Archive for the ‘Hardware’ Category

UHP Upgrade Update (Continued)

November 8th, 2020

1. Introduction

Good news — after a lot of hard work over the weekend by the CNI staff, we are most of the way back to normal operation with our upgrade to GE’s RX28 release.  The significant accomplishments over the weekend include solving the issues with the research SMS sequences as well as the spiral fieldmap sequence.  These are all operational again and the new sequences are fully integrated with Flywheel.

There are a few minor changes to the user interface with this revision of the scanner operating system and we describe these below.

2. Coil Selection

The tab for receiver coil selection has changed in RX28. Instead of a list of connected coils and checkmarks for indicating which coils are selected to be used, the new interface shows each port of the coil connection and highlights which ports are active. By default the Nova 32ch Head coil will have both Port P1 and Port P2 selected (as shown in the screenshot below).

3. Product Hyperband Updates

The Hyperband sequences on RX28 now have built-in calibration scans. The calibration can be integrated into the prescan of each Hyperband scan, so there is no need to do a separate ASSET Calibration before the Hyperband scans. The integrated calibration scan is turned on by default (as shown in the screenshot). If it is turned off, a separate calibration scan (named “Cal: [original series description]“) will be created automatically once the Scan button is pressed.  The Hyperband BOLD sequence also no longer requires an even number of slices / band to be acquired so it will improve the efficiency of some prescriptions over that in RX27.  In addition, the slice timing is correctly calculated and stored in the DICOMs. We will work with the developer of the dcm2niix utility to extract these timings and save them to the accompanying JSON file – a fix was needed because the developer had disabled this feature for GE Hyperband DICOMs given the erroneous timing values that were present for versions earlier than RX28.

RX28 UI Hyperband

4. High-Order Shim

GE has switched from a spiral-based acquisition for their high-order shim (HOS) process to a 3D fast gradient echo acquisition.  Users will need to replace their existing HOS series in their protocols with the new HOS template.  This can be accessed by first choosing Add Task, then selecting the GE protocol library, choose Adult->Other in the lefthand tabs and then select the hosGE entry.  We suggest using the GE HOS FOV28 template as this will most likely include all of your subject’s brain while still providing reasonable resolution. This new 3D FGRE approach may have modest benefits when subjects have significant B0 inhomogeneity due to orthodontics or large sinuses.

5. Remaining Tasks

As mentioned in our previous post, we will try to get the new coil configuration format for the NOVA16 research coil next week.  We are also working with Gary Glover on getting his spiral in/out (sprlio) pulse sequence installed.  The “show” system in the corner of the CNI scan room near the large TV screen is still running RX27 but we hope to update to RX28 in the next couple of weeks.

Thanks for your patience as we worked through this upgrade. We expect future s/w upgrades to be simpler now that the UHP is more closely integrated into GE’s standard product software.  Please don’t hesitate to contact us if you have any questions when using the new system.

Thanks and stay safe!

The CNI Team 

CNI Technology Update

October 22nd, 2020

The software controlling the new UHP system will be significantly upgraded in early November.  This post describes the system upgrade.

We also describe the valuable features of a new display we recently installed in the control room.

1. MR Software Upgrade

Our 3T Ultra-High Performance system is scheduled for a host computer and operating system upgrade over Nov. 2 – 4.  GE will be installing what is known as the RX28R04 release.

This upgrade is fully integrated in GE’s commercial release software.  The current version of the operating system is a special-purpose patch that was developed for our unique hardware.  The patched system had some restrictions.  The host computer will also be updated so we expect some modest increase in system speed and user interface responsiveness.

Here are two changes.

  • Some users have noticed that the timezone on our system is CDT — something we haven’t been able to change given the nature of the patch OS we’re running currently.  This will be fixed in the upgrade.
  • A behind-the-scenes feature of the Hyperband product that is currently running is that it restricts the number of slices per band to be an odd number — this means if a user specifies a Hyperband factor of 6x and a total number of slices being 60, the system will “round up” and acquire a total of 66 slices (6 of which are discarded) in order to keep the number of slices per band to 11 — an odd number.  This requirement will be removed in the new upgrade allowing odd or even number of slices per band.  If you’re interested there is more detail on our Slack channel regarding the why’s and wherefore’s of this issue — and may be an incentive to also join Slack if you’re not already a member (see our post If your Hyperband protocol already had an odd number of slices per band then this upgrade will not have any significant impact. 

