IN
DEGREE PROJECT MEDICAL ENGINEERING, SECOND CYCLE, 30 CREDITS
,
STOCKHOLM SWEDEN 2018
T2 Mapping compared to
standard MRI assessment
An assessment of the knee cartilage on distal
femur
JENNIE ANDERSSON
Abstract
Magnetic resonance imaging (MRI) has become the most important modality for assessment of pathological changes in the knee cartilage. The assessment of the cartilage is usually made by a set of anatomical MRI images with different sequences. Newer techniques, that map various in MRI parameters, have been devel-oped and allows changes in an earlier stage of the disease. One of these techniques is T2 mapping. The goal of this thesis was to compare this newer technique, T2 mapping, with the standard MRI assessment for assessment of articular cartilage on distal femur in the knee. The purpose was to assess the cartilage with these two different methods and analyze its outcomes.
Eight subjects were included in this study and scanned with a 3.0 T or 1.5 T MRI machine. A specific MRI knee protocol was used for the standard MRI assessment, and a multi-echo sequence was used for the T2 mapping. The T2 map was created and analyzed in the program IntelliSpace Portal.
Both the standard MRI assessment and the T2 map showed changes in the knee cartilage. The result showed either indication for damage cartilage or healthy cartilage. The standard assessment showed carti-lage lesion in three subjects and no lesion in five subjects. The same outcomes were with the T2 mapping. However, not all results were equal. The T2 mapping also showed higher values in the trochlea area where no indications for changes were found in the standard assessment.
This study showed similar results for both the standard assessment and the T2 map. Both methods could identify damage and is, therefore, useful for assessment of the knee cartilage. The outcomes of the different methods differ, and the assessment is therefore made in different ways. The T2 mapping can be analyzed both visual and quantitative. The outcomes were both a color map of the knee but also results in graphs and values. The standard assessment is only assessed from grayscale images. The best outcomes from the T2 mapping was when it only was changes within the cartilage and not when the cartilage lesion was adjacent to an underlying bone lesion. Based on what was examined in this work, the best result was when T2 mapping was used together with the anatomical images used in the standard assessment.
The conclusion is that the standard assessment is necessary when it comes to make a damage assessment and perform damage marking as for Episurf. The T2 mapping is, however, an interesting method and will be more useful with more applications in the future. It is therefore exciting to keep an eye on the technology and its development.
Keywords
Sammanfattning
Magnetisk resonanstomografi (MR) har blivit den viktigaste modaliteten vid bed¨omning av patalogiska f¨or¨andingar i kn¨abrosket. Bed¨omningen av brosket g¨ors vanligtvis med hj¨alp av anatomiska MR bilder som ¨
ar skannade med olika sekvenser f¨or att f˚a olika viktningar p˚a bilderna. En nyare teknik, T2 mappning, som kartl¨agger olika MR prameterar, har utvecklats f¨or att med hj¨alp av andra parametrar analysera kn¨abrosket. Den h¨ar tekniken har resulterat i att f¨or¨andringar i brosket kan uppt¨ackas vid ett tidigare stadie i sjuk-domsf¨orloppet. M˚alet med det h¨ar examensarbetet var att j¨amf¨ora de olika teknikerna, T2 mappning och MR-standardbed¨omningen, f¨or att bed¨oma ledbrosket p˚a distala l˚arbenet i kn¨aet. Syftet var att bed¨oma brosket utifr˚an dessa olika metoder samt att analysera och j¨amf¨ora dess resultat.
˚
Atta subjekt ingick i studien och skannades med en 3,0 T eller 1,5 T MR-maskin. Ett specifikt MR-kn¨aprotokoll anv¨andes f¨or att skanna sekvenserna som ingick i standard bed¨omningen och en multi-ekosekvens anv¨andes f¨or T2 mappningen. T2-mappningen skapades och analyserades sedan i programmet IntelliSpace Portal.
