Account Registration
Account Registration interface can be seen in Figure 1.1. A basic registration process that require the user to input their first and last name.
The email address provided by the user will also be used as the Sign In credential (Username). All passwords are hashed before storage to ensure securtiy.
Account registered by the user will always default to "Research" account, which will remove patient identifications before display in any part of the webpage.
Promotion of a "Research" account to "Clinician" account can be performed by Admin (the local admin at your institute that setup the platform).
Clinician account will gain access to not only the patient identifiers, but also access to all data uploaded by other "Clinician" account within the same institute.
On contrary, "Research" account's upload will not be added to the "Clinician" account nor any other "Research" account, everything will be specific to your account.
As a demo web application, no verification will be used on the registration email. Feel free to use made-up email address to register.
In addition, no "Clinician" nor "Admin" accounts are created for the demo application.
Account Sign-In
A user can sign in to their account through the Sign-In interface as shown in Figure 1.2. Session Cookies are used to store your credentials and further verification within the Web Application.
If the session idle for more than 60 minutes, the user will be required to login again before continue using the application.
De-Identified Patients (Research-Only)
In deidentied patient table, most contents are leave as blank unless labeled by the user.
In order to create a new deidentified patient in the table, follow the procedure in Figure 2.3.
After clicking "Add Patient" in Red Box 1, a pop-up dialog will be shown that allow user to enter the Patient Deidentification Information.
Among "Patient Identifier", "Patient Project Name", and "Diagnosis", only "Diagnosis" can be leave as blank.
Device information can all be left as blank. A random string will be generated as the device deidentified ID.
User may drag-and-drop session files to the upload area, or click the upload area to bring up file selector.
Multiple files can be selected at once and upload at once.
Once clicking "Create" in Red Box 2, a new row will be insert to the deidentified patient table.
If this patient has multiple device, follow instruction in Upload JSON Files (Research Only) to add new devices.
Identified Patient Table (Clinician-Only)
In the clinician account view, the patient table will be shown with identifiers.
Within clinician view, we eliminate the process to manually create patient from the table.
In contrast, all information are automatically populated when user uploads identified JSON files exported from Percept Neurostimulator.
The primary health information extracted are based on 1) Patient First and Last Name, and 2) Device Serial Number.
Data aggregation is based primarily on Device Serial Number, and Patient Identifiers are used to determine if multiple devices belong to the same patient or not.
Session Report (Beta)
Session Report is a function available to both Research view and Clinician view.
This function is accessible through the "View Single Session Report" button at the top of Patient Table.
This function intend to generate a brief report from a single Session JSON file uploaded by the user without aggregating past reports nor saving the reports for future reference.
IMPORTANT
This function DOES NOT DEIDENTIFY the session JSON file for you, and the resulting report page will contain idnetifiable information.
This function is primarily designed for clinicians who want to see a more detailed report than available on Medtronic's Report page.
Currently in Beta version. The information displayed includes 1) Basic patient identifier obtained from the uploaded JSON file, 2) Device Information and lead configurations, and 3) detail therapy configurations.
The platform will also calculate percent time in each group over duration of past visit.
Since identifers are present in Session Report, we take multiple measures to ensure as much security as we can provide.
First, the JSON session file will be immediately deleted from the server once the report is generated.
Second, the report page will only live in your browser for single-use, and user cannot access this page anymore.
However, despite all this, the user must comply with their institute's PHI policy and potentially edit the Session JSON files with modified PHI so the uploaded file doesn't contain real PHI.
Edit Patient Information
User can edit the patient information through "Edit Information" in Red Box 1 of Figure 3.2.
A pop-up dialog will be shown to user with existing patient information.
Edit the desire fields in Red Box 2 then click "Update" will prompt a database update.
The resulting edit can be seen in the post-update view in Red Box 3.
Upload JSON Data (Research-Only)
Here is where the Research account should upload their data.
Unlike Clinician account, the Research account is assumed to be working with deidentified files.
That means the PHI used to group uploaded JSON into respective Patient ID or Device ID will not be present in the uploaded JSON files.
This additional process required by the Research account ensure we can properly manage the data and organize them in correct group.
