RSS Feeds

Email a Friend Printer Friendly

Increase in the reliability of brain tumour diagnosis

February 09, 2004

A team of European researchers lead by Carles Ar'°s, professor at the Department of Biochemistry and Molecular Biology of the Universitat Autònoma de Barcelona, have developed a system that facilitates the interpretation of magnetic resonance spectra of brain tumours and improves their diagnosis. It is a computer-based tool that visually classifies the different types of tumours. The new system has significantly improved the reliability of the diagnosis in preliminary tests with 16 patients.

There are 50 different types and grades of malignant tumours. The malignancy of each type of tumour is what determines if and what type of therapy is the best one to carry out. Radiologists use images of the brain, obtained by different exploratory techniques, to diagnose the type of tumour as well as using magnetic resonance (MR) spectra of the tumour. These spectra are curves with different patterns, which are associated with the abundance of the different chemical substances in its composition. Explorations with images obtained from MR have an average reliability for diagnosing a type of tumour of between 75 and 80%. The only alternative currently available to increase this reliability is the biopsy with the consequent risks that are involved with that type of intervention.

Carles Ar'°s, professor at the Department of Biochemistry and Molecular Biology of the Universitat Autònoma de Barcelona, lead a team of researchers from various European institutions in carrying out the European project called, INTERPRET (International Network for Pattern Recognition of Tumours Using Magnetic Resonance). The scientists have developed a computer-based system that classifies on the screen, in a visual manner, different types of tumours according to their magnetic resonance spectra. The system is very flexible with the origin and technical characteristics of the spectra, which makes it quite useful in cases in which the MR spectra were obtained from different apparatus or from different clinics.

Thanks to this new system, radiologists, who are accustomed to making diagnoses based on images of the brain, do not have to have specific abilities in interpreting MR spectra to improve their diagnoses. This decision support tool is based on a database that includes information on 300 brain tumours. Each one was validated by following quality control protocols established over the course of the study. A point on a graph represents each case. Its position is determined by the characteristics of the MR spectrum in such a way that tumours with a similar origin appear represented on the graph in nearby positions. When a doctor receives the spectrum corresponding to a patient with a tumour of unknown origin, the computer system locates it on the graph taking into account its distinctive characteristics. In this way, the radiologist obtains visual information on the probability that the unknown tumour is either one type or another according to the area of the graph on which it is found.

The system has been successfully tested in a preliminary study on 16 patients. In that study, the combination of information that the new system offered from the images of tumours obtained from brain explorations made it possible to achieve a 92% degree of reliability in diagnoses. That means a 4% increase in the reliability obtained with the images of just that specific group of patients.

The system has been ceded for commercialisation to the SCITO S.A. company. In addition to the participation of the Department of Biochemistry and Molecular Biology of the Universitat Autònoma de Barcelona, the following groups and institutes have participated in the research carried out: l'Institut de Diagnòstic per la Imatge and the Centre Diagnòstic in Pedralbes (Barcelona), St. George's Hospital Medical School and the University of Sussex (UK), l'INSERM/Université Joseph Fourier in Grenoble, the French company PRAXIM SARL, the Katholieke Universiteit Nijmegen (Holland) and the German company Siemens AG.

Barcelona, Universitat Autònoma de