Statistical Questions: Background The study consists of 23 patients

advertisement
Statistical Questions:
Background
The study consists of 23 patients that have breast implants. Currently, the dimensional characteristics of
the implants given by the manufacturer are based on the implant being outside the body. However,
once they are inserted into the body, we do not know the dimensions. In this study, 23 patients were
put in to a standing MRI (Magnetic Resonance Imaging) and then measurements were taken of the
implants.
Analysis
The MRI has provided a lot of data that can be correlated to various parts of the breast anatomy, as well
as comparing them to the implant characteristics provided by the manufacturer. The measurements
that were taken are shown in the PowerPoint presentation provided.
I am studying any “correlation” between two sets of data. I do not really any thing about statistics so, I
will probably use the terms like “correlation” incorrectly. However, in simple terms, I would like to know
if there is any relation between the data – for example, is there a linear relationship, is there a
relationship but indefinable, completely independent, etc. So I have suggested what I think is happening
below for each point, however, if you have a better idea or suggestion, I am very open.
The data will be provided in a Microsoft excel (2007) format.
I would like to define some terms from slide (6):
Measured Values from MRI
A – Implant height (these are round implants, so this is the same as the diameter of the implant)
B – Implant projection at 2 cm – this is the amount of implant projection from the chest wall at 2 cm
below the apex of the implant
C – Maximal Thickness (Projection) Position – this is the location (relative to the apex of the implant)
where the maximum amount the implant projects out from the chest wall (There is a column on the
excel spreadsheet that is the “avg Location” it is the average of Point C per implant size
Maximum Projection (MP)
Maximum Width (MW)
Pectoralis Muscle thickness (PMT) – just as it seems – the thickness of the muscle is measured.
D – The amount of projection at 4 cm above the lowest point of the implant
E – The amount the pectoralis muscle extends below the apex of the implant
Measurements Provided by Manufacturer
Manufacturer Labeled Projection (MLP)
Manufacturer Labeled Width (MLW)(same as diameter because the implant is round)
Manufacturer labeled projection position (MLPP) – calculated value based on the fact that the implant is
round, so the point of maximum projection is at the center
Measured Values on patient
Superior Pinch Thickness (SPT) (measured value on patient, not from MRI)
Skin Stretch (SS)
The following is a list of the data that I would like analyzed (This will be provided in a Microsoft Excel
Spreadsheet)
1. Difference (=point B – predicted value of ellipse with diameter of A) of the actual measurements
of the projection at 2 cm from the top edge of the implant to the predicted projection at that
same position and how that correlates with the implant size. (the predicted value is obtained by
the assumption that the implant front part is an ideal ellipse and using the ellipse formula).
Since several patients had similar implants, the data can be averaged for the same implant or
used individually – which ever you think would be better. I am looking for a linear regression
analysis to show that as the implant size increases, the actual projection approaches the
predicted value OR it could show that the difference between the actual and the predicated
regardless of implant size. (Slide 9). The formula used is =SQRT((1-(((J2-2)^2)/J2^2))*D2^2) –
Where J is the column for the radius and D is the column for the MLW in the excel spreadsheet.
If you see a mistake, let me know. Averages were used, see column U in the spreadsheet.
2. Difference (= B – predicted ellipse projection) compared to pectoralis muscle thickness (PMT).
This would also be a correlation. Hopefully linear, but it would be whatever type of relationship
you could find. (see slide10). You may find that there is no correlation or that the difference is
constant.
3. Correlation of implant diameter variance from manufacturer specification (A – MLW) to implant
size. I am trying to show that the diameter is smaller than that reported by the manufacturer. If
there is a better way than just looking at the difference, I am open to that. Slide 11
4. Variance of Projection Manufacturer labeling- (MP – MLP) compared to the implant size. As
the graph shows, the actual projection is higher than that reported and that difference is greater
as the size of the implant get larger. (slide 12). Averages were used, see column T in the
spreadsheet.
5. Impact of Pectoralis Position on Implant projection. E compared to the difference of projection
from Manufacturer Label (MP - MLP). I am trying to see if there is any correlation, I am guessing
no correlation based on the graph. Slide 13
6. Impact of Superior pinch thickness on Implant projection . Same as above, except that now I am
looking to see if there is any correlation between implant projection difference (MP – MLP) and
the Superior Pinch Thickness. I am guessing that there I is no correlation (both are independent)
Slide 14
7. Impact of Skin Stretch to Implant Projection Variance – Same as above, except now looking at
the measured value of the skin stretch to implant projection (MP – MLP). I am guessing that
there is a correlation meaning that as the skin stretch is higher, the projection is farther from
the manufacturer labeled amount. (slide 15)
8. Variance of Maximum Implant Projection Position from predicted maximum implant position to
Implant size. The variance (C – Predicted manufacturer maximum implant projection )
compared to implant size. The predicted maximum projection would be at the center (the
radius length) because the implant is round. Based on graph it seems like the variance gets
larger as the implant size gets larger. Slide 16. Averages were used – see column S in the
spreadsheet
9. Any Corelation between Pont C and Column V (N to IMF distance). The N-IMF (Column V) is not
the n (minus) IMF. But rather just a label. N is “the nipple” and IMF “is the infra-mammary
fold). I am looking to see if there is any correlation between the position of the maximum
projection and the distance on the patient’s breast for where the maximum should be.
10. On sheet 1 column C is the is measured Width (MW), on Sheet 4 Column W is the height
(measured), or Point A. I would like to see if these correspond to a circle (which is what is
predicted to be), or an oval (ellipse).
Please let me know if you have any questions.
Download