| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, Surgical Radiology Section and Osteoradiology
Section, University of Vienna, Waehringer Guertel 18-20, Vienna A-1090,
Austria.
2 Department of Orthopaedics, LKH Stolzalpe, Stolzalpe, Austria.
3 First Orthopaedic Department, Orthopaedic Hospital Vienna-Speising, Vienna,
Austria.
4 Corporate Clinical Operations – Biometrics Europe, Schering AG, Berlin,
Germany.
5 Institute of CT and MRI Diagnostics, Schillerpark, Linz, Austria.
Received October 30, 2003;
accepted after revision December 12, 2003.
Address correspondence to M. E. Mayerhoefer
(marius_mayerhoefer{at}aon.at).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Ten patients with bone marrow edema of the knee were included in the study. Coronal STIR images of the affected knees were obtained using a 1.0-T MR scanner. Size and signal intensity of the bone marrow edema were assessed on the basis of gray-scale value analysis and calculation of a threshold value for differentiating normal and edematous bone marrow. All measurements were carried out three times for statistical analysis.
RESULTS. The intraobserver coefficient of variation was 0.89% for the volume and 0.94% for the signal intensity of the bone marrow edema, showing the small impact of manual interference on results produced with this method.
CONCLUSION. A computer-assisted method for quantification of bone marrow edema has been described. Intraobserver variation was very low, indicating excellent reproducibility of results. Although the method is too time-consuming for clinical use, it is recommended for research purposes.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This is mainly because, unlike X rays used in CT, the MR signal is sensitive to minimal changes of position of the examined object within the coil, leading to differences in the Q-factor caused by differences in filling factor and electric load, as well as to differences in radiofrequency homogeneity over the sample. Consequently, differences arise in the reference pulse for radiofrequency excitation and the receiver gain and lead to differences in MR signal intensities. Thus, even when using the same MR scanner with identical parameters to examine one patient twice, a slightly different distribution of gray-scale values will be visible in the two image series.
Manual measuring and segmenting of tissue abnormalities on MR images is widely used and might be an appropriate solution for diseases such as osteonecrosis [1], in which a clear demarcation of abnormal tissue is visible. When a gradual transition exists between normal and abnormal areas, however, this method is prone to high inter- and intraobserver variability, making reproducible segmentation and further analysis of the tissue abnormalities nearly impossible.
One model of a lesion with such a gradual transition is bone marrow edema. Bone marrow edema is associated with a variety of disorders, including early stages of osteonecrosis, bone marrow edema syndrome, bone bruise and microfractures, stress bone marrow edema and stress fractures, arthritis, osteoarthritis, and tumors [2].
It was our goal to develop a reproducible image processing method to precisely segment the bone marrow edema and measure its volume and its signal intensity in a way that would enable us to monitor the progression of bone marrow edema and its response to treatment.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Threshold Calculation
The goal of this step was to create a basis for segmenting healthy bone
marrow from bone marrow with abnormal signal intensity through the gray-scale
value distribution. For every examination of a patient, three images with
large areas of healthy bone marrow were selected from the coronal inversion
recovery sequence. One region of interest (ROI) was defined in each of these
three images. Because epiphysis and metaphysis sometimes show a slightly
different signal in these sequences, we attempted to include equal parts of
both regions in the ROI. Each ROI was then analyzed using MRI analysis
software (MaZda Institute of Electronics), and a gray-scale value higher than
or equal to that of 99% of the pixels inside the ROI was calculated. Then the
arithmetic mean and standard deviation (SD) were calculated from the 99%
values of the three ROIs. This arithmetic mean was used as the threshold value
between normal bone marrow and bone marrow edema for that particular
examination of the patient.
To determine the intraobserver variability and mean SD, this step was performed three times for every patient, providing us with three threshold values (threshold value 1, threshold value 2, threshold value 3) and three threshold SDs (TSD1, TSD2, TSD3) per patient.
Volume and Signal Intensity Calculation
To better define the examined volume of bone marrow, we measured the
maximal width of the proximal tibial and distal femoral condyle in the coronal
sequence using the EasyVision software package (Philips). This value was then
used as the maximal height of the total volume of investigation (TVI) as seen
from the most caudal (for the femur) or most cranial (for the tibia) point of
the condyles (Fig. 1A). Then an
orthogonal line maintained in a constant position for all MR slices was drawn
to separate the TVI from the adjacent diaphysis. The remaining margins of the
bone marrow were drawn manually, slice by slice
(Fig. 1B). The thus-defined TVI
was calculated in cubic millimeters by the software with reference to slice
thickness and interslice gap.
|
|
In the next step, EasyVision's threshold module was used in conjunction with the threshold value calculated in the previous step to perform a color demarcation of the edema inside the TVI (Fig. 1C). This color demarcation was applied to all slices automatically. Small and therefore irrelevant color holes and islands with fewer than four connected pixels were removed. The software then calculated the volume of the colored bone marrow edema in cubic millimeters. The average signal intensity (ASI) and corresponding SD of the bone marrow edema were also measured automatically.
|
Finally, the percentage of edema (PE) in the TVI was calculated for description of bone marrow edema size. For characterization of the edema's signal intensity, the difference in ASI-to-threshold (DAT) was used.
To determine intraobserver variability, we measured the PE and DAT three times for each patient using threshold value 1, which led to three PE (PE1a, PE1b, PE1c) and three DAT (DAT1a, DAT1b, DAT1c) values. Then PE2 and DAT2 were calculated using threshold value 2, and PE3 and DAT3 were calculated using threshold value 3.