B˚ade standard MR-bed¨omningen och T2-mappningen visade tydliga f¨or¨andringar i brosket. Resultatet visade antingen indikationer p˚a skadat eller friskt brosk. Standardbed¨omningen visade broskskador hos tre subjekt och inga broskskador hos fem subjekt. Samma resultat visades med T2-mappningen. D¨aremot skilde sig vissa resultat mellan T2 mappningen och standardbed¨omningen.
D˚a denna studie visade liknande resultat f¨or b˚ade standardbed¨omningen och T2-mappningen, ¨ar b˚ada metoderna anv¨andbara f¨or bed¨omning av kn¨abrosket. De olika metoderna har olika utfall vilket g¨or att bed¨omningen sker p˚a olika s¨att. I T2 mapping f˚ar man ut b˚ade en f¨argkarta ¨over kn¨at men ocks˚a grafter och v¨arden som kan anv¨andas. I standardbed¨omningen g¨ors bed¨omningen bara utifr˚an olika gr˚askalebilder. T2 mappningen var mest anv¨andbar n¨ar det var tydliga f¨or¨andingar i bara brosket och inte n¨ar skadan mest var i benet. Det b¨asta resultatet var d¨aremot n¨ar T2 mappning anv¨andes tillsammans med standardbed¨omningen. Slutsatsen ¨ar att standardbed¨omningen ¨ar n¨odv¨andig n¨ar det kommer till att bed¨omma skador och g¨ora en skademarkering s˚a som f¨or Episurf. T2 mapping ¨ar d¨aremot en v¨aldigt intressant teknik men ¨ar idag inte en vanlig teknik inom diagnostiken och saknar just nu n˚agot tydligt anv¨andningsomr˚ade. D¨aremot, finns det stor potential och kommer troligtvis bli vanligare och f˚a fler anv¨andingsomr˚aden i framtiden.
Nyckelord
Acknowledgments
I would first like to thank Episurf for providing me with this project and for giving me the opportunity to work on this topic. Especially, many thanks to my supervisor Ingrid Bratt, Medical Imaging Engineering at Episurf, for the support, help, and guidance throughout the work.
I would also thank Roberto Vargas Paris, Development and Research Responsible MRI-nurse at KS, for helping me with the T2-mapping.
Also many thank to Clas Aspelin at Philips, for letting me use the software IntelliSpace to create the T2 map.
Finally, I would like to thank my supervisor at KTH, Rodrigo Moreno, and my project group students Androula Savva, Dimitrios Gkotsoulias, Elin Wessel, Krishnadev Moothandassery Ramdevan, and Mahamad Alfakih for the social support and for helping each other out.
Abbreviation
Cor
Coronal
FoV
Field of view
FS
Fat saturation
MRI
Magnetic resonance imaging
NRM
Nuclear magnetic resonance
OA
Osteoarthritis
OCD
Osteochondritis dissecans
ON
Osteonecrosis
PD or PDW
Proton density weighted
ROI
Region of interest
Sag
Sagittal
SPAIR
SPectral Attenuated Inversion Recovery
TE
Echo time
TR
Repetition time
Tra
Transversal
TSE
Turbo spin echo
T2 or T2W
T2-weighted
WATSc
WATer Selective
2D
Two-dimensional
Contents
1 Introduction 1
2 Methods 2
2.1 Standard MRI assessment . . . 2
2.1.1 MRI knee protocol . . . 2
2.1.2 Assessment of the knee . . . 3
2.2 T2-mapping . . . 4
2.2.1 Assessment of the knee . . . 4
2.3 Analysis of the results . . . 5
3 Results 6 3.1 Damage assessment . . . 6
3.2 Standard MRI assessment . . . 7
3.2.1 Lesion subject 1 . . . 7 3.2.2 Lesion subject 5 . . . 8 3.2.3 Lesion subject 7 . . . 9 3.3 T2-mapping . . . 10 3.3.1 Subject 1 . . . 10 3.3.2 Subject 2 . . . 12 3.3.3 Subject 3 . . . 13 3.3.4 Subject 4 . . . 15 3.3.5 Subject 5 . . . 16 3.3.6 Subject 6 . . . 18 3.3.7 Subject 7 . . . 19 3.3.8 Subject 8 . . . 21
3.4 Comparison between T2 map and Standard MRI assessment . . . 22
3.4.1 Subject 1 . . . 22 3.4.2 Subject 3 . . . 23 3.4.3 Subject 5 . . . 24 3.4.4 Subject 6 . . . 24 3.4.5 Subject 7 . . . 25 3.4.6 Subject 8 . . . 26
4 Discussion and Conclusion 28 5 Future work 32 A State of the art 33 A.1 Magnetic resonance imaging . . . 33
