INESC Porto’s innovative software for breast cancer screening

Shorter waiting times and a diagnosis that is 100% reliable

In conjunction with a team from the Faculty of Engineering of the University of Porto (FEUP), INESC Porto has developed software that can automatically analyse mammograms. The process is based on comparing similar cases and the software automatically analyses and validates cases where a breast cancer diagnosis is clear. The project is called Semantic PACS and is an innovative system that is 100% reliable when detecting malignant tumours. It reduces the waiting time for a diagnosis, thus reducing the patient’s emotional stress, and represents a significant advance in the detection of breast cancer.

More accurate and reliable in the detection of malignant tumours

Semantic PACS is the name of this software created by INESC Porto and FEUP, with the Emílio Azevedo Campo company as a partner. The software automatically analyses and validates mammograms based on similar cases, when the diagnostic result of breast cancer is clear. It can automatically validate 30% of mammograms and guarantees 100% reliability in the detection of malignant tumours. This innovative software is funded by the NSRF (The National Strategic Reference Framework) and can be integrated in hospital IT systems.

According to Jaime Cardoso, the project coordinator at INESC Porto and a researcher at the Telecommunications and Multimedia Unit (UTM), “the final aim of this project was always to improve the quality of the medical services offered to patients”. This “useful tool for radiologists can facilitate the analysis and diagnosis of each case”. The project promises positive results as the doctors will be able to categorise each new mammogram using automatic description and search methods based on the comparison of the semantic contents of archived images. The database that supported the design of this software was created at the S. João Hospital in Porto.   

Based on the experience of analysing previous mammograms, the software is able to learn and thus provide relevant information for diagnosis. “In the clinical decision-making process, using similar images, in terms of their contents, is advantageous and often imperative. Therefore, it is important to develop techniques to search these images using the contents of the image. Furthermore, this technique can then be used in the breast cancer stage to support specialists”, explains Jaime Cardoso.

Benefits for doctors and patients

According to Jaime Cardoso, one of the main attributes is that the project “approaches the problem in a different way: it aims to significantly support the breast cancer scanning stage. In this area 93% of cases are normal, and in a significant number of them, the analysis and validation can be accelerated by using the software developed during this project. By helping to diagnose more simple cases, radiologists are able to concentrate on the more complex cases”, explains the researcher. When the doctor is not able to determine if the tumour is malignant or benign, the system searches similar cases to support the doctor in the evaluation of the severity of the tumour.

In addition to the benefits for doctors, this system can also benefit patients. In fact, according to Jaime Cardoso “from a patient’s point of view, this software has the advantage of reducing the waiting time for a diagnosis and thus reducing the emotional stress during this period”. Jakub Tkaczuk, another researcher at UTM who is also involved in the project, notes that “the system is the first to offer a diagnosis immediately after the mammography. Obviously this influences the patient’s wellbeing because in most situations the patients have to wait a long time to hear a doctor’s opinion”, he adds.

This innovative project has benefited from the commercial expertise of the Emílio Azevedo Campos company and will increase the competitiveness of their range of equipment to archive and share medical images. What does the future bring? Jaime Cardoso claims that the next step is to be able to “technically validate a broader range of images, and this was not possible during the length of the Semantic PACS project”, he explains. This is not INESC Porto’s first project in the area of breast cancer. INESC Porto is involved in the BCCT project that aims to evaluate the aesthetic results of breast cancer conservative treatment (BCCT) to make it possible to consistently compare the results of different clinics and centres that practice this type of treatment. Semantic PACS once again shows INESC Porto’s contribution in the area of health.

The voice of the people

The BIP team has spoken to some of the researchers on the project and they explain the importance of being part of this team and their pride in their work.

Jaime Cardoso (researcher at UTM)

“Scientific research is often quite a thankless task because the potential impact of our work on society is not immediately visible, or at least in the long term. Engineering projects in the area of medicine are very satisfying because we can see clear results and that the project has the potential to make a difference.

Furthermore, a multidisciplinary project of this nature has the difficulty of bringing together specialists from different areas that speak “different languages”. However, having overcome this communicational difficulty and having established a clear understanding of the requirements of the application and the possibilities associated with the technology, the return was high.

I would like to take this opportunity to congratulate the team involved in Semantic PACS and highlight Inês Moreira’s work, who was part of the team and also works at the S. João Hospital. Her role in creating the database to support the design of the software was integral. The algorithms that we developed can learn from the experience of specialists looking at past cases and therefore, the quality of the database was fundamental to the success of the system".

Maria João Cardoso (researcher at UTM and a doctor at the São João Hospital)

“Any type of tool that makes diagnosis easier, especially for radiologists, by identifying similar patterns automatically, makes the process more objective and quicker and this is more than welcome.”

Jakub Tkaczuk (researcher at UTM)

“It was really rewarding to take part in this project. Thanks to Professor Jaime Cardoso I had the opportunity to broaden my knowledge and be part of an extraordinary team. It is truly rewarding to know that my work can help so many people. We hope our system will be used daily by doctors to making the decision-making process much quicker”.

Inês Domingues (researcher at UTM)

“This project was very rewarding on a personal level. I have always been interested in signal and image processing in the area of medicine. In addition to covering these two areas, Semantic PACS also looks at a feminine problem and this was especially important for me.

The fact that my PhD thesis will be on image processing in breast cancer really shows the gratification I got from Semantic PACS”.