With large amount of patient vital signs, medical image data and diagnosis, as well as treatment results accumulated daily from healthcare enterprises, managing and distributing multimedia information becomes challenging. This presentation selects five independent concepts and technologies and their intertwined relationship to exemplify goal-oriented multimedia signal processing and distribution for healthcare delivery applications. They are: medical imaging informatics; Data Grid for image fault tolerant backup and disaster recovery; computer-assisted detection and diagnosis for daily clinical practice; biometrics technology for data access security in an enterprise level integrated system; and in vivo image-guided real-time surgical treatment. The success of implementing these novel technologies and concepts would have tremendous impact to better present and future healthcare delivery.

Storage and retrieval technology for large-scale medical image systems has matured significantly during the past ten years but many implementations still lack cost-effective backup and recovery solutions. As an example, a PACS (Picture Archiving and Communication system) in a general medical center requires about 40 Terabytes of storage capacity for seven years. Despite many healthcare centers are relying on PACS for 24/7 clinical operation, current PACS lacks affordable fault-tolerance storage strategies for archive, backup, and disaster recovery. Existing solutions are difficult to administer, and often time consuming for effective recovery after a disaster. For this reason, PACS still encounters unexpected downtime for hours or days, which could cripple daily clinical service and research operations. Grid Computing represents the latest and most exciting technology to evolve from the familiar realm of parallel, peer-to-peer, and client-server models that can address the problem of fault-tolerant storage for backup and recovery of medical images. We have researched and developed a novel Data Grid testbed involving several federated PAC systems based on grid computing architecture. By integrating grid architecture to the PACS DICOM (Digital Imaging and Communication in Medicine) environment, in addition to use its own storage device, a PACS also uses a federated Data Grid composing of several PAC systems for off-site backup archive. In case its own storage fails, the PACS can retrieve its image data from the Data Grid timely and seamlessly. The design reflects the Globus Toolkit 3.0 five-layer architecture of the grid computing: Fabric, Resource, Connectivity, Collective, and Application Layers. The testbed consists of three federated PAC systems, the Fault-Tolerant PACS archive server at the Image Processing and Informatics Laboratory, the clinical PACS at Saint John's Health Center, and the clinical PACS at the Healthcare Consultation Center II, USC Health Science Campus.In the testbed, we also implement computational services in the Data Grid for image analysis and data mining. The federated PAC systems can use this resource by sharing image data and computational services available in the Data Grid for image analysis and data mining application.In the paper, we first review PACS and its clinical operation, followed by the description of the Data Grid architecture in the testbed. Different scenarios of using the DICOM store and query/retrieve functions of the laboratory model to demonstrate the fault-tolerance features of the Data Grid are illustrated. The status of current clinical implementation of the Data Grid is reported. An example of using the digital hand atlas for bone age assessment of children is presented to describe the concept of computational services in the Data Grid.

Clinical assessment of skeletal maturity is based on a visual comparison of a left-hand wrist radiograph with atlas patterns. Using a new digital hand atlas an image analysis methodology is being developed. To assist radiologists in bone age estimation. The analysis starts with a preprocessing function yielding epiphyseal/metaphyseal regions of interest (EMROIs). Then, these regions are subjected to a feature extraction function. Accuracy has been measured independently at three stages of the image analysis: detection of phalangeal tip, extraction of the EMROIs, and location of diameters and lower edge of the EMROIs. Extracted features describe the stage of skeletal development more objectively than visual comparison.

Data security becomes more and more important in telemammography which uses a public high-speed wide area network connecting the examination site with the mammography expert center. Generally, security is characterized in terms of privacy, authenticity and integrity of digital data. Privacy is a network access issue and is not considered in this paper. We present a method, authenticity and integrity of digital mammography, here which can meet the requirements of authenticity and integrity for mammography image (IM) transmission. The authenticity and integrity for mammography (AIDM) consists of the following four modules. 1) Image preprocessing: To segment breast pixels from background and extract patient information from digital imaging and communication in medicine (DICOM) image header. 2) Image hashing: To compute an image hash value of the mammogram using the MD5 hash algorithm. 3) Data encryption: To produce a digital envelope containing the encrypted image hash value (digital signature) and corresponding patient information. 4) Data embedding: To embed the digital envelope into the image. This is done by replacing the least significant bit of a random pixel of the mammogram by one bit of the digital envelope bit stream and repeating for all bits in the bit stream. Experiments with digital IMs demonstrate the following. 1) In the expert center, only the user who knows the private key can open the digital envelope and read the patient information data and the digital signature of the mammogram transmitted from the examination site. 2) Data integrity can be verified by matching the image hash value decrypted from the digital signature with that computed from the transmitted image. 3) No visual quality degradation is detected in the embedded image compared with the original. Our preliminary results demonstrate that AIDM is an effective method for image authenticity and integrity in telemammography application.

