Article Text
PDF
Abstract
Background: Precise diagnosis and follow up treatment of rheumatoid arthritis (RA) requires objective quantification, which is still lacking. For this purpose, radiological analyses are considered to be the most appropriate method.
Objective: To develop computer assisted quantification software that is particularly applicable to joint scoring in rheumatic disorders.
Methods: 3914 radiographs from hands and feet of 190 patients with RA were collected, expertly examined, analysed, and statistically evaluated. Radiographs were quantified using the conventional Larsen score and the “X-Ray RheumaCoach” (XRRC) software. The XRRC is a Java stand alone application which can support and accelerate, but not fully automate, the scoring procedure in RA. The scorer can apply both the Larsen and the Ratingen-Rau scores.
Results: Compared with conventional scoring procedures, the XRRC software accelerated quantification time by ~25%. The program, which is now available on the internet free of charge, ran stably and proved to be a consistently valuable tool.
Conclusions: Compared with conventional scoring methods, the XRRC software offers several advantages: (a) structured data analysis and input that minimises variance by standardisation; (b) faster and more precise calculation of sum scores and indices; (c) permanent data storing and fast access to the software’s database; (d) the possibility of cross calculation to other scores; (e) “user friendly” technology and a dedicated help program; (f) fast access and data transfer through the internet if desired; and (g) reliable documentation of results in a specially designed printout.
- computer assisted quantification
- Larsen score
- radiographs
- “X-Ray RheumaCoach” software
Statistics from Altmetric.com
In rheumatoid arthritis (RA), quantitative radiological evaluation of the affected joints is crucial to establishing a diagnosis and planning appropriate therapeutic intervention. Radiographic assessment of RA yields important information beyond that obtained from other variables, such as tender and swollen joint counts.1 Moreover, radiological evaluation correlate well with functional outcome.2
Many attempts have been made to develop efficacious and economic scoring methods that improve on those of Larsen3,4 and Rau,5 and their modifications, which currently represent the most widely used and valid scoring systems.
Compared with conventional scoring methods, computer assisted quantification offers several advantages: structured data analysis and input that minimises variance by standardisation, faster and more precise calculation of sum scores and indices, the possibility of cross calculation to other parameters, “user friendly” technology and a dedicated help program, fast access and optional data transfer through the internet, and reliable documentation of results in a specially designed printout.
Efforts to fully automate radiological scoring of joints affected in RA have so far not produced accurate results and have therefore not yet been used in clinical practice.6,7
Analogous scoring is a tedious and time consuming procedure, and long term studies require that many parameters and values be ascertained and statistically correlated with laboratory findings. We, therefore, developed the “X-Ray RheumaCoach” software (XRRC) to quantify RA radiographs, and will be reporting its application in recent rheumatological and radiological studies shortly (Wick M et al, unpublished data). Our studies concentrated on the advantages offered by new computer assisted, but not fully automated, quantification software, over current approaches to radiological scoring.
THE “X-RAY RHEUMACOACH” SOFTWARE (XRRC)
The XRRC is a fully operated Java stand alone application developed to support the scoring procedure in RA.
Computer hardware and software requirements
The requirements are an IBM compatible personal computer Pentium with at least 32 MB RAM and at least 200 MB hard disk space; Windows 95 or higher or Windows NT. This software can also be installed on Apple and Linux hardware and software components with equal specifications. The software is designed for easy use and networking across hospitals, clinics, and private radiological and rheumatological practices. The scorer can apply both the Larsen score and the Ratingen-Rau score.
The XRRC consists of three modules: (a) tutorial; (b) data input sheet; and (c) compute sheet. The comprehensive html format online tutorial (http://www.univie.ac.at/radio/radio.htm) encompasses the basic pathology of RA, principles of non-biased scoring, the importance of optimised and standardised radiographic projection technique, radiographic appearance of RA, and the proper use of the Larsen score.
The data input sheets (figs 1 and 2) also contain assessments of radiograph quality, brightness, contrast, spatial resolution, artefacts, and limb positioning. Scoring can also be combined with routine reporting, allowing the addition of preformatted phrases and self written comments for each joint measurement.
Data input sheet for assessing radiographs of the hands of a patient with RA.
Data input sheet for assessing radiographs of the feet of a patient with RA.
Film reading is easy to perform using the computer mouse or touch pad, clicking on the joint symbol on the screen for comparison with the automatically displayed inserts of the original Larsen score tableau and reference radiographs.
After scoring, the results are semiautomatically calculated and can be viewed on the compute sheet (fig 3), compared with data from further examinations, saved on the computer, filed in Microsoft-Excel format, printed in individual departmental layouts, and sent through the intranet or internet for interdisciplinary cooperative study designs. Data stored in a database decisively facilitate long term follow up studies. The software is programmed to accommodate permanent implementation of new modules.
Computer sheet providing detailed information on the Larsen score obtained from patients with RA.
METHODS
The XRRC software was installed on customary notebooks, and standardised evaluation sheets provided for conventional scoring.
