MIM Safety Recall Notification - SUV Statistics Calculation

MIMvista Corp has issued this safety recall notification for MIM versions 4.0.2 through 4.1.2. An error has been discovered in the way SUV and contour statistics are measured and reported under some specific circumstances. Users of these versions should immediately contact MIMvista Support at (866) 421-2536 or visit http://www.mimvista.com/recall/ to obtain instructions to work around this error.

A new version that corrects this problem is available on the Downloads page.

Does this recall apply to me?

The recall applies to you if you are using MIM versions 4.0.2 through 4.1.2. Most MIM users are currently using versions 4.0.1 or lower, which are not affected by this problem. The affected versions never went into wide release, and were only released to beta testers, customers who requested them, or to customers who needed them in order to take advantage of the enhanced contouring tools and workgroup networking features available in these versions. You can locate the version number by opening MIM and looking in the upper left hand corner.

Not all users of the affected versions will experience the problems that are prompting the recall, but it is nevertheless important to update these versions as a preventative measure.

When can this problem occur?

It is possible for this problem to occur when ALL of the following are true:

• You are using MIM 4.0.2, 4.0.3, 4.0.4, 4.0.5, 4.1.0, 4.1.1, or 4.1.2.
• You are fusing a PET series with another modality such as a CT or MR.
• The Z-Axis slice separation in the primary volume (CT or MR) is larger than that of the secondary (PET).
  • In other words, you have a hybrid PET/CT with significantly less CT slices than PET slices when obtained over the same scan range (e.g. 154 CT slices to 203 PET slices), then the Z-Axis separation is larger on the CT. If the difference is only a couple slices, than the problem does not affect you (205 CT slices to 204 PET slices).
• You are using either the SUV tool or Contouring tools to measure statistics.

Is there a workaround to this problem with my current version?

Yes. Close the fusion row, and then calculate SUV statistics on the PET row.

What happens when the error occurs?

If a PET series is fused with a CT or MR series which has a larger Z-Axis slice separation than the PET series (less CT slices than PET slices for a hybrid PET/CT scan), data in some of the PET slices will be ignored when calculating statistics such as Maximum SUV and Mean SUV.

Example: a PET/CT with 5mm CT slices and a PET with 4mm slices. In this example, every 4 slices of the CT will correspond with 5 slices of PET. The SUV tool and contouring tools are based on the CT's resolution, and the problem becomes that 1 out of 5 PET slices will be ignored when calculating statistics.

What types of scanners does this problem occur with most often?

Philips scanners (all models) typically will produce PET/CT scans with less CT slices than PET slices, therefore possibly affected by the problem. Siemens/GE scanners typically produce PET/CT scans with equal numbers of slices, and therefore are not generally affected, except when a protocol is used that acquires the CT at a lower Z-Axis resolution then the PET.

What if I don't have a PET/CT, does this affect me?

If you use the SUV or contour statistics on images fused with CT or MR images from a different scanner, this problem is likely to affect you because some CT or MR scans may have Z-Axis slice separation larger than the secondary image.

If I used versions 4.0.2 through 4.1.2, is every report I dictated incorrect?

There is a possibility, but even if you used PET/CT scans that are affected by this problem most reports which include statistics will still be diagnostically viable. We are not trying to minimize the effect of this problem, as certain tumors (especially smaller tumors <5-8ml) where the hottest SUV fell on an ignored slice will report lower SUV values than if this problem had not occurred. However, most average SUVs reported by MIM will be very close to the true average, and the Max would have to fall on the missed slice (1 out of 5 slices, for example). Because of the finite resolution of PET, slices near the Max will likely have an SUV close to that Max.

In other words, if your scans are affected by this problem, the average SUV value will be slightly to moderately lower, and the Max generally has a 4 in 5 chance, for example, of being correct. In the chance that the slice with the Max was skipped, there will likely be an SUV close to the max that will be reported. There are some cases, especially with small tumors, where the SUVs reported will be significantly different than the true values.

If you want to check previous scans to compare whether or not you were affected by this problem, we recommend you go to scans with small lesions where you thought the SUV value reported was borderline, and recalculate the statistics with the fusion closed.

This error does not affect the display of PET or Fusion images. Voxels with higher glycolytic activity will still be rendered correctly on screen, allowing for accurate qualitative judgment of significance. The error only affects the statistics reported by the SUV Tool or Contour Statistics.

When was this error introduced?

The error was introduced in MIM 4.0.2, which was released to limited audiences in September, 2007. The error is present in version 4.0.3, 4.0.4, 4.0.5, 4.1.0, 4.1.1, and 4.1.2 as well.

When was this error discovered?

The error was discovered on Friday, April 18th, 2008, when a customer called in to report unusual behavior in MIM's SUV measurement tool. MIMvista's support staff was able to reproduce the error and MIMvista's engineering staff was notified immediately. MIMvista engineers found the source of the error within a day of testing and were able to identify the circumstances in which the error can occur.

Why was this not discovered until now?

The error only occurs in certain specific circumstances, and even in those circumstances it can be difficult to detect.

What is MIMvista Corp. doing about this problem?

• We have notified the FDA of the problem.
• We are in the process of notifying all customers that our records show as using the affected versions by phone, email, or postal mail.
• We have specified a workaround (see above) for current affected customers.
• We have produced a newer version of MIM which fixes the SUV statistics problem.