Medical imaging diagnosis (MID) is a technique and process used to produce visual representations of the human body parts, tissues, or organs for clinical diagnosis. It uses image-guided intervention procedures and is becoming increasingly important in the field of medical treatment. MID have innumerable applications, solutions and innovation that can benefit the healthcare fraternity. In the same context, computer assisted imaging is revolutionizing the medical imaging field and as such, has resulted in the steady growth of research interests among researchers worldwide. Medical images contain useful information of body parts for medical analysis, diagnosis and investigation for both research and education (Tagare et al. 1997)
Various medical imagining models, such as PACS and DICOM, short for Picture Archival and Communication Systems (Lehmann et al. 2003) and Digital Imaging and Communications respectively (Bidgood et al. 1997 ) permit easy medical images storage and transportation and thus, enhance interoperability. More often than not, varied modalities of medical images are structurally difficult and require general image processing techniques before the potential computer assisted diagnosis can be developed. To name a few, those varied modalities is Magnetic Resonance Image (MRI), Computerized Tomography (CT), Positron Emission Tomography (PET), Signal Photon Emission Computed Tomography (SPECT) ultrasound and microscope pathology and histology images (Müller et al. 2004; Robb 1999). Accordingly, radiologists or physicians in most occasions check and inspect these medical images of varied modalities in the natural ways based on their own skills and knowledge. However, such practice involves laborious and exhaustive effort and consequently in the digital and high computing era, these painstaking practices can be simplified by automating such practices which imply that.
The medical images produced by an imaging source can be automatically evaluated and match up with those stored images in a database. Subsequently, using the medical diagnosis system the potential normalities/abnormalities can be identified and recognized. With such a system, the role and function of medical imaging would expand and more focus can be given to efficient and accomplished processing, organization and interpretation rather than on image acquisition and generation.
To afford such a system, that is, a computer assisted automated diagnostic system three major modules are necessary and incorporated in, namely indexing module, classification module and retrieval module. In addition pre-processing approach can be consisting of in necessitating for image enhancement as a solution to recover the image visibility for indexing by structure contents. Over the last decade, Medical imaging diagnosis (MID) systems have been one of the mainly fastest and exciting rising research areas in the domain of medical imaging (Müller et al. 2004; Robb 1999).
1.2 PROBLEM STATEMENT
Vertebral fracture or vertebral irregularity is an extremely familiar complication of osteoporosis that has become a major public health concern. Since, a timely pharmacologic intervention can reduce the danger of further vertebral fractures, thus an early detection of vertebral fractures is very important. Even though vertebral fractures are visible on lateral x-ray radiographs, investigators have noted that vertebral fractures observable on lateral radiographs are frequently undetected by clinicians (Gehlbach et al. 2002; Probst et al. 2002) and under diagnosed by radiologists (Delmas et al. 2005; Gehlbach et al. 2000) even with the severity of the fractures.
Two major issues taking in considerations as a significant challenge in the development of such MID system, first issue concerning the development of practical algorithms and techniques for the vertebral spine fracture diagnosis, where the second issue concerning the MID design and architecture.
Developing algorithms and techniques is a general difficulty and one of the grand challenges for the automated indexing by structure contents (Antani et al. 2002). Generally, so in the case of medical images where the interest structures are usually irregular and possibly will be incompletely blocked. Technically, the computer assisted segmentation quality is affected by three significant factors.
1. The x-ray radiographs quality factor These x-ray images more often are not clear and are of inferior quality while the segmentation methods used are prone to mixing-up the tissue and vertebra boundaries.
2. Another factor related to the region of interest (ROI) whereby the vertebra has a wide range of shapes, sizes and orientations and the determination of vertebrae shape boundary is significant challenges need to be considered.
3. Another important factor is the size and resolution of the image as the cervical or lumbar spine x-ray images are usually too large. In short, it remains a challenge to achieve a fully automated and ideal segmentation of spine x-ray images.
In terms of design and architecture, our medical imaging diagnosis (MID) system targets to tolerate medical practitioners and researchers to admittance images directly through their content. In addition, the system predicts that its expansion would have various advantages in medical research, education, clinical trials and other diagnosis applications, detect., etc. example, a member of medical discipline, who is a specialist in identification spine fracture and disease, could query for cases of slight disc space narrowing in the cervical/lumbar spine for both sexes. Similarly, a clinician can exploit the system to search for a similar image to a patient's presented with such pathology.
1.3 RESEARCH OBJECTIVE AND SCOPE
With respect to the above-stated problem, this work presents the implementation research in the medical image indexing, classification and retrieval area and major trends. It also draws attention to several promising research directions such as developing combined structural design system for automatic and practical image diagnosis in the hospital and clinical environment.
As more healthcare professionals acknowledge and skills are employing the use of medical image-guided diagnosis in the course on their daily basis research and activities. The necessity for a valid, effective and practical system based image diagnosis is rapidly expanding even with the complexities to build a project class system that is robust and reliable; the benefits and profit on it would be gaining to the health community are unimaginable. The main research goal is to develop a computer assisted automated diagnostic system for vertebral fractures on lateral x-ray images with an intention to support radiologists' image interpretation and understanding and thus allowing the faster diagnosis of vertebral fracture. The early diagnosis of the spine vertebral fractures is reminiscently important for clinical trials of osteoporosis treatments. Hence, the specific activities required to achieve the goal are:
To obtain and annotate a large spine images database consist of cervical and lumbar x-ray radiographs.
2. To investigate the need of pre-processing techniques as a key to improve the image visibility for indexing by structure contents.
3. To examine algorithms that automatically locate and extract models of all vertebrae in both cervical and lumbar x-ray images.
4. To develop techniques for vertebrae contents feature extraction and geometrical measurement.
5. To develop an automated detection and classification system of the relevant pathology, consequently the pathology classification in biomedical images necessitates the decision system to be trained with several variations that can be set up.
6. To examine techniques for vertebrae image retrieval so as to extract alike images as compared to a query image using full or partial vertebral shape.
7. To develop and generate a tool that can facilitate radiologists and clinicians for osteoporosis diagnosis and treatments.
8. To develop a completely automated system for possible use in large-scale clinical trials and research.
9. To extend in the long term, the use of the computer-aided diagnosis system tools and techniques to different biomedical application.
To attain this goal, a software based medical diagnosis imaging (MDI) tool for assessing vertebral fracture of lateral x-ray images is being set as the target application to be developed. The proposed system will consist of four major sub-systems, mainly for image training and verification, image measurement and decision, image registration and image retrieval. The proposed software system will be capable of indexing, classifying and retrieving vertebral fractures by evaluating the shape and appearance of vertebrae in spine lateral x-ray radiographs. A training data set comprising x-ray images with pre-marked setting of corresponding landmark points on every vertebral contour shape performed by the medical expert will be utilized to test and validate the system in general and the algorithms, specifically.