| | | |
|
 | |
| |
| Resouces INDEX | |
| Patient Information | |
| Bone Densitometry | |
| Refs. & Reviews | |
| Policies | |
| Grants/Awards | |
| Journal Subscription | |
| ASM Resources | |
| Links | |
 | | | | | ||||||||||
| | |
APPENDIX 6- Construction of a Cusum plot CUSUM plots are optional and may be difficult to interpret. This is a method for the sensitive detection of systematic drift in a measured QC parameter, and may be particularly useful for monitoring of long-term (ie over several years) "constant" QC parameters such as the BMD of a phantom. Note that random errors (e.g. a deterioration in precision due to failure of electrical circuitry, or wear in a mechanical component) may not be detected by a CUSUM chart. The use of CUSUM (as distinct from, say, a Moving Average method) is largely a matter of personal preference and prior experience. Prior to deriving a CUSUM chart a "Target" value is required. The mean value of the first 10 phantom scans can be used as the target value, denoted as T. Let Xp denote the phantom BMD at time p. The deviation (d) of this value from the target value is calculated as d = X-T. At any time (t ), the cumulative sum of deviations (CUSUM or St ) is calculated by summing all the prior values of d up to the current measurement. This is expressed mathematically as : St =  p
<= t (Xp-T)The CUSUM (St ) is plotted against time to provide a cumulative sum chart. A suitable vertical scale is one unit per 0.5 gm/cm2 and a suitable horizontal scale is one unit per 30 days. Since any variations of the phantom BMD from the target value T, shall be randomly positive or negative, the CUSUM shall not deviate significantly from 0. A deviation of the chart away from 0 indicates a potential instrument problem. This will occur if a systematic error in BMD occurs.
| ||||||||||||
|  |
| HOME | ABOUT
US | MEMBERSHIP | NEWS | EVENTS | RESOURCES | CONTACT
US Content copyright 2005 © Australian & New Zealand Bone & Mineral Society Design copyright 2005 © Anne Kovach, eRadiate | |
-- This page last edited: 06 Dec 2002 -- |