A.1.1 MRI principle . . . 33
A.1.2 Relaxation processes . . . 34
A.1.3 MRI image . . . 34
A.2.1 Structure of articular cartilage . . . 37
A.2.2 Pathology of articular cartilage . . . 38
A.3 T2 mapping . . . 40
A.3.1 T2 values . . . 41
A.3.2 T2-mapping techniques . . . 41
A.3.3 T2-mapping assessment of articular cartilage . . . 42
A.3.4 IntelliSpace Portal . . . 43
A.4 Previous Studies . . . 44
A.4.1 Different methods to evaluate articular cartilage . . . 44
A.4.2 Studies about T2-mapping . . . 44
Chapter 1
Introduction
Magnetic resonance imaging (MRI) has become the most important modality for the assessment of patho-logical changes in the knee cartilage. MRI is a non-invasive medical imaging technique used in radiology to produce detailed anatomical images of the human body. This technique produces better images of organ and soft tissue and has a higher resolution than other medical imaging technologies without using ionizing radiation. MRI provides detailed images of the structure within the knee joint, including bones, cartilage, tendons, ligaments, menisci and muscles. The high resolution enhances differences in signal intensity be-tween cartilage and synovial fluid and bebe-tween cartilage and subchondral bone. The visualization of articular cartilage and focal osteochondral injuries has been improved by the development and advancement of the MRI technology [1].
Progress in MRI has developed new methods for assessment of the articular cartilage. Techniques to map various in MRI parameters had been developed and visualize the biochemical and biophysical changes of the cartilage. Compared to other methods, these newer techniques allow changes in an earlier stage of the disease. One of these techniques, to distinguish normal and abnormal tissue, is the T2-mapping which use T2 relaxation time as a parameter [2]. T2 mapping provides both a visual evaluation with high image quality and quantitative analysis, by drawing ROIs on the map. The T2 map consists of a color map based on the T2 values and is shown as a fused overlay on the original series (anatomical image). The T2 map provides an easy way to see variations within the cartilage and assess damage [3].
The T2 relaxation time is a function of the water content, collagen content and the collagen fibrils orien-tation in the extracellular matrix. By measuring the spatial distribution of T2 relaxation times through the cartilage, areas of increased or decreased water content can be identified. These changes correlate to carti-lage damage and render T2 mapping to be a useful technique to depict early changes of articular carticarti-lage [4]. Episurf Medical AB treats painful joint injuries with patient-specific technology. MRI data is used to assess the knee and estimate cartilage lesion. Episurf Medical AB has developed a specific MRI scanning protocol to get a detailed view of the joint. It is optimized for best possible assessment and a tailored 3D sequence, as well as conventional diagnostic sequences (2D sequences), are included [5].
Since T2 mapping also is a technique to assess the changes in the knee cartilage, it is interesting to analyze this method and compare it with Episurf standard assessment.
The main goal of this master thesis is to compare two different methods, T2 mapping and a standard MRI assessment with anatomical images, to estimate damage in the articular cartilage on distal femur in the knee. The purpose is to analyze the outcomes from the different methods and compare the result from the T2 map with the result from the standard MRI assessment. The goal is to study what is counted as damage in the anatomical images and in the T2 mapping.
