benefit of the proposed method is that the resulting seeds are suitable for identifying small fragments of shattered bones. ACM is used to capture the edges of larger bones that, due to their natural varying densities, and consequently varying grey levels, cannot be correctly segmented by solely using seed growing. The preliminary results produced by the proposed method are very promising. The proposed method performs the challenging task of identifying the fragments of fractured bone, as well as accurately detecting the edges of bones in the pelvic region. Moreover, separation between bones is identified even in challenging areas such as hip joints.

Hierarchical object recognition in Pelvic CT images

This paper introduces a hierarchical method of recognizing bone tissue from regions extracted from Pelvic CT Images. The method allows distinguishing among segmented objects with similar grey level values, such as bone tissue and regions of active hemorrhage. The method addresses the challenge of correctly segmenting and classifying bone as well as assessing presence of active hemorrhage.

A Signal Processing Method to Explore Similarity in Protein Flexibility

Understanding mechanisms of protein flexibility is of great importance to structural biology. The ability to detect similarities between proteins and their patterns is vital in discovering new information about unknown protein functions. A Distance Constraint Model (DCM) provides a means to generate a variety of flexibility measures based on a given protein structure. Although information about mechanical properties of flexibility is critical for understanding protein function for a given protein, the question of whether certain characteristics are shared across homologous proteins is difficult to assess. For a proper assessment, a quantified measure of similarity is necessary. This paper begins to explore image processing techniques to quantify similarities in signals and images that characterize protein flexibility. The dataset considered here consists of three different families of proteins, with three proteins in each family. The similarities and differences found within flexibility measures across homologous proteins do not align with sequence-based evolutionary methods.

Outcome prediction in traumatic pelvic injuries using maximum similarity and quality measures

Traumatic pelvic injury is frequently life-threatening due to its association with severe hemorrhage and the high risk of complications. Immediate medical treatment is therefore of utmost importance; however, decisions regarding treatment are often very difficult to make due to the amount and complexity of patient information. The use of a computer-aided decision making system to help trauma surgeons assess the severity of a patient’s condition, and to make more reliable and rapid treatment decisions, could improve care giving standards and reduce the cost of trauma care. This paper focuses on creating such a system based on the rules derived through CART and C4.5. The system is designed to predict the eventual outcome of a trauma case home or rehab - by using maximum similarity, measure of discrimination, specificity and sensitivity to form a reliable decision regarding a patient’s condition.

Predicting pelvic trauma severity using features extracted from records and X-ray and CT images

Computer-aided decision making systems can assist physicians in prompt and accurate treatment of the high-energy pelvic trauma injuries by rapidly analyzing patient data and generating recommendations based on a large database of prior cases. However, no current system incorporates information contained in medical images. This paper presents a method which combines demographic information, standard medical measurements and features extracted from both X-ray images and Computed Tomography (CT) scans, to predict whether a patient will be sent to ICU after initial triage and stabilization. Predictions are presented in the form of rules, extracted from trees generated using the C4.5 algorithm. Results are promising and indicate that the image features are statistically significant in patient outcome prediction.

Traumatic Pelvic Injury Outcome Prediction by Extracting Features from Relevant Medical Records and X-Ray Images

Traumatic pelvic injuries are complex and difficult to treat, due to the high risk of complications. Prompt and accurate medical treatment is therefore vital. Computer-aided decision-making systems can assist physicians in this task, but none of those proposed so far incorporate features extracted from medical images. The study in this paper uses demographic information, standard medical measurements, and features extracted from X-ray images to predict a patients length of stay in ICU via rules extracted from decision trees generated by the CART algorithm. The X-ray features are extracted by using a spline/ASM segmentation technique to detect structure position, then calculating measures of displacement. The results are promising and compare well with SVM and C4.5 algorithms, indicating that the rules represent true data patterns. Significantly, an X-ray feature is selected as highly important to injury severity, indicating that medical image features are important in providing accurate recommendations and predictions.