Use of data base techniques to store medical records has existed for more than forty years. However, some aspects still remain unresolved. This includes the management of textual data and image data within a single data base. Object-orientation techniques applied to a database management system (DBMS) allow the definition of suitable data structures (e.g. to store digital images): some facilities allow the use of predefined structures when defining new ones. Currently available object-oriented DBMS, however, still have to be improved, both in the schema update and in the query facilities. We developed a prototype of a medical record which includes some multimedia features, managing both textual and image data. The prototype considers data from the medical records of patients subjected to percutaneous transluminal coronary artery angioplasty (PTCA). We developed it on a Sun workstation with Unix operating system and ONTOSTM as an object-oriented DBMS. The prototype has been named ARCADIA, which stays for A Realization of a Coronaric-artery Angioplasty Data and Image Administator.
ATime-Oriented Medical Record is a medical record based on a temporal data model. Both in the clinical field and in the research field there is a need for managing temporal aspects of data in a more functional and complete way in order to satisfy particular needs of that application field. The model we developed has been applied to data from patients who undergo coronary-artery angioplasty. These patients, after a 2-3 day long hospitalization, go through some periodical follow-up visits aimed at verifying the efficacy of angioplasty. Management of temporal aspects of this data is therefore relevant. The temporal data model has been implemented by means of an Object-Oriented Data Base Management System (OODBMS). Descriptions of the model and of some temporal queries which have been made available by the implementation of this model are provided.
After designing the temporal data model, which has the ability to manage different time granularities of the clinical information, we focused our attention on designing a window-based user-interface. We designed a window-objects hierarchy for the management of data related to PTCA patients. The user interface allows the user to:
After developing the temporal clinical data model, we are designing a Temporal Object SQL. The Temporal Object SQL relies on the non-interactive version of OSQL of ONTOS to make queries on temporal data having different time granularities (i.e. time units). Our temporal SQL uses a three-valued logic (true, false, undefined) in order to manage the uncertainty coming from relationships between temporal data at different granularities. The temporal SQL has been implemented by C++ and ONTOS.
We are strengthening the integration between textual and image data into a database. Particularly, we are designing a system which is able to integrate more image data coming from different sources: angio-images having heterogeneous clinical contents, image sequences, echocardiographic images, and images coming from nuclear medicine. We are also making efforts toward integrating textual, image and graphical data to allow a deeper analysis of the collected data.
We defined and implemented three new irreversible compression techniques for digital angiocardiographic static images: Brightness Error Limitation (BEL), Pseudo-gradient Adaptive Brightness Error Limitation (PABEL), Pseudo-gradient Adaptive Brightness and Contrast Error Limitation (PABCEL). To scan digital images we implemented an algorithm based on the Peano-Hilbert plane filling curve. We applied our compression techniques to 168 static images selected from angiocardiographic 35mm films. We achieved best compression results applying the PABCEL method, obtaining a mean compression ratio of about 8:1. Consulted cardiologists did not find significant diagnostic differences between original images and reconstructed ones.
The quantity of public-domain medical software available is huge. A classification schema may therefore be helpful. We developed the schema that includes identification data (e.g. name of the software, author...), description (hardware and software requirements), classification (software category, application domain...) and evaluation data (external quality and internal quality factors, according to some software engineering criteria). We tested the schema on the public-domain software available at the SCAMC® meetings (about 36 Mbytes). We classified the software also by employing students from a medical informatics master course in computer science. We stored the high quantity of information collected into a database we developed by Paradox.
