OOSE 2014 Team 16

Customizable Brain Analysis Database

Iteration 2

Group Members

• Rachel Coston
• Ezgi Ergun
• Guannan Ren
• Kai Zhang

Vision Statement

This application will allow the users (IACL, JHU) to collect, organize, and use medical image and experiment data for research purposes. It is challenging to organize and maintain medical data because of the longitudinal nature of the studies, inconsistent data collection procedures, and the constant influx of new datasets.

Normally, when a patient schedules a visit, they go through medical examinations followed by a medical imaging procedure performed by different groups of people (imaging and testing can be done individually or within short intervals as well). Researchers are often interested in analyzing the image data with a corresponding examination result. Therefore it is important to organize the data in a way that makes it easier to associate related data (related by visit, patient, family members, diagnosis, etc.).

We are aiming to:

1- Develop a database management system that could parse, restructure, and organize different types of medical data through an intuitive user interface. We want to incorporate some basic image processing with the database system to create more applicable data from the incoming datasets. Please note that database management can only be done by authorized personnel.

2- Develop an interface to retrieve desired data from the dataset. This can be incorporated into the JIST (Java Image Science Toolkit) framework as a module.

3- Develop a database version control system (strech goal)

Feature List

• Import Data
• Upload folders of data to the database
• Data is automatically organized
• Data is automatically correlated
• Images will be processed according to the user defined input image processing along with basic image processing. The resutls will be added to the database
• Data from the image's xml file will be parsed for database fields
• Test Results will be parsed for indivual tests into database fields
• If Demographic data is entered, it will update the subject's demographic data
• All origanal images and test results will be added to the database unchanged
• Feature Extraction (key feature)
• Using the xml image file extract the metadata for the image. This includes subject id, image type, date taken, etc.
• Extract the test results from one patient and parse it into test and result pairs.
• Image Organization
• Create new directories with the format subjectID_modality_yearMonthDay.
• Place associated data in the folder
• Data Correlation (key feature)
• Associate raw images, processed images, and test results taken on the same day for the same subject
• Associate all images and test results of a subject
• Image Processing and Analysis
• Preform default image processing when data is added to the database, adding the data to the database
• Preform image processing given an image and a process, generating a report for the user. Data is not added to the database
• Database Versioning (extended feature)
• Keep a copy of the old database when data is changed in the database
• Allow users to query old versions of the database

Actors

• Database Administrator: The database administrator is the only user that can change or add the data in the database.  The database administrator has the ability to define the image processing that will occur for every input.
• User: A user has read-only access to the database.  They are able to preform image processing. but the results will not be added to the database.  A user can preform queries on the database.
• Database: The object-oriented database that holds subject data (images, test results..)

Use Cases

• Data Entry

Primary Actor: Database Administrator

Secondary Actor: Database

Goal: Insert data into the database

1. The database administrator inputs a variety of data by pointing to a directory containing the information to enter.
2. The application organizes the data to prepare it for entry
3. The application extracts the features of images from the xml file or extracts test results from the given excel document.
4. The application checks if for each test result if an image was taken of the subject on the same day.
5. The application checks if there is demographic data. If so it updates the subject in the database
6. The application processes the images for the user specified image processes. The resulting data is stored.
7. The application enters the data into the database. This includes the image files.
8. The database administrator sees the report from the entry. The report contains the types of image processing preformed, a list of images entered, a list of test results entered, and a list of test results that do not correspond to an image.

• Default Image Processing

Primary Actor: Database

Goal: Apply basic image processing and the user specified default image processing to images being inserted into the database

1. For each image the database administrator inputed, the application starts running several image processing algorithms.
2. The application will filter images (Laplacian, edge detection) to increase the quality of images.
3. The application will save the processed images into the database by rename the image data by adding the information of applied filteration.
4. The application resamples the filered images to make the data comparable to other data. The orientation of images is aligned into a uniform orientation. The extracted xml file is used as the source of information to rearrange images.
5. The application will save the processed images into the database. The processed image will be renamed by adding the information of applied resampling.
6. The application applies histogram equalization and save the processed data.
7. The application applies skull stripping (BSE, BET) to the previous processed data, and the processed data will be renamed and inserted into the database.
8. The application will preform the default image processing, defined by a database administrator.
9. The application closes the session, when all image processing is over and has been inserted into the database.

