Multi-Classifier System

Budget: 100$

Our project is a research project with a goal to deliver a complete web-based medical imaging platform. We are using a similar source like prototype-cornerstone [1, 2] as a tool to display Medical Images (DICOM) from a DICOM Server [3, 6].

At the moment, our system composes a DICOM Server that serves and manages DICOM Images as a PACS. On the other hand, we have a prototype-cornerstone [2] similar system, that uses those DICOM Images to produce several operations over them.

The prototype-cornerstone is just an incomplete sample of a similar code. The code regarding this project description is not the same since it has more functionality than this prototype-cornerstone. The structure is similar.

The tasks regarding the present project are as follows:

1. Clinical Covariates;

2. Breast Densities Classifier;

3. BI-RADS Classifier;

Understanding the task number (1) is trivial since it is the development of a new feature for the management of clinical covariates. For this feature we will need: (i) Age; (ii) Personal Background (PB); (iii) Family Background (FB); and (iv) Obesity.

For the task number (2), we aim to develop a new feature to the breast density classification on a fourfold scale. The scale is ordered by A, B, C and D, where A is the less dense (or adipose) and D is the densest.

Finally, the task number (3) is the development of a classification system for each medical image so that each modality can generate meta-information regarding the respective BI-RADS [10] of each medical image. This new dataset feature will be of chief importance to the future Machine Learning (ML) Algorithms since it will feed those algorithms with more solid, reliable and complete information.

References:

[1] [login to view URL]

[2] [login to view URL]

[3] [login to view URL]

[4] [login to view URL]

[5] [login to view URL]

[6] [login to view URL]

[7] [login to view URL]

[8] [login to view URL]

[9] [login to view URL]

[10] [login to view URL]

CSS HTML JavaScript Node.js

Prosjekt-ID: #17606783

Om prosjektet