Paroxysmal Atrial Fibrillation (PAF) is a typical cardiac pathology where both Holter monitoring and traditional ECG are inadequate instrumentation systems. Probability to record PAF events - symptomatic episodes occurring in unpredictable moments - with 24 hours recording is low. Production of an objective instrumental documentation based on ECG for on ambulatory patients is still a problem. Documentation might be obtained with a portable instrument given to the patient for self-documentation of pathological events. The device must permit many hours of active recording even if it stays in standby for months. This is allowed by the Paroxysmal Atrial Fibrillation Histogram-based Portable Identifier (PAFHPI) which we have built. The PAFHPI is a low-cost and easy-to-use portable solid state recorder. The method used to investigate atrial fibrillation events is based on morphology of some histograms: the first based on RR intervals values series and the second based on one of time-
A number of new quantifiers recently suggested for clinical practice and already provided by some monitoring instruments are considered to be insufficiently tested on the normal population. The CD-ROM based technology allows the construction of easy-to-use dynamic archives of biosignals and bioimages. They can give truly positive contributions to the needed testing procedures. This situation refers to the validation of computer algorithms for ECG analysis. We built The Politecnico Biosignals Archives on CD-ROM, where two biosignal databases from young normal subjects are stored. The first database is the Politecnico Ca Granda VCG/ECG Database on Young Normal Subject. It was developed to investigate unevident ECG variability. It consists of 23 recordings about 30 minutes long, from 23 young normal subjects. Recordings consist of the three orthogonal ECG leads corrected according to Frank. A concerted effort was made to ensure the maximum fidelity of recordings. Main technical features are: sampling frequency of 500Hz, resolution of 2.4uV/unit on 12 bits. The second database is the Poli/Medlav Database on ECG-and-Respiration. It was developed to investigate interaction among cardia and respiratory systems. It consists of 20 recordings, about 23 minutes long, from 20 young normal subjects. Recording consists of three ECG leads (the three orthogonal leads corrected according to Frank), and two respiratory traces: the respired airflow and the body volume variations, both measured with a plethysmographic chamber. Main technical features are sampling frequency of 500Hz, ECG resolution of 2.4uV/unit, respiration resolution of about 0.125ml/s/unit, both on 12 bits. Data format respects the practical standards used for similar applications in recent years.
In the occasion of non-invasive measurement of the systolic and diastolic blood pressures values, some instruments currently available, allow recording of complete pressure signals. Recent advances in biomedical signal processing have suggested new methods for a deeper investigation of such signals both in the time and in the frequency domain. The Riva-Rocci & Poli-Mi Signal Base Management System is a software package that is easily used by physicians for testing these methods in several different pathophysiological conditions on a necessary large-scale basis.
One question to be answered by any teacher relates to ways to help students when they make mistakes. To give such help in the field of teaching dynamic electrocardiography, we built up an effective integration of two CAI tools that we have developed in recent years. The first CAI tool is the D.E.K.G.-Trainer system, an instructional aid oriented towards environmental simulation and summative evaluation in the field of dynamic electrocardiography teaching. The second CAI tool is the Clinic Arrhythmia Database - a database which holds a teaching-oriented version of the arrhythmia knowledge. The integration we have implemented is the D.E.K.G-Manager system. It provides a formative evaluation of a student who has made classification mistakes. The D.E.K.G.-Manager intends, therefore, to correct the erroneous beliefs which could have driven the student to formulate a wrong diagnosis. The system has two working modes. The first mode is dedicated to formative evaluation. When the student classifies a heart beat in a wrong way using the D.E.K.G.-Trainer, our system allows the student to compare the morphological characteristics of the beat really present in the tracing with those pertaining to the erroneous diagnosis, highlighting the difference between the two descriptions. The student can also examine the segment of tracing related to a beat classified in a wrong way and analyze the classes of rhythms compatible with the class of the current beat. The second mode consists in the query of the Clinic Arrhythmia Database. In this case, the student can freely compare the different description contained in the classes of beats present in the database end examine the rhythms compatible with them. The descriptive information obtained from the Clinic Arrhythmia Database is displayed within a graphical schema similar to the Entity-Relationship model. Graphical symbols and colors are used to underline a different degree of importance of the morphological attributes typical of each considered diagnostic class. Finally, the D.E.K.G.-Manager system, though not all-encompassing, intends to be an effective and easy-to-use aid to a student who wants to fill the gaps in his knowledge of dynamic electrocardiography.
Many clinics are interested in using software packages in daily practice. Such packages, if not integrated, however, are seriously limited in their scope. This means that their daily use requires switching from one program to another and, at times, interrupting the running of an individual program. A multi-task approach would not solve the common problem with separate packages, as it does not do away with the need to type in the same data repeatedly. We developed a Multi-Service Medical Software package (MSx2), which was also developed as an example of practical integration of some clinically relevant functions. The package runs on a personal computer in an MS-DOS environment and integrates a time-oriented medical record management unit (TOMRU) and a drug information management unit (DIMU). TOMRU handles the follow-up data of ambulatory patients. DIMU performs a drug-information reference service as concerns posology, content, effects and possible interactions. Of the possible database configurations allowed by MSx2, the cardiology patient database (MSx2/C) and hypertensive patient database (MSx2/H) were developed and are described here. Configuring MSx2 should in any case be left to an expert user, in charge of database management. The clinical information to be included in the configurations was obtained by discussing the needs of and obtaining the consensus of clinical practioners. MSx2/C was handed over to hundreds of clinical centers during the computerized itinerant courses held to train would-be users. MSx2 can easily transfer patient data to statistical processing packages.