EnglishPortugueseSpanish

Aid System in the Diagnosis of Breast Calcifications Using Artificial Intelligence

2X place

Project summary

According to the WHO, in 2020, more than 2,2 million cases of breast cancer were diagnosed in the world. In Brazil, it was the main cause of death in the female population when it comes to cancer (INCA, 2021). The early diagnosis strategy contributes to the reduction of the cancer presentation stage (WHO, 2007). According to the Guidelines for the Early Detection of Breast Cancer, mammography is the only test whose application in screening programs is effective in reducing mortality from the disease. Calcifications are acellular abnormalities that can be identified on mammograms and can be associated with cases of cancer. Half of the biopsies performed are due to calcifications, and among these, about 35% represent malignant cases. These cases are difficult to diagnose, and in case of uncertainty, the risk can vary from 2% to 95%, and a biopsy is recommended for an accurate diagnosis, which can be an inconvenience to patients. Considering these facts, a tool to aid diagnosis was developed, using a convolutional neural network, which seeks to classify benign and malignant cases of calcifications through mammography images, helping the opinion of the mastologist on the case and the patient to have a better perspective to decide on the performance of the biopsy. Tests were carried out with several pre-trained convolutional neural networks, they were DenseNet121, ResNet152V2, InceptionResNetV2, MobileNetV2, DenseNet169, where the number of neurons and their layers were varied, to size them in the best possible way. For these networks to learn to recognize the cases of calcifications, the database used was the Digital Database for Screening Mammography, in it there were 1147 cases of calcification, which underwent pre-processing and augmentation methods. After a conversation with a physician specialized in the area of ​​mastology, Dr Damásio, the indication was given to add information such as the patient's age and breast density, as this information is also very relevant for a pre-diagnosis of the case. By finding this information for each case in our database, we implemented a functional API, it works with two processing layers, one used to read and interpret the information from the mammography image, and another layer to receive the information that they didn't do. part of the image, such as age and breast density. When testing and improving the neural networks within the API, ensemble methods were introduced, which consist of joining more than one neural network and assigning a definitive value to the case based on the result of most networks. With these methods, the results reached a margin of 86% accuracy. The mastologist physician Dr. Damasio was asked about the usefulness of a system like this. According to him, these results can already be very useful to aid in the diagnosis, especially in cases where there is no definition of conduct.

Students

Guilherme Moutinho da Silva
Felipe Matos de Campos
- - - - - - - - - - - - - - - - - - - - - - - - - -

Guidance counsellors

Augusto Bemfica Mombach
- - - - - - - - - - - - - - - - - - - - - - - - - -
- - - - - - - - - - - - - - - - - - - - - - - - - -

Institution

FEICIT - INTERNAL SCIENCE AND TECHNOLOGY FAIR
  RS -
  Brazil

Popular vote*

Did you like it? So vote and share now:

317+

Leave your comment

What did you think of this project? Participate by leaving your comment below:

Subscribe now
Notify of
guest
47 Comments
Oldest
Newest Most Voted
Inline feedback
View all comments
Vinicius Bitencourt
Vinicius Bitencourt
6 months ago

very interesting project

4+
Luciano Rodrigues
Luciano Rodrigues
6 months ago

Great job. The initiative to link artificial intelligence to healthcare is extraordinary. Eliminating the discomfort and suffering of a biopsy is everyone's dream, not to mention the cost reduction in healthcare. Congratulations to the students and the project.

3+
Danielle Hikaru
Danielle Hikaru
6 months ago

Beautiful! Congratulations on the work!!

1+
Edward Engel
Edward Engel
6 months ago

Very well done work and an important topic.

1+
Eduardo Godoi
Eduardo Godoi
6 months ago

Great work, friends!

1+
Silvana de Souza
Silvana de Souza
6 months ago

OI, very interesting project, because the sooner breast cancer is diagnosed, the probability of cure without removing the breast is greater and the less aggressive the treatment, the better for the patient. With a short-term treatment, expenditures become much lower, thus managing to treat a wider range of women.

2+
Karin W. Moutinho da Silva
Karin W. Moutinho da Silva
6 months ago

Congratulations on the work done.
Without a doubt, it is of great importance, as with less aggressive resources, without a doubt, both patients and the health area can only be thankful. ❤️

3+
Val Purper Lamb
Val Purper Lamb
6 months ago

A current theme and one that needs improvements and more advanced studies that can help in prevention. I loved the project!