There will be additional new features, and we will provide more information regarding this upgrade closer to the installation date.  We plan to provide time on Nov. 4 for users to have some time to ensure their restored protocols work as expected. More on the new system features in a future blog.

2. New TV in CNI Control Room

We have installed a new Sony Bravia X900H display in the scan control room.  This display can correctly receive the signal from the scanner host computer.  This enables us to direct the video matrix switch input from the Scanner Host (#1) to the TV Wall monitor in order to replicate the host display for other researchers in your group if desired.  We will be using this system for training, as it will allow for greater distancing between researchers.
Thanks and stay safe!

The CNI Team

CNI upgrade planning

July 2nd, 2019

Planning for an MRI scanner upgrade at the CNI

The MRI scanner we have (GE DVMR 750) is about 8 years old. There have been advances over the years that we incorporated through compute infrastructure and software. It is now time for us to consider the opportunity for a hardware upgrade.

In anticipation of the hardware upgrade, we participated in GE’s ‘Evergreen’ program. This is a technology non-obsolescence agreement.  The path we were on was to upgrade to the GE Premier 3.0T, a system that you can see in Lucas as 3T2.  That system has some computational upgrades, more receiver channels, and a wider bore.

A second option has emerged that the CNI staff believes is preferable.   GE is marketing this upgrade as the DVMR 750 Connectome plus Edition*.  The Connectome plus includes most of the computational elements of the Premier.   In addition the upgrade replaces the gradient coils and drivers with a new system that has better performance in two ways. The coils generate a higher gradient (50 mT/m for the DVMR 750 and 100 mT/m for the Connectome plus), and the coils are more stable with respect to heating.

The CNI team thinks that the Connectome plus will be a better technology choice for our community. We also believe that the Connectome plus upgrade will help PIs justify the equipment infrastructure when they write grants, enabling them to correctly claim that the GE equipment is competitive with the best Siemens products.

The same decision is being faced by our colleagues in other departments; it is our understanding that other groups also are planning to install Connectome plus systems.

Hardware down time planning

We are planning to be closed for data acquisition from January 6, 2020 to the end of February, about six months hence.  Stanford’s financial rules and the contract with GE leaves us with very little flexibility in when the upgrade can take place.  We will try hard to get the system online by February 21, 2020 for protocol testing.  Please remember that this is a major install and there is always some uncertainty about timing.

You may remember that this timing differs from what we anticipated during our community meeting last fall.  At that time we hoped to do the upgrade during the winter closure.  That turns out to be impossible because, well, people want to visit with their families.  We fully understand and hope you do, too.

During the next six months we will help labs plan for the down time and transition to the new system. We will send additional messages that describe how we will collaborate with the Lucas Center to make it possible for some studies to be carried out there; and, we  will collaborate with GE to quantify the system performance to help understand how to best extend longitudinal studies.

* Some people refer to this system as the Ultra High Performance (UHP) system


2017 Recap and Upcoming at CNI

January 12th, 2018

As we embark on a new year (Happy New Year!), the CNI team would like to take a moment to highlight some of the things that happened last year and to preview some of the things we’re working on for 2018.

Looking back at 2017

Personnel Changes

In May we said farewell to Bob Dougherty, who ventured off to a new life at a startup company here in the Valley. Bob remains close to the CNI and at the risk of being sappy, we’ll just say we miss him very much.

During his tenure at the CNI Bob had developed great working relationships with others on campus. For this reason, we were both confident and excited to welcome two new additions to the CNI team, both of whom were very active in CNI work. The first is Adam Kerr, who is the new Research Director at the CNI. Hopefully many of you have had a chance to meet and talk with Adam. Adam is a Stanford alum (PhD, EE) and a wonderful addition to the CNI staff.