• Define Default Image Processing

Primary Actor: Database Administrator

Secondary Actor: Database

Goal: Define the default image processing

1. The database administrator logs into the application.
2. The database administrator selects the define default imaging processing
3. The database administrator is presented with a popup that allows them to choose from a list of existing image processing algorithms or import a new image processing algorithm
4. If the database administrator chose to import a new algotithm it will appear in the list of existing image processing algorithms
5. The database administrator clicks OK and the default processes are saved.

• Image Processing

Primary Actor: User

Secondary Actor: Database

Goal: Query database for necessary data processing

1. The user interacts with the software through a client-side itnerface
2. The user can query the database with keywords such as name, date, modality, etc.
3. The user can choose to do image processing techniques with the returned image
4. The user will select which processing technique will be applied to the image
5. The user will receive a report on the processed image from the application
6. The user, who is not allowed to store the processed image, will now log off and end session.

GUI Sketches

The Database Administrator is used to maintain and update the database. Authorization is required. Generates a report after each update.

The Database Administrator can also view past reports and define default image processing

The image processing pipeline is implemented using Java Image Science Toolkit (JIST). JIST is a Java-based imaging processing environment which was developed as an extension to MIPAV (Medical Image Processing, Analysis, and Visualization). Using this environment, multiple modules that contain one or more algorithms can be connected into a pipeline to process the data. Output from one module is connected to another to create a continuous flow of analysis. The pipelines can be distributed using layout files (.LayoutXML). The user interface looks like this:

We can incorporate out data retrieval into the JIST framework so that the user can integrate the software into their pipeline designs. The module can be used independently to produce output files (.txt) that contain desired information from the database.

Class Design Proposal

Please click here for the JavaDoc Documentation

Activity Diagram

Domain Model

Class Diagram

Component Diagram

We didn't include component diagram here, because in the concept of our software the number of nodes are limited, and the relationships are obvious from class diagram.

Architecture

A basic client server interaction is used for this project. The client-side application allows a user to connect via own computer to the database server. The server is running Hibernate, MySQL, and JIST functions for image processing and storage. The results for processing jobs are stored on the server-side. The two main packages that we will implement includes one for the client and one for the server.

We are going to build our project on an existing Java based image processing and analysis environment (JIST). JIST is a framework for algorithm development and large-scale image processing which combines a fully functional, graphical pipeline environment with a framework to take advantage of advanced multi-modality imaging libraries available within Java. JIST integrates closely with MIPAV (Medical Image Processing, Analysis and Visualization, National Institutes of Health), a widely used and well-supported multidimensional imaging, visualization, and processing package from the National Institutes of Health which provides for user-friendly visualization and exploration of the multi-dimensional imaging data as well as three-dimensional structures.

JIST has a layout panel which provides a graphical interface to arrange image processing algorithms as modules (see GUI sketches). Modules can be dragged and dropped from the module panel into the layout panel. When a module is selected in the layout panel, a dialog appears in the parameter panel that allows the user to edit input parameters for that module [Lucas, Blake C, et al. The Java Image Science Toolkit (JIST) for rapid prototyping and publishing of neuroimaging software, Neuroinformatics, 2010 Mar].

Using this environment, multiple modules that contain one or more algorithms can be connected into a pipeline to process the data. Output from one module is connected to another to create a continuous flow of analysis. The pipelines can be distributed using layout files (.LayoutXML). There are 2 parts to our project:

1 - Database management: This may or may not be incorporated into JIST framework. It will use Hybernate to update the database.

2 - Interaction with the database: This will be implemented as a JIST module (like an image processing module). This way, users can incorporate it into their image processing pipelines. Users can specify the data they want from the database using the parameter panel.

Resources

• Java Image Science Toolkit (JIST)
• Medical Image Processing, Analysis, and Visualization application (MIPAV)
• Hibernate ORM - An Object/Relational Mapping (ORM) framework