Performing Cell Boundary Analysis for Biomedical Research Using NI LabVIEW

Author(s):
Carlos Yapura - G Systems, Inc.
Diego Lopez - G Systems, Inc.
Giovanni Sanchez - G Systems, Inc.
Dave Baker - G Systems, Inc.

Industry:
Biotechnology

Products:
NI Vision Acquisition Software, LabVIEW, IMAQ A6822

The Challenge:
Developing an integrated solution that interfaces different brands of specular microscopes for image capture, analysis, and reporting using a single software platform, as well as developing a cell geometric analysis algorithm while complying with FDA 21 CFR Part 11 regulations.

The Solution:
Creating a custom application based on NI LabVIEW software to interface with several manufacturers’ specular microscopes using an NI-IMAQ capture board and NI vision tools to develop a user interface, perform image processing, and execute an analysis algorithm.

Creating a Single-Application System

We needed to create a system that could capture images from various microscopes using a single application. Previously, we used several software packages to acquire the images and perform analysis. The new integrated system we created using LabVIEW, NI-IMAQ hardware, and NI vision software tools provides a single-system solution for image capture, analysis, and reporting on cell geometry. The system allows the user to compare the results obtained from different microscopes using a single software application platform and performs the analysis tasks more efficiently and accurately than previous systems.

Image Capture Module

We developed the image capture module to run independently from the analysis module so that multiple users can simultaneously perform image acquisition and analysis. With the image capture module, users can acquire images from a single channel at one time from the NI-IMAQ A6822 connector block after defining the active channel in a configuration screen. Because image size and resolution may vary, the image capture module has a built-in calibration procedure so that the user can calibrate a microscope. Then the user can manually move cursors on the screen to match the calibration grid.

Additionally, the user can rotate the image in case the image calibration grid is not aligned with the image coordinate system. The video output from the active microscope is shown live on the screen and the user can snap an image at any time and save it. By implementing third-party image conversion utilities, users can save the images in multiple formats. Immediately after saving the image to disk, the cyclic redundancy check (CRC) value is calculated and stored in the database along with user-selected demographic information. If an image is reacquired, then a version number is also attached to the image file name.

Manual Center Dot Analysis

We developed an independent algorithm to determine the cell borders from user-selected center points. We used tools from the NI Vision Development Module to create the analysis algorithm, for user interaction with the image, and for overlaying analysis results on the image. We made the following basic assumptions while constructing the analysis algorithm:

• Because there is no characteristic cell length parameter, the cell image can contain cells of differing sizes.

• A unique nonintersecting triangular mesh can be drawn connecting the centers of the cells. To create this mesh, we start with an arbitrary point and its closest neighbor to form a first segment, find the closest point to the first segment to form a first triangle, and search outward from the original triangle.

• A maximum surface angle is specified to determine if a cell next to the border of the mesh is a boundary cell or an inner cell. The maximum surface angle is a configurable parameter, and cell borders can be drawn for inner cells only.

Even though we created the analysis algorithm using basic assumptions, we recommend recursive checks to ensure geometrical consistency of the triangles in the mesh as well as to determine all of the boundary cells.

Analysis GUI

The user starts the analysis procedure after an image is loaded and displayed on the screen. Then the user can choose points on the image to reposition or delete the selected point. In addition, all features work on a zoomed screen. An advantage of the current application over other similar applications is that changes to the brightness, contrast, or gamma corrections to the image do not interfere with the analysis procedure. Also, the user can choose the colors of the overlay information displayed on the screen. After performing the cell analysis, the user can still go to edit mode to view the triangle mesh and add/remove inner cell, change the neighbor relationships between any two cells, and change inner cells to border cells to redo a recursive surface angle check.

When the user is satisfied with the analysis results, a report is generated and the results are saved to a database.

Complying with FDA 21 CFR Part 11 Regulations

To comply with the FDA regulations governing electronic records, we developed the system with data integrity as the highest priority. The network administrator set procedures to limit access to network computers, shared drives, and the database to authorized personnel only. Then we implemented the following items in the application to help preserve data integrity and enable tracking of the data stored:

• The login screen requires a valid database username and password as well as a privilege setting on the database.
• We use time and date stamps to record entries to the database for tracking purposes.
• We store CRC values for the image files in the database and check when the images were loaded to ensure file integrity.
• We tag version numbers to images acquired from the same specimen and to repeated analysis reports.
• A reason for change is required for recalibration or repeated analysis.
• Database triggers enable tracking historical changes in case of an FDA audit.

We took advantage of the power and versatility of NI off-the-shelf products to create a system that exceeds the capabilities of existing commercial applications and complies with federal regulations.

Author Information:
Carlos Yapura
G Systems, Inc.
United States
carlos.yapura@gsystems.com

Browse All Case Studies »