We were also happy that Michael Perry could increase his efforts at the CNI to become our new IT Director. Michael fully supports the extensive compute infrastructure at the CNI, including the NIMS data management system. Michael is a Berkeley product, but he is excellent in many ways that compensate for this one flaw.

Policy updates

The Advisory Board made some adjustments to the scanning policy at the end of 2017. First, there was a change in how after-hours time (10pm – 7am) can be scheduled. Groups must first meet with Laima to address the increased risks of scanning after-hours.

Second, there was a change to how protocol development time can be used. There is now a cap of 10 hours per grant and the protocol development time for seed grants starting in FY18 will no longer be available.

Please take the time to read through all the scanning policies that are available on the Getting Started page of the WIKI.

DV26 upgrade

After weeks of testing, patching, and more testing, we upgraded the scanner software to DV26 on Monday, Nov 12, 2017. This brought along with it some small changes in operational protocol (e.g. SAR limit settings, spectroscopy prescriptions). For more information about the upgrade, please see this post.

EEG Refresh

This past year we sent out our EEG equipment for service. The EEG caps were refurbished and the software was upgraded to the current release. Please contact the CNI staff for information on how to get started if your group is interested in using the EEG equipment in your study.

Penetration Panels

We had two new 10” x 12” penetration panels installed in the scanner wall common to the equipment room. These penetration panels are used to support the introduction of new instrument connections into the scan room and were a critically required expansion as we had already exhausted all of our research penetration panel space.

Looking forward to 2018

Here are just a few things that we are planning for this next year.

New laptops and E-Prime upgrade

We have purchased new Windows and MAC laptops to support stimulus delivery at the CNI. In the upcoming weeks we will retire some of the old laptops. To help make this transition smooth, please migrate your scripts to the new workstations as soon as possible. Please get in touch with the CNI staff if you have concerns or issues transferring your scripts.

In addition to the workstations, we purchased E-Prime 3 upgrade licenses for our laptops. This version of E-Prime will be installed alongside existing versions, so code requiring older versions will continue to function.

Fee structure updates

Starting March 1, 2018 scan time will increase slightly. Fees will be as follows:

Time SlotHourly Rate

Peak: 8:00 AM to 6:00 PM weekdays $425
Off-hours: 6:00 PM to Midnight weekdays. 8:00 AM to Midnight weekends $325
Owl: Midnight to 8:00 AM weekdays. Midnight to 8:00 AM weekends $100
EEG: Anytime usage $50

CNI Innovation Seed Grants

This year we are again in the fortunate position of being able to support innovation. We now welcome applications for funds to purchase scan time at the CNI. Innovative projects that require up to 30 hours of scanner time for data acquisition will be considered. Note that these funds can only be used for scan time at the CNI. The deadline for CNI Innovation grant applications is Wednesday, January 31st, 2018; award announcements will be made in early February. You can apply for an award using the Google Form here.

NIMS transition to Flywheel

In October of last year we started mirroring NIMS on a new data management system, Flywheel. Data collected on or after October 1, 2017 are already in the Flywheel system, in addition to being in NIMS. This has allowed a few groups to begin testing this next generation system. While NIMS will continue to operate as usual for the time-being, we expect to transition to Flywheel during this year, with new data going only to Flywheel by the Summer 2018.

Flywheel is running on Google Cloud, which allows us to expand data storage without purchasing and installing new hardware. We also save costs of system management (e.g., backups, disk replacements, operating system upgrades). The hardware that we purchased in the past, and which is running out of warranty, will be maintained for the time being. But we expect that it will start to fail over the next few years and be replaced by Cloud systems.

The Flywheel system originated from NIMS, but it has a large number of features that make it feel different and also enable certain new capabilities. We believe the data management features include everything in NIMS but extend them significantly. Most importantly the Flywheel system will integrate tightly with scalable Cloud Computing. We anticipate that this system will support reproducible research and data sharing.

