Top Topics
Recent Topics
Hello Again,
If I interpret the comment. The question is regarding your own data. The goal of this forum is to answer questions pertaining to data hosted on this site. You might consider posting the question on the Linkedin EEG Signal Processing Group. Most questions received several responses within a few hours.
Best, Dennis
Hi Kisoong, Need a little more information. Would you mind sending me a figure, some details on how you are computing and which subject set you are looking at? You can email me at ddean@research.bwh.harvard.edu. Best, Dennis
Hi Edward,
I scanned Ben Shneiderman's recent work since I haven't worked with tree maps much. I believe you are on to a potentially rich research area. You might want to consider focusing on creating visualizations related to issues associated with the signal processing pipeline. (1) There are currently no good ways to look at the annotation files which are written in an XML format. Being able to access and view the contents could be the start of many research directions. (2) Viewing the signal specifications of a large cohort to look for inconsistencies. We are also beginning some large scale analyses that would benefit for post analysis visualization. I would be happy to discuss the type of analyses that we are doing, if you are interested in post processing/analysis visualizations. Although the visualization could be done from a CSV, I would encourage you to consider working from raw files; possibly with a schema/search driven search component. Accessing and organizing raw sleep output is still some what of a challenge for novice to intermediate programmers.
P.S. Ben Shneiderman is one of my CS idols.
Not exactly. The program requires two types of inputs per subject. The first input is the signal information which is stored in the EDF. The second type of input is the technician scored sleep stages which can be stored in an XML file. Thus, you need an accompanied XML file. I would suggest that you start by looking at the XML file in the SpectralTrainFig release. You will need an XML file with scored sleep stages. I would recommend that you make an XML file with a text editor (not MS Word). On the first step, you want to get the program to get to the point where the XML file is loaded. All of the figures and EXCEL outputs are configured to look for NREM and REM epochs. Getting the program to run all the way through may take a little editing. Email me if there are specific bugs that arise or your not sure how to get the program to run all the way through.
Here are some questions you should direct at your preceptor if you are unclear of the EDF contents: Does the EDF contain sleep and wake data? If the EDF contains sleep data is the scoring data available. If there is sleep data, you will approach editing of the program differently. If NREM and REM doesn't make complete sense, look them up in Wikipedia to start.
Dear Bridgette,
Thank you for posting your question. The run is failing at the data check level. We see these errors during the data check step. My colleague Mike Morical is developing a guide for data checking. For now we need to put on our detective hat when these errors occur. It isn't clear from your post, whether you are using your own data or the test data. Let's continue the conversation by email and then we can post the solution to the forum once you are up an running.
Followup:
Bridgette and I exchanged several emails. She was able to get the spectral program to run on a Windows machine. We have not tested or run the Spectral program on a Macintosh computer. In theory, all of the functionality in the spectral program can work on a Macintosh platform. If you do your analysis on a Macintosh platform and want to update the code. There are two areas that need to be reviewed. The first is file access commands, which in the past differed across platforms. This should be relatively easy to do. The second area is saving the generated figures as a power point files. Currently, a window mechanism is used. There is no reason a more cross platform approach couldn't be used. Please send me a note if you are interested in using the spectral program on a MAC, so we can proceed accordingly.
Kisoong, I found this: Automatic Stage Scoring of Single-Channel Sleep EEG by Using Multiscale Entropy and Autoregressive Models.
Hi Farhad, Are there any published/opensource methods that you believe do a good job of sleep-wake detection?
The software described in Gross et al. can be found online and is freely available. You can download the software here: https://secure.nouvant.com/umich/technology/4582/license/142.
Hi Kisoong, It isn't clear to me what your question is. Annotations are available and we have some tools for accessing annotations. Let me know if you need more information.
Hi Korey,
Welcome to the site. It is very exciting to feel your enthusiasm. I will break down your questions into discrete steps. I would recommend approaching staging systematically. I believe that you could proceed in a way that you can develop some nice tools while generating important research findings.
Sleep Staging Papers. I would recommend that you start by reviewing Gross et al. and Leclercq et al.. I have always wanted to go through the Gross paper in detail. Please respond with your thoughts.
Please let us know how you intend to proceed.
SignalRasterView was developed specifically to view signals and annotations. There are a lot of possibilities. Our plan was always to convert the GUI from generating a figure to writing to a figure in a GUI. With just a few buttons, it would be straight forward to scroll through the signal and see the data.
(1) SignalRasterView and Sleep Stages. I would recommend adding the annotation file as an input. If the input is present, sleep stage information would be added to the signal raster view. I would recommend changing the background color of each epoch by stage.
(2) SignalRasterView and Scoring. Our scorers tell us that they want to look at the data at multiple scales and include other signals. Consider adding a second figure panel. A revised GUI could be added as a 'scoring' configuration option.
I wrote SignalRasterView, so feel free to discuss strategies for adding sleep stage scoring. It may be possible to coordinate effort.
Good Luck!