Chapter 2
Methods
The study group in this thesis consist of eight subjects with or without knee pain. For the standard assess-ment, Episurf MRI knee protocol was used (see section 2.1.1) and scanned with two different MRI scanner, one MRI scan from Philips with a magnetic field of 3.0 Tesla and one MRI scan from GE (General Electric) with a magnetic field of either 1.5 T or 3.0 T. MRI T2-mapping sequences was performed with the 3 T MRI scan from Philips. Information about the subject and each MRI scan are found in the table 2.1. A knee coil, designed to fit the anatomical part of the knee, was used in order to obtain higher spatial resolution.
Table 2.1: Test data Subject Scanner Magnetic field
strength (T)
Knee side Sex Age
1 Philips 3 Right K 31 2 Philips 3 Right K 31 3 Philips 3 Left M 62 4 Philips 3 Right M 34 5 Philips 3 Left M 31 6 GE 1.5 Right K 26 7 GE 3 Right K 78 8 GE 1.5 Right M 35
2.1
Standard MRI assessment
2.1.1
MRI knee protocol
The Episurf MRI Knee protocol for Philips MRI machines or GE MRI machines consists of five MRI se-quences and are described in table 2.2. These include one 3D sequence and four 2D sese-quences with different orientation. One additional 3D sequence has also been used in this thesis. All sequences had a field of view (FOV) that covers the whole femoral bone and the articular cartilage. The required for the pixel size, slice thickness and slice gap for each scanner and sequences are described in the table 2.3 below [6] [7].
Table 2.2: MRI sequences included in Episurf MRI knee protocol for Philips and GE machines [6], [7] Type Orientation Pulse sequence
Philips GE
3D Sagittal 3D WATSc 3D T1 VIBE
3D Sagittal 3D PD (additional) 3D PD (additional) 2D Sagittal TSE Dual Echo PDW/T2W TSE Dual Echo PD/T2 2D Sagittal TSE PDW SPAIR TSE PD FS
2D Coronal TSE PDW SPAIR TSE PD FS 2D Transversal/Axial TSE PDW SPAIR TSE PD FS
Table 2.3: List of required,[6] [7]
Scanner Sequence Pixel size (mm) Slice thickness (mm) Spacing between slices (mm)
Philips 3D-sequence 0.5x0.5 1.0 1.0
Philips 2D-sequences 0.5x0.5 3.0 3.3
GE 3D-sequence 0.5x0.5 2.0 1.0
GE 2D-sequences 0.5x0.5 3.0 3.3
2.1.2
Assessment of the knee
The standard assessment was done using Materialise Mimics Medical v.20.01and Materialise 3-Matic
Med-ical v.12.02. The assessment has been done on distal femur and the region of interest was the cartilage on
the medial femoral condyle, lateral femoral condyle, and the trochlea. To perform a full damage assessment according to Episurfs routines, all MRI sequences with Episurf MRI knee protocol (table 2.2 and 2.3) were needed. All sequences (both 3D and 2D-sequences) was used for the damage marking, and the 3D-sequence was also used for segmentation. The diagnostic 2D sequences have higher resolution making it easier to detect the damaged cartilage. The assessment of each knee was done by checking through all the different sequences, one by one, and looking for abnormal variations in the cartilage. In each sequence, a segmentation mask was created and used to mark out the damaged part. From each sequence, 3D meshes were calculated from the masks. The different 3D meshes were compared and analyzed with each other, and if the markings differed a comparison was made between these sequences. A final assessment was established and merged together to create a single model. The size and dimension of the lesion were then calculated. In addition to my assessment, an independent assessment was also made by one radiologist and one medical engineering. In order to assess the articular cartilage, an understanding of the MR images and the normal and pathologi-cal conditions within the tissue is required. Each pixel in the MRI image consist of different signal intensity and appear with different grey values in the image, due to different tissues gives different signals. In a PD- or T2-weighing image, used in the standard assessment, the articular cartilage appears ”medium grey”, synovial fluid white and bone black.