1+
Jaqueline Silva
Jaqueline Silva
6 months ago

Great job. Congratulations!

1+
Angelica Kraus
Angelica Kraus
6 months ago

Congratulations on the project!! 👏🥰

1+
Lorraine
Lorraine
6 months ago

Wonderful

1+
Neuza
Neuza
6 months ago

beautiful work very well explains

1+
Telma Eunice Metz
Telma Eunice Metz
6 months ago

Very important theme, well prepared work, congratulations to the participants.

1+
Tatiane Barbosa Luz
Tatiane Barbosa Luz
6 months ago

Congratulations on the work👏👏👏it was great.

1+
Murilo Moutinho da Silva
Murilo Moutinho da Silva
6 months ago

Interesting justification and very relevant topic!!!

2+
Gabriel silva
Gabriel silva
6 months ago

Congratulations on your work, the health area can benefit a lot from these types of initiatives that support the work of its professionals.

2+
Last edited 6 months ago by Gabriel Silva
Julio Cesar Escarrone Correa
Julio Cesar Escarrone Correa
6 months ago

Congratulations on the beautiful work, everything that benefits health is important and very welcome!

1+
Silvia
Silvia
6 months ago

Important work, congratulations to everyone involved!

1+
Gilberto Roberti
Gilberto Roberti
6 months ago

Congratulations on the very interesting project….

1+
Lucinda Maria Barbieri
Lucinda Maria Barbieri
6 months ago

We need to encourage and motivate the application of young student reasoning in integrated science for the promotion and protection of public health.
I congratulate the young authors for their work and I hope that the wisdom acquired in this study will continue for their professional lives.

2+
Samuca Souza
Samuca Souza
6 months ago

Success there, God bless!!!!

1+
JANDIRA KLEIN DE OLIVEIRA
JANDIRA KLEIN DE OLIVEIRA
6 months ago

Great job, congratulations.

1+
Valtoir
Valtoir
6 months ago

This project is very interesting, I hope it reaches
The expected goal b.

1+
Leonor Sofia Ciceri of the Rock
Leonor Sofia Ciceri of the Rock
6 months ago

Important work.
Health needs much more attention in terms of economically viable diagnoses and quality when dealing with cancer patients.

2+
Anita
Anita
6 months ago

Very good

1+
Alexandra Maria Da Silva
Alexandra Maria Da Silva
6 months ago

Congratulations Guilherme and everyone in the group. Beautiful and important work!

1+
Jessica de Matos
Jessica de Matos
6 months ago

Congratulations on your work, women's health thanks you❤

1+
Victor Gonçalves
Victor Gonçalves
6 months ago

Spectacular work by young people who have come into the world with concerns about improving people's living conditions. Congratulations. Go ahead, you will have a wonderful future.

2+
Simone
Simone
6 months ago

Congratulations!!! Great job. Very happy with the chosen theme. Pride of this youth 🥰

1+
Marli
Marli
6 months ago

Beautiful work.important matter.

1+
Nicole
Nicole
6 months ago

very important project

1+
Veronica Michaut
Veronica Michaut
6 months ago

Very interesting and current theme. We need more and more to unravel everything that involves this disease so that, who knows, one day it will stop being so dangerous and frightening. Congratulations to the students!

2+
Silvana Rodrigues
Silvana Rodrigues
6 months ago

Congratulations to the students for choosing the Project's theme. It meets the search for instruments that will facilitate more accurate diagnosis of breast cancer, without needing a biopsy or unnecessary surgical incision. And to obtain the result, which is often benign, but unfortunately, doctors are only able to give a more accurate diagnosis using this procedure, which is already archaic in today's times. In addition to being a scientific and financial advance in the health field.

1+
Melissa Mello
Melissa Mello
6 months ago

Congratulations on the project! A relevant and fully applicable topic.

1+
Leila de Freitas Rodrigues
Leila de Freitas Rodrigues
6 months ago

Great project very relevant to everyday life. Excellent initiative. Congratulations.

0
Prof. deise muller
Prof. deise muller
6 months ago

Congratulations, for the work!!
Good luck with the review!!