Expect to hear a lot from our team about the new Flywheel system during the upcoming weeks. We will offer group and individual training sessions that cover a number of topics, including:

  • Data Migration Efforts
  • DV26 Impact on reconstruction and data capture in Flywheel
  • Flywheel Center Edition training
  • Flywheel Lab Edition computational methods
  • User rights management training and demonstrations

If you or members of your group would like to be part of the pilot program with Flywheel, please contact Michael Perry, who is leading this effort.

See you around the magnet!

- The CNI Team

CNI scanner upgrade to DV26

November 9th, 2017

After weeks of testing, the CNI will upgrade the scanner’s software to DV26 on Monday, Nov 12, 2017. All existing protocols will be migrated to the new platform, and we fully expect they will perform as they do on the current platform. Data will continue to be stored in NIMS as they are now as we continue to make progress on migrating NIMS to our new data management platform, Flywheel (more on that soon). Please contact us if you notice any issues with your protocol or data.

The DV26 interface will feel a whole lot like the current platform (DV25), so we don’t expect any of you to be disoriented when we switch. There are, however, some very minor differences that we want you to be aware of.

SAR limit settings. What follows are step-by-step instructions for setting the SAR limit in DV26 — note that these changes don’t affect most sequences:

  1. When starting an exam, in the dialog window for selecting operating mode the first level dB/dt and SAR limit are selected by default: you can click the “Accept” button directly to enter the exam unless you want to change the operating mode.
  2. By default the system will automatically run a “SAR Scout” scan for SAR limit calculation before the localizer in each exam. This scan will show up in the list of Task as series 0. The Scout is 6s long and doesn’t generate images.

Spectroscopy. If you are using spectroscopy sequences, please refer to the CNI wiki for a description of how to prescribe a rotated voxel acquisition. GE has committed to working on a solution which does not require this workaround, but we don’t yet have an expected date for when this will be available.

SMS BOLD and DTI. Our current SMS BOLD and DTI sequences (muxarcepi and muxepi2) will be available, and they will still work as in DV25. While the DV26 software supports the product SMS DTI sequence known as Hyperband DTI, we recommend users to continue using our research sequences until we have had the opportunity to fully evaluate the Hyperband DTI sequence.


Planning for hardware upgrades

May 23rd, 2016

The CNI is planning for a significant equipment upgrade, which we expect to be available in about a year. This message describes our plans, which are rather uncertain. Despite this uncertainty, we thought it best to write this note to give you plenty of time to plan.

About a year ago, Siemens released a new scanner (Prisma) that has particularly strong gradients. Many NIH grants are now being written with the specification of matching the Prisma performance. Bob has worked hard at the CNI to implement the sequences and reach the quality of images that are required for these grants. The CNI team have documented the SNR of the data, and we think it is pretty good and should make us eligible for these NIH awards.

GE is responding to the competition. In particular, they have told us that they expect make available a gradient system upgrade within the next year. The stronger gradients will make our scanner competitive with (GE says better than) the Siemens Prisma.

We are funding this upgrade through a GE non-obsolescence service contract that costs about $100k per year for the next three years. The funds for this service contract are coming from user fees (which have barely gone up over the 5 years of operation). We can afford this because of the heavy usage at the CNI and our efforts to keep other costs in check. An additional $200k was needed to bring our system up to the current GE hardware platform, and this funding was provided by the Dean of Research. This upgrade was implemented last year and involved a faster host computer and reconstruction server. Some of you may have noticed this upgrade in the form of reduced delays between multiband scans. (It did not affect the data acquisition systems, so it has no impact on data quality.)

We will keep you informed as we learn about the likely delivery dates for the scanner upgrade.  We expect the new hardware to be available in about a year, and the system will likely be down for about 2 weeks during the installation process. The new gradients can be run at the current system’s lower performance level to maintain compatibility with data collected pre-upgrade. Diffusion sequences and a few others will benefit significantly from the new gradients (not such much for fMRI).

Feel free to post questions about the upgrade here. We will let you know more as we learn more.

Brian and Bob