Three thousand nine hundred and fourteen radiographs of the hands and feet of patients with RA were “blinded” and scored in chronological sequence by four experienced radiologists applying the Larsen method.4,8 Two scorers used the XRRC software, the other two scored conventionally. In each case, 64 joints were scored (hands: 4 proximal carpal, 4 distal carpal, 10 metacarpophalangeal, 8 proximal interphalangeal, 2 thumb interphalangeal, 8 distal interphalangeal joints; feet: 10 metatarsophalangeal, 8 proximal interphalangeal, 2 toe interphalangeal, 8 distal interphalangeal joints).
The exact time for quantification was taken in minutes and seconds. Time needed for intraobserver calculations and documentation was excluded.
All data are given as mean (SEM). Statistical comparisons were performed using analysis of variance following ad hoc Student’s t test analyses. Differences were considered statistically significant at p values ⩽0.05. All statistical analyses were performed using Microsoft Excel 97 and SPSS 11.0 for Windows (SPSS Inc Headquarters, Chicago, IL; http://www.spss.com).
RESULTS
Data analyses and input
Significantly decreased reporting times were recorded for all readers who applied the XRRC software (p<0.05). The mean time spent by XRRC scorers was 5.7 (0.1) minutes v 7.5 (0.5) minutes for those using conventional methods,9 an improvement in scoring time of about 25%. Radiologist preparation time and semiautomatic calculation and documentation by the software were excluded. No influence on interobserver variability was detectable and no data were lost.
Calculations and transfer of data
Sum scores, scores for individual joints, and a broad range of optional indices were calculated and stored in the software database. Export to statistical software was easy to perform and yielded data compatible with the XRRC software in clinical long term follow up and comparative therapeutic studies (Wick M, et al, unpublished data).
Reporting
No systemic errors occurred during more than 3914 scorings. Documentation of results on the department’s printout improved interdisciplinary communication and was easy to interpret. The possibility of exportation to other programs and electronic distribution facilitated and accelerated the scientific work.
DISCUSSION
By evaluating radiographs using the XRRC software, the results of our studies in patients with RA describe (a) structured data analysis and input that minimises variance by standardisation; (b) faster and more precise calculation of sum scores and indices; (c) possibility of cross calculation to other scores; (d) user friendly technology and a dedicated help program; (e) fast access and optional data transfer through the internet; (f) reliable documentation of results in a specially designed printout; (g) balanced patient joint score card; (h) optimised patient tracking; (i) quick identification of trends, even within normal ranges; (j) reduction of manual transcription time; (l) elimination of paperwork; (m) optimised access to data; and (n) dedicated online support, development, and implementation for those unfamiliar with computer applications.
The XRRC offers total overview of a patient with RA at a glance, providing up to date and complete radiological information in one place.
To our knowledge, this is the first radiological software package to present a balanced patient joint score card in the form of integrated graphs on which doctors can see the direct effect of their prescribed treatments and quickly identify trends, even within normal ranges. Treatment decisions can thus be based on the most current radiological evaluation, ensuring accuracy and completeness as well as saving time. Moreover, all affected joints can be specifically followed as well as being part of the general assessment of radiological progression for long term follow up. Data transfer through intranet or internet reduces manual transcription time and permits transmission of results electronically to healthcare providers, other health centres, or patients. The XRRC does not, however, currently allow fully automated scoring.
This program allows far greater and faster access to all data by exporting the data for multidimensional analysis. Of note, the online tutorial (http://www.univie.ac.at/radio/radio.htm), which is available to all interested parties, offers detailed information on RA, radiological quantification, software installation, and the correct use of this software, as well as allowing free downloads of the most recent versions.
In contrast with some other electronic modalities and software in radiological and physician settings,11–13 the XRRC is well established and robust software that facilitates routine and scientific work and saves time. Because the XRRC software was introduced as a routine tool for doctors in rheumatological and radiological settings, it has proved to be an important time and cost saving tool.
In conclusion, we have designed and successfully applied a new computer program that significantly improves clinical and scientific evaluation of radiographs in RA. More precise data analysis and computation of scores are a crucial part of quality control and offer the potential to replace present scoring techniques in RA. This program is currently in extensive use by the authors and several other groups. One of the major aims for the foreseeable future will be the implementation of the XRRC software to fully automated digital radiography.
Note: The authors will make the software program described herein freely available (http://www.univie.ac.at/radio/radio.htm) to those who agree not to (a) copyright; (b) patent; or (c) use the program for commercial purposes. Furthermore, user comments are invited and will be included in a collection of comparative data on conventional application of the Larsen and Ratingen-Rau scores and our new XRRC software, respectively.
REFERENCES
Footnotes
-
M Wick and P Peloschek contributed equally to this work.
-
↵Present address: M Wick, Rheumatology Unit, Department of Medicine, Karolinska Hospital, S-171 76 Stockholm, Sweden; marius.wick{at}ks.se