Comparison of 2D and 3D MRI Texture Analyses of Functionally Different Hip Muscles
Nketiah, Gabriel (2013)
Nketiah, Gabriel
2013
Master's Degree Programme in Biomedical Engineering
Tieto- ja sähkötekniikan tiedekunta - Faculty of Computing and Electrical Engineering
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2013-03-06
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tty-201303181092
https://urn.fi/URN:NBN:fi:tty-201303181092
Tiivistelmä
The need for detailed image information to enhance radiological decision making has necessitated computerized analysis of medical images. The superior sensitivity of MRI in detecting subtle changes in soft tissues has facilitated automatic analysis of MR images through texture analysis. Though there have been a number of recent studies in this area, most of them have focused on using 2D MRI texture analysis in detecting and classifying pathological tissues from healthy ones.
The objective of this thesis it to examine whether textural differences exist in hip muscles due to exercise-loading differences, if so, the effectiveness of 2D and 3D MRI texture analyses in detecting and characterizing these differences will be examined.
Ninety-one high-level female athletes representing five distinct loading sports (High-impact, odd-impact, high-magnitude, low-impact and non-impact exercise-loading) and 20 healthy non-athlete (referent) female subjects were used in this study. A 1.5T MRI scanner (Siemens, Erlangen, Germany) was used to acquire axial T1-weighted FLASH sequence images of the hip muscles. Two-dimensional(2D) and three-dimensional(3D) texture analyses were performed on four specific load-bearing muscles (gluteus maximus, gluteus medius, iliopsoas and obturator internus) using texture analysis application software – MaZda (TUL, Poland: COST Action B11). The computed texture parameters were statistically analyzed (using SPSS, Chicago, Ill.) to ascertain differences in texture between the four muscles, the non-athlete group and the athlete groups, and to characterize them accordingly. A comparative evaluation of 2D and 3D texture analyses was also made.
Significant differences (p-value < 0.00833) in texture were recorded between the four muscles. All the four muscles were found to be linearly separable from each other. Moreover, muscle texture of athletes who were involved in high-impact (triple-jumpers and high-jumpers), odd-impact (soccer and squash players) and low-impact (endurance runners) exercise-loadings differed significantly (p-value < 0.01) from that of the non-athletes. Subsequently, the high-impact, odd-impact and low-impact exercise-loading groups were completely separable from the non-athlete group. Contrarily, muscle texture of the high-magnitude (power lifters) and non-impact (swimmers) exercise-loading groups were not found to differ significantly from the non-athletes, some level of overlap was noticed in their classification from the non-athletes. Finally, 3D texture analysis was more effective in detecting and characterizing textural differences in skeletal muscles than the 2D texture analysis.
In conclusion, the 3D texture analysis of MR images provides a more accurate quantitative method for detecting and classifying textural differences in skeletal muscles that are associated with specific exercise-loading types.
The objective of this thesis it to examine whether textural differences exist in hip muscles due to exercise-loading differences, if so, the effectiveness of 2D and 3D MRI texture analyses in detecting and characterizing these differences will be examined.
Ninety-one high-level female athletes representing five distinct loading sports (High-impact, odd-impact, high-magnitude, low-impact and non-impact exercise-loading) and 20 healthy non-athlete (referent) female subjects were used in this study. A 1.5T MRI scanner (Siemens, Erlangen, Germany) was used to acquire axial T1-weighted FLASH sequence images of the hip muscles. Two-dimensional(2D) and three-dimensional(3D) texture analyses were performed on four specific load-bearing muscles (gluteus maximus, gluteus medius, iliopsoas and obturator internus) using texture analysis application software – MaZda (TUL, Poland: COST Action B11). The computed texture parameters were statistically analyzed (using SPSS, Chicago, Ill.) to ascertain differences in texture between the four muscles, the non-athlete group and the athlete groups, and to characterize them accordingly. A comparative evaluation of 2D and 3D texture analyses was also made.
Significant differences (p-value < 0.00833) in texture were recorded between the four muscles. All the four muscles were found to be linearly separable from each other. Moreover, muscle texture of athletes who were involved in high-impact (triple-jumpers and high-jumpers), odd-impact (soccer and squash players) and low-impact (endurance runners) exercise-loadings differed significantly (p-value < 0.01) from that of the non-athletes. Subsequently, the high-impact, odd-impact and low-impact exercise-loading groups were completely separable from the non-athlete group. Contrarily, muscle texture of the high-magnitude (power lifters) and non-impact (swimmers) exercise-loading groups were not found to differ significantly from the non-athletes, some level of overlap was noticed in their classification from the non-athletes. Finally, 3D texture analysis was more effective in detecting and characterizing textural differences in skeletal muscles than the 2D texture analysis.
In conclusion, the 3D texture analysis of MR images provides a more accurate quantitative method for detecting and classifying textural differences in skeletal muscles that are associated with specific exercise-loading types.