It is within the age range of 40 to 69 years that most of the cases of deaths of Brazilian women with breast cancer are registered, many of which are discovered only when the illness is at an advanced stage. Early detection, fundamental for reverting estimates that point to 48,930 new cases and 9,335 deaths in 2006, passes through a clinical examination and mammography, normally done with an X-ray device capable of pointing to initial lesions with cancerigenic potential. In many cases, to reach a more precise diagnosis it is necessary to resort to a biopsy, which consists in removing a fragment of the suspect tissue for analysis. On average, for every eight biopsies carried out only one is positive, or that is, confirms the doctor’s initial suspicion. However, this number of unnecessary procedures represents extra costs for the Public Health Service. In order to correct this distortion a new system for processing the images, which improves the diagnosis’s quality and sensitivity, was developed at the University of Mogi das Cruzes (UMC).
“With our system the doctor is helped in the evaluation of the X-ray, principally in the identification of micro calcifications, one of the indicators of breast cancer” explains professor Annie France Frère Slaets, the research coordinator at UMC. Preliminary results of tests made at the Hospital das Clínicas of the Medical School of Ribeirão Preto, University of São Paulo (USP) have shown that the researchers are on the right track. “It’s possible to reduce by 25% the number of benign biopsies or from eight to six of the negative results, without significantly diminishing the level of detection of malignant cases” says professor Paulo Mazzoncini de Azevedo Marques, from the Image Sciences and Medical Physics Center of the Medical Clinical Department, Medical Faculty and one of the members of the research group.
The evaluation of the image processing system at the HC of Ribeirão Preto does not interfere in the medical routine of the hospital-school. There the mammography examinations carried out are initially passed to a resident doctor for making a written report, reviewed by radiologists or hired doctors. Parallel to this the same images are taken to the information technology laboratory where they are digitalized before being evaluated by software that is as yet not in a commercial format. The results of the medical written report are compared with those emitted by the computer. “We’re in the process of validating the processing system, an intermediary step so that it can be applied into the hospital’s routine” says Marques.
The images worked on up until now have allowed for the verification that, with the application of the new processing system, some procedures could have been avoided or other complementary exams could have been requested before the request for a biopsy. In the United States, similar systems approved by the American Food and Drug Administration (FDA), already make up part of their clinical routine. “There, the doctors who work with these support systems receive an additional payment from the convention fore their use” explains Marques.
Researchers at the University of Chicago were the pioneers in studies in this area. During 1998, in partnership with the company R2 Technology, they launched the first commercial system for mammography image processing approved by the FDA, called ImageChecker. Afterwards, another two similar programs were launched in the United States and one in Europe. The major difference in the United States is that digital mammography has already been inserted into the clinics’ and hospitals’ routines. This facilitates the utilization of the image processing systems, since there is not the necessity to adopt another additional procedure, such as the digitalization of images.
Here in Brazil the situation is very different. The few digital devices around are found in private clinics. The Public Health System and the majority of the private health insurance plans work with conventional X-ray mammography machines, similar to those used at the start of the 1990s when professor Annie France began to take an interest in the theme, which resulted in a project initiated in 1997 and funded by FAPESP.
At that time she was teaching at the Electrical Engineering School of USP at São Carlos, and professor Kunio Doi, from the University of Chicago, was in São Carlos to talk about the research that had resulted in the first commercial mammography image processing system. Before retiring from USP in 1998, Annie France brought together a group of researchers interested in the same area. Today many of them are spread through different and distant locations, among them places such as Mogi das Cruzes, Ribeirão Preto, Sorocaba, Canada and Japan, but they are still working on perfecting the system of mammography image processing.
In this type of exam, breast size influences the image result. “The larger the breast the fainter is the image” says the biomedical engineer Henrique Jesus Quintino de Oliveira, a member of the research team. This occurs because the X-rays deflect away when passing through the breast tissue, which results in image deformation. The effect, called dispersion, can reach up to 30% of the radiation and interfere in the image quality of the film that is going to be analyzed. In order to attain the system that will be transformed using commercial software, it was necessary to develop a computer simulation system of the breast and their images. The simulations allow for all of the parameters that contribute to the image formation to be controlled and individually evaluated, which is not possible with real images. One of these simulations has as its objective to evaluate which is the ideal exposure technique in order to detect micro calcifications.
The analyses involve all breast conditions, such as density, proportions of fat and glandular tissue, as well as the technical conditions of the X-ray equipment, the filter to be used and the exposure time.
The result is a program that analyzes details of a digitalized X-ray and acts as an interlocutor with the doctor. “It draws attention to possible details that were not detected during the visual analysis” says Annie France. In practice, it functions like a second medical opinion, a recommendation from the World Health Organization (WHO) for the evaluation of mammography. In Brazil there are more than 40 million women who find themselves within the age group indicated for mammography tests, a potential market for the dissemination of this new system. For the female population below 30 years, the most indicated examination is the ultra-sound, because in this phase the woman has more glandular tissue, difficult to be visualized when using an X-ray because it does not bring about much contrast.
Although the research had begun with the objective of improving the sensitivity of the images obtained by the X-ray machines that use film, the system that has been developed can also be used in digital mammography equipment. As the signal is already digitalized, one only needs to use the software in the digital device. All of the process becomes even simpler, and the diagnoses more precise. In this way it is possible to lower the number of unnecessary biopsies, without counting that the lesions can be diagnoses more precociously, making possible that the patient’s treatment can begin sooner.
The lowering of the number of negative biopsies obtained in the trials carried out at the Hospital das Clínicas of Ribeirão Preto, which reached the proportion of one positive case per six procedures realized, is as yet not ideal, since that recommended from the point of view of the patient’s health guarantee and service cost is one positive biopsy for every four tests carried out. But the new system has already advanced considerably and everything indicates that is should attain, with some more modifications, a relationship close to ideal.
Computerized Auxiliary Diagnosis (CAD) in the detection of micro calcification (nº 97/05019-7); Modality Regular Line of Research Assistance; Coordinator Annie France Frere Slaets – University of Mogi das Cruzes; Investment R$ 263,699.70 and US$ 74,289.61 (FAPESP)