Statistics
To assess the reproducibility of the method, we determined the
intraobserver variability. This was accomplished by calculating the
intraobserver coefficient of variation (ratio of SD and arithmetic mean as
percentage) for the final results of the quantification as well as for the
threshold value and the manually outlined TVI independently. The variation
coefficients were determined as follows (VT = threshold value):
|
|
Threshold variability describes the intraobserver reproducibility of the threshold value calculation. Volume and signal variability (TVI) provide information about the intraobserver reproducibility of the manual outlining of the contours of the TVI by measuring the effect on volume and signal intensity. The total volume and signal variabilities show the impact of both threshold and TVI variations on the final results.
In addition to the intraobserver variability of the threshold value, the
arithmetic means 
i (for i:
TSD1, TSD2, TSD3) of the standard deviations were calculated.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Volume variability (TVI) and signal variability (TVI) were respectively 0.0025 (0.25%) and 0.0036 (0.36%), which was much less than expected (Table 2). Finally, the total volume and signal variabilities were calculated as 0.0089 (0.89%) for the volume and 0.0094 (0.94%) for the signal intensity (Table 3), showing the overall reproducibility of the results produced with this method.
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
A simple way to assess edema volume was described by Roemer et al. [3], who manually measured the maximal diameters of the bone marrow edema in all three dimensions using coronal and sagittal STIR images. These authors then calculated bone marrow edema volume by multiplying the three diameters. However, because of the fuzzy appearance of the bone marrow edema margins, measurement of the diameters seems highly subjective. For that reason, the results from this study are unlikely to be reproducible. In addition, the calculated volume is cubelike, which adds to the inaccuracy of the results.
An improved method was presented by Schmid et al. [4], who measured edematous signal abnormalities in the foot and ankle with STIR and contrast media–enhanced T1-weighted sequences. Again, the volume of these abnormalities was assessed manually therefore relying on inaccurate outlining of the abnormal tissue and was then multiplied by the thickness of the slices. In addition, an effort was made to measure the signal intensity of the bone marrow edema by comparing the mean gray-scale value of a subjectively defined ROI inside the edematous tissue with the mean gray-scale value of an ROI in the adjacent normal-appearing tissue. In our experience, however, drawing an ROI in the diffuse edematous tissue is problematic because the distribution of gray-scale values within that ROI is representative only of the ROI itself but not necessarily of the whole of the bone marrow edema. Thus, the mean gray-scale value of different ROIs inside the bone marrow edema may vary significantly.
The lack of a gold standard for quantification of the bone marrow edema led us to create a method that is largely independent of the observer's experience. Manual interference is necessary only to outline the contours of the examined bone marrow and to draw an ROI of no particular shape inside the healthy tissue. By using a 99% gray-scale value as a threshold between healthy and edematous tissue, we have minimized the influence of small pixel clusters with high signal intensity that would otherwise have damaged the result. This threshold value allows us to automatically and precisely segment the abnormal tissue, thus eliminating intra- and interobserver variability for this step. In addition, by using the threshold value as a dynamic reference for the description of the bone marrow edema signal intensity (DAT, difference in ASI-to-threshold), changes in both the volume and the signal of the bone marrow edema can be mathematically expressed.
The calculated intraobserver coefficients of variation suggest that this method provides highly reproducible results and may therefore be valuable for sensitive quantification of bone marrow edema and, probably, also for monitoring the progression of these lesions.
The major drawback of this method lies in the amount of time necessary for the analysis, about 20 min per condyle. In addition, bone marrow edema can in no way be automatically differentiated from other abnormal tissue changes with similar signal intensity (e.g., from enchondromas). For this type of characterization, a pattern-matching algorithm would be necessary that takes into account not only the gray-scale value threshold of healthy bone marrow but also other parameters such as entropy (quantifying signal inhomogeneities within an ROI) and position inside the bone marrow. Until a software package with these capabilities is released, a semiquantitative scoring system for clinical use must be created that shows an adequate correlation with the measurements produced with this method.
In conclusion, a computer-assisted method has been described that is less subjective and more reproducible than human estimation. Similar technology in the field of CT has significantly improved patient treatment (e.g., through calcium scoring of the heart and bone mineral densitometry). In MRI, this technology could vastly improve specificity, especially if applied to multiple pulse sequences.
The foremost strength of this method lies in the largely observer-independent segmentation of abnormal from healthy tissue, which is the most critical step when measuring the volume of lesions with irregular morphology such as bone marrow edema. The use of a reproducible, dynamic gray-scale value threshold between edema and surrounding healthy tissue allows comparison of measurements not only among different examinations of the same patient but also among different patients. Another possible application is the comparison of different MR sequences with regard to contrast material behavior. Although the method is too time-consuming for clinical use, it is recommended for scientific purposes.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  S. S. Boks, D. Vroegindeweij, B. W. Koes, R. M. D. Bernsen, M. G. M. Hunink, and S. M. A. Bierma-Zeinstra MRI Follow-Up of Posttraumatic Bone Bruises of the Knee in General Practice Am. J. Roentgenol., September 1, 2007; 189(3): 556 - 562. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. E. Mayerhoefer, J. Kramer, M. J. Breitenseher, C. Norden, A. Vakil-Adli, S. Hofmann, R. Meizer, H. Siedentop, F. Landsiedl, and N. Aigner Short-term outcome of painful bone marrow oedema of the knee following oral treatment with iloprost or tramadol: results of an exploratory phase II study of 41 patients Rheumatology, September 1, 2007; 46(9): 1460 - 1465. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K.-P. Jungius, M. R. Schmid, M. Zanetti, J. Hodler, P. Koch, and C. W. A. Pfirrmann Cartilaginous Defects of the Femorotibial Joint: Accuracy of Coronal Short Inversion Time Inversion-Recovery MR Sequence Radiology, August 1, 2006; 240(2): 482 - 488. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
