Development and prospective evaluation of an automated software system for quality control of quantitative 99mTc-MAG3 renal studies

Abstract

Quantitative nuclear renography has numerous potential sources of error. We previously reported the initial development of a computer software module for comprehensively addressing the issue of quality control (QC) in the analysis of radionuclide renal images. The objective of this study was to prospectively test the QC software.

Methods: The QC software works in conjunction with standard quantitative renal image analysis using a renal quantification program. The software saves a text file that summarizes QC findings as possible errors in user-entered values, calculated values that may be unreliable because of the patient's clinical condition, and problems relating to acquisition or processing. To test the QC software, a technologist not involved in software development processed 83 consecutive nontransplant clinical studies. The QC findings of the software were then tabulated. QC events were defined as technical (study descriptors that were out of range or were entered and then changed, unusually sized or positioned regions of interest, or missing frames in the dynamic image set) or clinical (calculated functional values judged to be erroneous or unreliable).

Results: Technical QC events were identified in 36 (43%) of 83 studies. Clinical QC events were identified in 37 (45%) of 83 studies. Specific QC events included starting the camera after the bolus had reached the kidney, dose infiltration, oversubtraction of background activity, and missing frames in the dynamic image set.

Conclusion: QC software has been developed to automatically verify user input, monitor calculation of renal functional parameters, summarize QC findings, and flag potentially unreliable values for the nuclear medicine physician. Incorporation of automated QC features into commercial or local renal software can reduce errors and improve technologist performance and should improve the efficiency and accuracy of image interpretation.

FIGURE 1

(A) QuantEM-II software displays dialog box that asks user to enter patient’s height (cm), weight (kg), and age (y). Technologist entered 178 kg as patient’s weight. (B) After entry, QC software highlights questionable weight and gives user option of changing weight if 178 really referred to pounds rather than kilograms.

FIGURE 2

Algorithm that is followed when values are input by user. Area within black outline represents QC-related steps. There are 3 opportunities for QC event (flag) to be set when values are entered and tested: when value is detected on entry to be outside expected range (trigger flag); when value is changed by user (change flag); and when final, accepted value is outside expected range (out-of-range flag). These are indicated on flowchart by italicized type.

FIGURE 3

Early phase of renogram curve, indicating arrival of radiopharmaceutical bolus in kidneys. (A) In usual case, a fit to upslope of curve is extrapolated to x-axis, and this intercept identifies bolus arrival (vertical line). (B) If camera acquisition was started too late, earliest part of uptake phase will be missing, and bolus arrival cannot be identified with certainty. A default early frame (vertical line) is used in calculations, but accuracy is reduced.

FIGURE 4

Summary of all technical QC events (white bars) and clinical QC events (hatched bars) that were detected. Trigger events (events initially displayed as being outside expected range) are included; these events occur when an unusual value is entered by user. Purpose of these flags is to trigger redisplay of data entry dialog. Presence of trigger may or may not indicate error, because value may be corrected by user or may have been correct as originally entered. If redisplayed value is accepted by technologist, it is recorded as an out-of-range event.

FIGURE 5

Posterior injection site image from patient determined to have significant infiltration (>1% of injected dose; arrows). (A) Image displayed at default window level, which is scaled to counts in kidney. (B) Window level adjusted to show body outline and to better show injection site in right arm.

FIGURE 6

Posterior images from study in which QC event (right-kidney ROI significantly smaller than left-kidney ROI) led to discovery of incorrectly placed left-kidney region. (A) Original region on 2- to 3-min summed image. Little uptake of radiopharmaceutical is seen in left kidney. (B) Nineteen- to 20-min summed image, in which faint outline of left kidney is seen. (C) Two- to 3-min image with redrawn ROI.

FIGURE 7

Example output of QC module. As integrated into existing QuantEM-II software, this display is accessed by selecting a button on main review screen. All QC items are listed. Items that were checked without error are indicated by “ok”, whereas those that represent potential errors are indicated by “check”. Items with neither “ok” nor “check” were not verified by QC module. On actual computer display, “ok” items are listed in green text and “check” items in red text, as an additional visual cue.