After initial deidentified patient creation in Deidentified Patient Table , the patient overview will be shown without any associated device.
The user may manually add a Percept PC device via "Add Device" in Red Box 1 of Figure 3.3.
The fields can all be leave as blank, a fake device ID will be generated in place with most information unavailable to the user.
Once generated, look at the device table and you will see "Upload" in Red Box 3 of Figure 3.3.
The user then can upload one or more files associate with that device.
If a patient is using bilateral Percept Device, the user should create a second blank device and upload files separately.
Primary Analysis Navigations
Delete Devices or Delete Patient
Patient entry and device entry can be manually deleted by the following procedure.
It is very unlikley for a clinician account to perform such task, nevertheless this is available to both Clinician Account and Research Account.
Clicking either Edit Device or Edit Patient Information, a pop-up window will be shown with a "Delete" button.
Clicking "Delete" will prompt user for confirming if they want to delete the device and all associated data from the server.
Once confirmed, data will be removed and the page will be refreshed to reflect new information.
If patient entry is deleted, the page will automatically redirect back to Patient Table view.
If user want to delete individual session file, they may use "Sessions Management" tab shown in Figure 3.4c.
Therapy Change Log
Therapy Change Log is a trend generated from Medtronic Session file's DiagnosticData.EventLogs JSON Field.
A typical Therapy Change Log looks somewhat like the following code snipet:
{
"DateTime": "2021-10-25T22:35:01Z",
"ParameterTrendId": "ParameterTrendIdDef.ActiveGroup",
"NewGroupId": "GroupIdDef.GROUP_B",
"OldGroupId": "GroupIdDef.GROUP_D"
}
The datetime field indicate the time of group changes, based on UTC timezone and not patient's local timezone.
In our platform, all time are presented as the user's local timezone.
Past Therapy Configuration
The therapy configuration extracted from GroupHistory typically contains all information about the stimulation.
For example, in Figure 4.2 we present a typical therapy configurations for a patient before their session on July 19th, 2021.
The reason that these configurations are "Before" certain date, because the GroupHistory JSON field's time indicate the date when the setting was changed.
A typical configuration contain up to 4 Groups (A,B,C,D). Each group may contain Left/Right hemisphere, each each hemisphere may contain up to 2 program groups.
Therefore, each group may have up to 4 rows.
Right after the Group ID is a calculated percent usage based on Therapy Change Log.
Warning
BrainSense may show 0.0Hz, because GroupHistory doesn't always maintain good storage of the BrainSense Frequency.
Power Spectrum across Channels
Power Spectrum are calculated with Welch's Periodogram method.
The Session JSON files divide one survey into multiple simultaneous recordings of different channels.
We organize them by performing clustering of timestamp. Recordings perform close to each other are shown side by side for comparison.
The survey are interactive and user may selectively display channels to display.
Figure 5.2 and 5.3 demonstrate how Segmented Survey can be better compared when dividing into 3 segmented groups.
Figure 5.4 also shows the power of Interactive Plot capability, which allow hover display and custom zoom and pan.
A dropdown menu is shown at the top of the page. The user may choose which Survey group to view.
Left and right hemisphere are shown in different figure. Different channels are colored differently.
Power Spectrum across Time
BrainSense Survey across time is an analysis perform across time.
It present all surveys recorded on the same channel organized by order of acquisition, colored by gradient of colormap.
A dropdown menu is presented at top-right of the figure block. User may choose which channel to view (Figure 5.6).
This allow user to visually identify disappearance and emergence of certain brain signals.
For example, in Figure 5.5 we can see that an changes in recorded signal between August to September 2021.
Select Recording to View
Similar to BrainSense Survey, BrainSense Streaming data are aggregated for all patients.
Recordings are organized by date of collection. The platform will also attempt to merge simultaneous Left/Right hemisphere recordings into one recording if detected.
An example of the selection table is shown in Figure 6.2.
As shown, if recording contain both Left and Right hemisphere, the table will display information for both in one single row.
The table provide essential information regarding the recording, such as recording duration and therapy configurations.
Warning
The only information require manual update is Stimulation Mode (Ring Stimulation vs Segmented A, B, C) because Percept Session file does not store those information in the recording data.