It is the intensity changes that will give information about the healthiness. In healthy articular cartilage, the signal intensity is more or less unchanged, which gives a smooth, clear surface with the same grey value. In damaged cartilage, the signal intensity is instead changed and will deviate from the normal pattern. Damage cartilage can be distinguished in several ways; it can give a darker signal than normal as in Figure 2.1b or a brighter signal if the cartilage has been replaced by synovial fluid (Figure 2.1c). The thickness and the surface can also change and be visible as thinner/thicker or have irregular contours (Figure 2.1d)[8]. Figure 2.1a shows an MR image of a knee with a cartilage lesion on the medial condyle (right side of the image) and healthy cartilage on the lateral condyle (left side of the image).
1https://www.materialise.com/en/medical/software/mimics (2018-05-15) 2https://www.materialise.com/en/software/3-matic (2018-05-15)
For more information about the MR images and the pathology of articular cartilage, see appendix A.1.3 and A.2.2.
(a) Healthy vs damage cartilage.
(b) Darker signal (c) Brighter signal (d) Irregular surface
Figure 2.1: Signal intensity changes in the cartilage, results in changes of the normal pattern. (a) and (b) Coronal PD FS MR image, (c) and (d) Sagittal PD FS MR images
2.2
T2-mapping
MRI T2 maps were calculated from a multi-echo sequence with seven different TE-time and constant TR-time. The image sets with TE and TR time can be found in the table B.1 - B.4 in Appendix B. The pixel size for the T2-image was 0.5x0.5 and the slice thickness 3.0 mm. The T2-mapping was created and analyzed using Philips software IntelliSpace Portal v.93.
The signal is sampled from the different TE and the intensity of each pixel is fitted to one decay expo-nential. The T2 map is generated by a model fitting, defined by:
Si= M0exp
−T Ei T2
(2.1) where Si is signal intensity, M0 is a lumped parameter that includes the equilibrium magnetization and
local receiver coil gain, and T E is the T2 preparation time.
2.2.1
Assessment of the knee
The T2-mapping was created and analyzed with IntelliSpace Portal, and the package cartilage assessment was used to enable the visualization of cartilage structures integrated with color-coded T2 maps. To determine the degradation of the cartilage, a layered ROIs (region of interest) was used to assess the variation of T2
values across the cartilage. The cartilage surfaces were divided into segments and layers. For each knee, a curved layered ROI was drawing one the medial femoral condyle, lateral femoral condyle and in the trochlea area to segment the complete cartilage structure. In order to place the ROIs in the same locations on each knee, the anterior cruciate ligament (ACL) was used as a guideline for the lateral side, the posterior cruciate ligament (PCL) for the medial side and patella for the trochlea. The ROI was defined by placing multiple points at the anatomical image, first by drawing a curve along the bone interface and then drawing a curve along the cartilage surface. The ROIs included the entire cartilage thickness from the subchondral bone to the surface.
The ROI was divided into three different segments (A, B and C) and three layers (superficial, intermediate and deep). These are the standard-setting for analyzing of the cartilage.
The knee showing indications for having cartilage lesion was analyzed in more detail, and additional ROIs was created. Both a rectangular layered ROI and a curved layered ROI were created. The rectangular was used to analyzed the cartilage within the damaged region and the curved layered to analyzed the complete cartilage. The curved layered was divided into three segments where one segment was resized to the damaged area.
