H. K. Huang

Computer-assisted bone age assessment: image preprocessing and epiphyseal/metaphyseal ROI extraction

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.

Feature selection in the pattern classification problem of digital chest radiograph segmentation

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.

Skeletal Age Determinations in Children of European and African Descent: Applicability of the Greulich and Pyle Standards

This study assesses the value of the Greulich and Pyle method in determining the skeletal ages of healthy American children of European and African descent born after the year 1980. The hand and wrist radiographs of 534 children (265 boys, 269 girls; 260 European-Americans [EA], 274 African-Americans [AA]), ages 0 to 19 y, were analyzed by two experienced pediatric radiologists blinded to the chronological age of the subjects. A difference score was calculated for each subject by subtracting chronological age from the mean bone ages scores provided by the two raters. One group t-tests were performed to verify the hypothesis that the mean difference score was equal to zero. Skeletal age determinations by the two radiologists showed a high degree of agreement by intraclass correlation coefficient (r = 0.994). The range of values for differences in skeletal and chronological ages was very wide, indicating great individual variability. Comparisons between skeletal and chronological age only reached statistical significance in EA prepubertal girls, whose skeletal ages were delayed, on average, by three months (t = −2.9;p = 0.005). Mean difference between skeletal and chronological age in prepubertal children of African descent was 0.09 ± 0.66 y, while that in children of European descent was −0.17 ± 0.67 y; (t = 3.13;p = 0.0019). On average, the bone ages of 10% of all prepubertal AA children were 2 SD above the normative data in the Greulich and Pyle atlas, while the bone ages of 8% of all prepubertal EA children were 2 SD below. In contrast to the racial differences observed in prepubertal children, EA postpubertal males had significantly greater values for bone age than AA postpubertal males (t = 2.03;p = 0.05). In conclusion, variations in skeletal maturation in prepubertal children are greater than those reflected in the Greulich and Pyle atlas; prepubertal American children of European descent have significantly delayed skeletal maturation when compared with those of African descent; and, postpubertal EA males have significantly advanced skeletal maturation when compared with postpubertal AA males. New standards are needed to make clinical decisions that require reliable bone ages and to accurately represent a multiethnic pediatric population.

Advances in Medical Imaging

The field of medical imaging, stimulated by advances in digital and communication technologies, has grown tremendously. New imaging techniques that reveal greater anatomical detail are available in most diagnostic radiology departments. We discuss vascular imaging with ultrasound, high-resolution computed tomography of the thorax, magnetic resonance imaging applications, and picture archiving and communication systems. Vascular imaging with ultrasound requires duplex and color flow Doppler, which combine gray-scale ultrasound and the Doppler phenomenon. High-resolution computed tomography modifies conventional computed tomography technology and results in images with higher spatial resolution. Magnetic resonance imaging applications for all areas of the body are being investigated and are replacing older roentgenographic techniques such as computed tomography, arthrography, myelography, and even angiography in a growing number of indications. With these new digital imaging modalities, image management has become an important consideration that can be addressed by picture archiving and communication systems.

A CT/MR/US Picture Archiving and Communication System

A picture archiving and communication system (PACS) module for CT, MR, and US images is being developed in our department. The PACS module is composed of three subsystems: (1) image acquisition, (2) data management and (3) image display. A digital communication network links these subsystems into a functional unit. The system hardware configuration is described. Daily and yearly totals for the amount of CT, MR, and US images generated in our department are provided. The software modules for the system operation are presented. We also report on our previous experience with a MR/US PACS module for a single radiology section.

High Speed Image Compression System, Prototype And Final Configuration

A hardware implementation of the full frame image compression method developed at UCLA has been completed and is undergoing evaluation. Variable compression ratios can be achieved according to user input parameters. Current clinical acceptance studies focus on a compression ratio of 10:1. The compression and reconstruction process for 1K x 1K images each take approximately 8.5 sec. Future upgrades can enhance the performance to 1.1 sec for 1K images and 4.3 sec for 2K images. Peripheral functions incorporated into the compression module include 1K image display and parallel data link to a VAX base PACS system.

Magnetic resonance image synthesis from analytic solutions of spin-echo and radio frequency-spoiled gradient-echo images.

RATIONALE AND OBJECTIVES.To synthesize magnetic resonance images (MRI) in real-time using a minimal data set obtained with routine clinical protocols and stored in a picture archiving and communication system (PACS) database. METHODS.Analytic solutions for T1 and T2 were obtained from a double and a single spin-echo set, with routine parameters. Analytic solutions from radio frequency-spoiled gradient- echo images, with TRs as low as 33 mseconds, also were used to synthesize gradient-echo images. RESULTS.Phantom studies showed that the errors in the synthesized images were significantly smaller than the errors in the T1- and T2-calculated images and similar to the source images. The gradient-echo images resulted in significant scan time savings. CONCLUSION.MRI synthesis from analytic solutions of T1, T2, and p saves computational time and yields accurate values for the intrinsic parameters while allowing the use of routine clinical protocols. The availability of clinical images in the PACS database and the ability to synthesize images in realtime has allowed the development of a practical interactive teaching module.

Centralized vs. Distributed PACS for Intensive Care Units

One clinical environment which can immediately benefit from the implementation of a radiologic PACS is the intensive care unit (ICU). Our previous study has demonstrated the feasibility and timeliness of routine image transmission to an ICU. In anticipation of future expansion of this service, we have investigated two different models for a hospital-wide ICU PACS. These models included a centralized and a distributed processing PACS configuration. Their comparison indicated that although the distributed model offers some major advantages over the centralized model, the latter may hold a rightful place in the inter-departmental service, especially if the cost issue is a critical factor.

Performance Evaluation of a Clinical PACS Module

Picture archiving and communication systems (PACS) are now clinically available in limited radiologic applications. The benefits, acceptability, and reliablity of these systems have thus far been mainly speculative and anecdotal. This paper discusses the evaluation of a PACS module implemented in the pediatric radiology section of a 700-bed teaching hospital. The PACS manages all pediatric inpatient images including conventional x-rays and contrast studies (obtained with a computed radiography system), magnetic resonance images, and relevant ultrasound images. A six-monitor workstation is available for image review.

Interfacing a Computed Radiography System in a Centralized PACS System through a Microcomputer

Establishing control of the image data flow between a computed radiography system and a host computer presents many problems in a centralized PACS system design. Dedicating the main computer of the PACS system to service on-line acquisition devices will soon overload the system, especially when multiple image acquisition units are present. At UCLA, our approach is to distribute the task of buffering the input image data to a smart file server. Using a low-cost microcomputer as our image file server, we can reduce the burden to the central PACS computer when an image is to be acquired. This paper describes the design of the smart file server for a PCR/SP computed radiography system.