Neural Recording Summary
Once a user selected "View" in Selection Table, data will be processed in the server and transmitted to the web application for display.
The display for a typical bilateral recording in shown in Figure 6.4 and unilateral in Figure 6.3.
All figures provided in the graph are interactive with x-axis alignment fixed. All time are presented based on user's local timezone.
Time alignemnt with bilataral recording can be easily identified via Red Box 3 in Figure 6.4.
The presence of pathological beta activity is supressed unilaterally when unilateral stimulation is turned on for Left and Right separately.
The alignment shows that the stimulation artifact align with changes in stimulation parameters.
User may choose to export the raw data via Red Box 1 in Figure 6.4. The export will generate a CSV file easily loaded in any scientific programming languages.
The data are aligned if left and right hemisphere both present in the recording. Timestamp are provided as UTC timestamp in seconds.
Aligned stimulation values are provided for identification of stimulation period.
The basic summary uses default short-time Fourier Transform (Spectrogram) method to generate Time-Frequency Analysis.
However, user can also choose to use Wavelet Transformation (usually more time-consuming) via Red Box 2.
Once method is changed, the processed data are cached on the server and available to user in the future.
Similarly, user may also choose to use a template matching cardiac filter to remove cardiac artifacts if present.
Figure 6.6 and Figure 6.7 shows the performance of the cardiac filter. It selectively remove signal without altering stimulation artifact spikes.
Effect of Stimulation
In the effect of stimulation presentation, the platform will automatically segment period with different level of stimulation and calculate average power spectrum for different stimulation amplitudes.
The segments are sorted with increasing amplitude and color gradient indicate a changes of brain signal with increasing stimulation (Figure 6.8).
The user may also toggle the switch on top-right of the figure to view segmentation based on contralateral stimulation.
For example, this is a bilateral recording with only left hemisphere stimulation. Figure 6.8 shows no changes in stimulation amplitude for right hemisphere.
But user may toggle the "Contralateral Reference" switch and calculate power spectrum segmentation based on contralateral stimulation.
This may be useful to understand effect of stimulation on contralateral hemisphere.
Select Recording to View
The recording selection is performed through toggle buttons. Each button indicate the time and duration of the recording.
After recordings are selected, user can retrieve the data from server, and simple display will be used to allow interactive visualization of Indefinite Streaming data (Figure 6.1).
Since there are no extra label provided by neurostimulator. User may use external label such as biosensors or questionaires to indicate events.
Data can be exported similar to BrainSense Streaming.
Multiple Selection
The toggle selection actually allow multiple selection.
User can select multiple recording from the same day and visualize them on the same time-axis.
Segment without data will be leave as blank.
Figure 7.2 shows an example of recording separated by one hour but presented on the same axis.
Figure is edited to highlight the major difference from Figure 7.1.
Chronic LFPs Summary
Figure 8.1a shows a typical aggregated chronic LFP recordings over multiple months. Patient Events are overlayed on the ChronicLFP trend at corresponding time.
As mentioned above, different therapeutic group or sensing setting can lead to different interpretation of the signal.
In this example, the initial segments of recordings are collected with 67.38Hz as the primary BrainSense frequency, which has significantly lower power than the following recording that collect the 24.41Hz power.
The exact therapy configurations can be identified using cursor hover.
Circadian Rhythms
Circadian Rhythm is one of the additional processing examples for Chronic LFP recordings (Figure 8.1b, top-left).
The circadian rhythm calculation divide all Chronic LFP samples based on therapy settings and sensing settings, then calculated 24-hour trend of brain signal.
This graph demonstrates changes in brain signal between awake state and sleep state.
In addition, consistent medication cycles will also show up on the graph.
Event-locked Power Trend
Event-locked Power Trend (Figure 8.1b, top-right) allow user to visualize power 3 hours before and after onset of an event.
This is especially helpful for understanding changes in power with respect to medications or symptoms.
Event Power Spectrum
Patient Events that contains PSDs will be averaged within group and compare to other events (Figure 8.1b, bottom-left).
The shaded area is one standard-error from mean.
Number of sample is usually different from event-locked power trend because not every recorded event contain PSD snapshot.