1+
Raquel Vieira Sebastiani
Raquel Vieira Sebastiani
6 months ago

Congratulations! As one of the possible beneficiaries of this type of technology, I want you to take this project forward and for it to be incorporated into medical analyzes in this area.

1+
Vladimir Simões da Luz Junior
Vladimir Simões da Luz Junior
6 months ago

Very interesting how they approached the problem. Congratulations on your work, friends!

1+
Camila da Silva
Camila da Silva
6 months ago

Great job!!! Congratulations to the students involved🙏❤

1+
Mirela MMCStoll
Mirela MMCStoll
6 months ago

I highlight the relevance of this work, which is fundamental for taking care of people! Congratulations Congratulations!

1+
Carine Carvalho
Carine Carvalho
6 months ago

Congratulations nice job !!

1+
Aline de Souza
Aline de Souza
6 months ago

An important and very relevant project for advances in public health, especially in the early diagnosis of breast tumors. Happy birthday!

1+
Leo Weber
Leo Weber
6 months ago

Excellent project. Congratulations!!

1+
GABRIEL DORNELLES MONTEIRO
GABRIEL DORNELLES MONTEIRO
6 months ago

The State of the art for classification today is EfficientNets-V2 (convolution-based, I believe attention models are a little higher lately), hardly emsemble applied in this way is usually something applicable in the final model. These networks have a loss function with the landscape large enough to classify with 80%+ accuracy among 1000 classes, it is extremely optimizable, when that happens the data is the problem and not the model (in 99% of cases).

Giving voice to a model with much smaller computational capacity like a MobileNet is a risk to be taken as well, since probably the network with the highest performance is more likely to get it right on its own (all used architectures use residual or similar blocks like the inverted block in MobileNet ), that is, there is nothing very different, except details that increase the ability to increase performance at its limit, but these networks will never be at the architectural limit with 60~70% performance, as they reach 80+ on ImageNet with 1000 classes.

Congratulations on the work, I know that maybe you chose this because of the data limitation, but just giving a touch that in your real objective, in the real model at the end, the problem will never be the model, but the data. Success for you and continue in this research area!

1+
William Moutinho
William Moutinho
6 months ago

Hello Gabriel Dornelles, yes, we agree with you and we know that our biggest challenge at work since the beginning has been data, it is a very big challenge when you join artificial intelligence and medicine because there is a very big scarcity of data, so it is very It is common to use augmentation methods within the area. As we did not have many options for how to solve the data problem, as there were not many other databases available, we chose to try to take advantage of those we had in the best way possible, as we considered that our objective at the beginning of the work was to obtain the best possible accuracy within the our problem. We appreciate your support and feedback, and we know we can always improve within the research area. If you want to get in touch for any tips or suggestions on how we can work better with this problem, follow my email below. Thanks in advance.
guismoutinho@gmail.com

0
Ana Izabel Fernandes
Ana Izabel Fernandes
6 months ago

Congratulations guys!!! For the important choice of the theme and the project!!!

1+
Aline de Matos
Aline de Matos
6 months ago

Congratulations on your work 💙

1+

Popular vote*

Did you like it? So vote and share now:

317+

Students

Guilherme Moutinho da Silva
Felipe Matos de Campos
- - - - - - - - - - - - - - - - - - - - - - - - - -

Guidance counsellors

Augusto Bemfica Mombach
- - - - - - - - - - - - - - - - - - - - - - - - - -
- - - - - - - - - - - - - - - - - - - - - - - - - -

Institution

FEICIT – FEIRA INTERNA DE CIÊNCIA E TECNOLOGIA
  RS –
  Brazil

Educational Incentives and Awards Offered

Research Areas

Get to know the projects that are competing in all of Mostratec Virtual's research areas:

Cellular and Molecular Biology and Microbiology
Biochemistry and Chemistry
Environmental Sciences
Animal and Plant Sciences
Computer Science
Planetary and Earth Sciences and Mathematics and Physics
Health Sciences
Education and Humanities
Environmental and Sanitary Engineering
Engineering and Materials
Electrical engineering
Electronic Engineering
Mechanical Engineering
History and Social Sciences
47
0
Click to leave your comment and participate!x
()
x

Popular Jury Awards

For the People's Jury awards, only single votes (one per user) and that are cast between 0/00 and 26:10 on 23/59 will be considered valid.

Entrepreneur Profile Test

Fill in your details below to perform your test: