Neuroscience
fMRI Overview
|
fMRI, Vision Laboratory, MSKCC, Jan-Mar, 1994 |
|
|
a PV-Wave application for analyzing, visualizing, and publishing data from functional MRI experiments |
|
| How it is used |
|
| My role |
|
|
Applicability to Medical Informatics |
|
| Overview | As a research rotation, created a graphically oriented system for automating the analysis, visualization, and presentation of data from functional MRI and PET studies in humans. Each experiment collected one gigabyte of 256x256x18 brain volume time-series while subjects performed neuropsychological experiments. This 20,000 line, object-oriented, event-driven program, written in PV-Wave, is still the primary analysis and visualization tool at the Vision Lab and its affiliates. |
| Image Preprocessing | Brain volumes had to be reconstructed from K-space via fast Fourier transforms, realigned to correct for motion artifact, filtered to increase signal-to-noise, and optionally segmented to remove the skull artifact and normalized to the Tailarach standardized anatomical space. |
| Data Analysis | Created and validated tools for traditional and proprietary analyses of time-series and spatial extent. Incorporated and tested tools for performing principle component and wavelet analysis. |
| Data Visualization | Created PV-Wave and X windows-native meta-widgets for retrieving, viewing, manipulating, and printing the data and analyses, as well as graphical tools to design the time-series analysis paradigms. Tools included multiple zoomed windows on data, region of interest analysis, color coded Z maps of analyses blended atop grayscale source images, annotated time-series graphs, cine tools for assessing motion artifact, and an analysis tracking tool to automatically label each image with a summary of the analyses performed on it. Output was conditionally sent to the screen or any of several printers. |
| Batch Mode | All data manipulations were graphically scriptable. Thus, once an analysis was fully specified and saved, the user only needed to pick the new source files and press the start button to automatically have multiple experiments processed from realignment through presentation quality output. |