Dynamic images, a sequence of static images displayed in rapid succession and perceived as a continuous motion by the human eye, are widely used in medicine. One of the primary objectives of telemedicine is the transmission of such images to a distant location to manage clinical problems remotely. A broad variety of methods is available to acquire, store, transmit, and display these images. However, the context of the clinical problem determines which of these methods can be deployed in a telemedicine solution. This paper discusses the advantages and disadvantages of the different technologies and presents an example of a teleconferencing system for interventional cardiology. This system acquires cardiac angiography and intravascular ultrasound images and transmits them over an existing Internet connection to a distant location. It is specifically optimized for clinical conferencing, where time is limited for each case presentation during the conference, compared to the relatively long time available for the conference preparation. The system takes advantage of this characteristic by transmitting the images well in advance of the clinical conference and displaying them synchronously in both locations during the conference. This allows for the preservation of the original image quality.

Real-time consultation between referring physicians or radiologists with an expert is critical for timely and adequate management of problem cases. During consultation, both sides need to: 1) synchronously manipulate high-resolution digital radiographic images or large volume MR/CT images; 2) perform interpretation interactively; and 3) converse with audio. We present a specifically designed teleconsultation system in a digital imaging and communication in medicine picture archiving and communication systems clinical environment. The system uses bidirectional remote control technology to meet critical teleconsultation application requirements with high-resolution and large-volume medical images operated in a limited-bandwidth network setting. We give the system design and implementation methods, and also describe the teleconsultation procedure and protocol used in this system. Finally, laboratory and clinical evaluation results are discussed.

The early detection of breast cancer increases the survival rate in women. Today, film-screen mammography is the most common and effective technique for the detection of breast cancer. However, the film-screen image recording system of current mammography has several technical limitations that can reduce the breast cancer diagnostic accuracy. A state-of-the-art technology, full-field direct digital mammography (FFDDM) has the potential to increase the sensitivity of clinical and screening examinations and emerge as a film-screen mammography replacement. It is believed that efficiently delivering the superior image quality of FFDDM to expert mammographers will significantly increase the breast cancer diagnostic accuracy. This concept motivates the studies of FFDDM telemammography. In this paper, we will report the FFDDM telemammography project that is being conducted in our facility. The fundamental technology of the FFDDM system and the characteristics of FFDDM images are described. Our research approach is through three protocols: telediagnosis, teleconsultation, and telemanagement. Each of these study protocols is defined. To conduct this FFDDM telemammography project, an asynchronous transfer-mode-network-based telemammography system is developed across two remote campuses in our facility. The architecture of this system is detailed. Unlike other medical imaging modalities, the matrix of an FFDDM image is huge. How to present the image information to mammographers via currently available display media is a challenging task. Our display approaches for the FFDDM images are described. Some preliminary study results from the current phase of this study are reported.

A novel method to automatically recognize and remove background signals in computed radiography (CR) images caused by X-ray collimation during projection radiographic examinations is presented. There are three major steps in this method. In the first step, a statistical curve is derived based on many hierarchical CR sample images as a first approximation to loosely separate image and background pixels. Second, signal processing methods, including specific sampling, filtering, and angle recognition, are used to determine edges between image and background pixels. Third, adaptive parameter adjustments and consistent and reliable estimation rules are used to finalize the location of edges and remove the background. In addition, this step also evaluates the reliability of the complete background removal operation. With this novel method implemented in a clinical picture archiving and communication system (PACS) at the University of California at San Francisco, we achieved 99% correct recognition of CR image background, and 91% full background removal without removing any valid image information.

The trend in healthcare information technology is increasingly digital and multimedia oriented. The next generation of health care information systems will consist of a vast network of heterogeneous, autonomous, and distributed imaging scanners, databases, information systems, knowledge intensive applications, and large quantities of multimedia medical data. A key challenge facing system researchers and builders is to provide a new organizational framework that can integrate this varied collection of resources into what appears to be a uniform and logical conglomeration of data and knowledge store in order to increase the availability of global or previously nonaccessible information and to address demanding new information processing requirements for diverse image-assisted medical applications. The purpose of this paper is to present the authors' research toward the development of a hospital integrated framework of multimodality image base management (MIBM) for digital radiology of the future. This evolutionary framework consists of three hierarchical components: a hospital-integrated picture archiving and communication system (HI-PACS), a medical image database system (MIDS), and a set of image-based medical applications that relies on the support of MIDS and PACS. In this paper, the authors describe the system architecture, guiding principles, and design specifications of HI-PACS and MIDS and illustrate their functions and capabilities with three implemented applications, namely, patient folder workflow, distributed object management, and multimodality imaging studies. In addition, the authors conclude their findings with a summary of challenges and research directions.

The objective of radiologic image compression is to reduce the data volume of and to achieve a low bit rate in the digital representation of radiologic images without perceived loss of image quality. However, the demand for transmission bandwidth and storage space in the digital radiology environment, especially picture archiving and communication systems (PACS) and teleradiology, and the proliferating use of various imaging modalities, such as magnetic resonance imaging, computed tomography, ultrasonography, nuclear medicine, computed radiography, and digital subtraction angiography, continue to outstrip the capabilities of existing technologies. The availability of lossy coding techniques for clinical diagnoses further implicates many complex legal and regulatory issues. This paper reviews the recent progress of lossless and lossy radiologic image compression and presents the legal challenges of using lossy compression of medical records. To do so, we first describe the fundamental concepts of radiologic imaging and digitization. Then, we examine current compression technology in the field of medical imaging and discuss important regulatory policies and legal questions facing the use of compression in this field. We conclude with a summary of future challenges and research directions.>

In pattern classification problems, the choice of variables to include in the feature vector is a difficult one. The authors have investigated the use of stepwise discriminant analysis as a feature selection step in the problem of segmenting digital chest radiographs. In this problem, locally calculated features are used to classify pixels into one of several anatomic classes. The feature selection step was used to choose a subset of features which gave performance equivalent to the entire set of candidate features, while utilizing less computational resources. The impact of using the reduced/selected feature set on classifier performance is evaluated for two classifiers: a linear discriminator and a neural network. The results from the reduced/selected feature set were compared to that of the full feature set as well as a randomly selected reduced feature set. The results of the different feature sets were also compared after applying an additional postprocessing step which used a rule-based spatial information heuristic to improve the classification results. This work shows that, in the authors' pattern classification problem, using a feature selection step reduced the number of features used, reduced the processing time requirements, and gave results comparable to the full set of features.

Sampling a radiographic film containing grid line patterns during digitization may produce aliasing artifacts (Moire pattern). The authors propose a mathematical model based on the laser spot size of the digitizer, the distance and the angle between the grid line and the sampling direction to predict the amplitudes and the frequencies of aliasing artifacts. The predicted results are compared to the experimental results. Effective ways of avoiding or reducing aliasing artifacts without sacrificing too much image quality are proposed.

Hand-bone analysis with image processing techniques using a digital radiograph can be used to assess skeletal age. The analysis consists of two steps: phalangeal and carpal bone analysis. The carpal bone analysis is discussed. First, the carpal bone region of interest (CROI) is defined using a standard thresholding technique to separate the hand from the background. Then, a dynamic thresholding method with variable window sizes is used to differentiate between the bones and the soft tissue. Next, the radius, ulna, and metacarpals intersecting the borders of the CROI are removed by using mathematical morphology. Finally, all objects included in the corrected CROI are separated and described in terms of features. These features describe the size, shape, and location and include some gray-scale pixel value information. On the basis of this analysis, the separation of the noncarpal bone objects from the carpal bone is possible. The feature selection step removes features of low discriminant power and reduces the space dimension. The remaining carpal bone parameters are used for further analysis leading to skeletal age assessment.

An emerging concept in picture archiving and communication systems (PACS) is the PACS infrastructure. The critical components in this infrastructure are communication systems, cluster controller, database integration, fault-tolerant design, and systems integration software. The architecture of a comprehensive communication system within this infrastructure, which will support a digital-based radiology operation, is described. This communication network consists of a real-time analog video and digital components. The video network utilizes broadband and fiber-optic communication technologies. The digital communication is based on a three-tiered network integrating Ethernet, FDDI, and Ultranet technologies. The digital network uses the standard TCP/IP protocol and has a fault-tolerant design. Both components have been in clinical operation since the summer of 1991.>

The significance of each bit of the pixel in digital image data from various radiological modalities is tested to determine the contrast resolution. Two statistical methods, join-count statistic and chi-square goodness of fit test, are used to perform the test. Join-count statistic is used to measure the spatial coherence among pixels, while the chi-square test is used to determine if the bit data are randomly distributed. A residual image is formed by subtracting an original image from its smoothed version. The contrast revolution is determined by applying both statistics on each bit plane of the residual image starting from the least significant bit up to the bit plane whose statistic does not show a random pattern. Images from three digital modalities, computerized tomography, magnetic resonance, and computed radiography, are used to evaluate the gray-level dynamic range. Both methods are easy to implement and fast to perform.

CTIP (CT Image Processing) package is an on-line image processing software developed through the CT scanner's computer system for manipulating CT scans. The purpose of developing such an on-line system is to minimize the disadvantages from the off-line processing method, namely, the requirement of an additional image processing hardware in order to be interactive, the tediousness in handling pictorial input and output transfer, and the difficulty in using it in a clinical environment. The architecture and operating procedure of the CTIP are first given. Program descriptions together with their applications are then followed.