2.3
Analysis of the results
The result was analyzed in three different step.1. General analysis of the T2 mapping results where each subject was analyzed and compared with each other
2. Analysis of the result from the T2 mapping compared to the anatomical images in the standard assessment
3. Analysis of the result from the standard assessment with anatomical images compared to the T2 mapping
Chapter 3
Results
3.1
Damage assessment
The result from the damage assessment for both the standard MRI assessment and the T2 mapping is shown in table 3.1. The standard assessment showed cartilage lesion in three subjects and no indications for five subjects. In the T2 mapping, a cartilage lesion was also identified in three subjects. Furthermore, the T2 mapping showed indications for a small cartilage lesion in one subject, possibly a cartilage lesion in one subject and also indications for increased T2 values in the trochlea area for the subjects with healthy cartilage. Each assessment is described in the following section 3.2 and 3.3.
Table 3.1: Damage assessment of the knee
Subject Damage assessment Agreed
Standard T2 map
1 Medial condyle lesion Medial condyle lesion Yes 2 No cartilage lesion No cartilage lesion Yes 3 No cartilage lesion Indication for a small cartilage lesion No 4 No cartilage lesion No cartilage lesion Yes 5 Trochlea lesion Trochlea lesion Yes 6 No cartilage lesion No cartilage lesion Yes 7 Two small trochlea lesions One small trochlea lesion No 8 No cartilage lesion Possible cartilage lesion Maybe
3.2
Standard MRI assessment
This section presents the results from the standard MRI assessment of the anatomical images. A cartilage lesion was found in three subjects (subject 1, 5 and 7), and each result is described in this section. The other subjects showed no indication of a cartilage lesion.
3.2.1
Lesion subject 1
A cartilage lesion with an underlying bone marrow lesion was identified on the medial condyle on distal femur in this knee. The size and position of the lesion can be seen in the 3D-visualization in figure 3.1. The pink marking indicates degenerated or regenerated cartilage and the red marking indicates a possible full depth cartilage lesion. The lesion has a length of 16.04 mm and a width of 23.93 mm. The total area of the lesion is 254.7216 mm2.
Figure 3.1: Upper image: 3D visualization of the knee. The yellow part visualizes femur, the white cartilage, the pink and red the cartilage lesion.
Lower images: 3D visualizations of the knee showing the length(left) and width(right) of the lesion. MRI images of the lesion
3.2.2
Lesion subject 5
A cartilage lesion was identified in the trochlea area on distal femur in this knee. The size and position of the lesion can be seen in the 3D-visualization in figure 3.2. The pink marking indicates degenerated or regenerated cartilage. The lesion has a length of 14.93 mm and a width of 12.6 mm. The total area of the lesion is 134,7731 mm2.
Figure 3.2: Upper image: 3D visualization of the knee. The yellow part visualizes femur, the white cartilage and the pink the cartilage lesion.
MRI images of the lesion
3.2.3
Lesion subject 7
Two cartilage lesions were identified in the trochlea area on distal femur in this knee. A bone overgrowth has also been identified underlying the cartilage lesion on the medial side of the trochlea. The size and position of the lesions can be seen in the 3D-visualization in figure 3.3. The pink markings indicate degenerated or regenerated cartilage. The lesion on the medial side of trochlea has a length of 7.94 mm, width of 7.66 mm and a total area of 36.9575 mm2. The lesion on the lateral side of trochlea has a length of 15.54 mm, width of 0.88 mm and a total area of 10.6327 mm2.
Figure 3.3: Upper image: 3D visualization of the knee. The yellow part visualizes femur, the white cartilage and the pink the cartilage lesions.
MRI images of the lesion
3.3
T2-mapping
This section presents the results of the cartilage assessment from the T2 map and showed with images, plots, numerical results in tables and analysis. The color map represents the values of the T2 relaxation times per voxel, and each color stands for a specific T2 value. Dark red to red indicates low T2 values, yellow to green intermediate values and blue indicate high T2 values. The table lists the calculated T2 relaxation times in milliseconds with standard deviation (in the parenthesis) for each layer and segment. The graphical results in the diagram plot the T2 relaxation time versus the relative cartilage depth. The white lines at the upper part of each bar indicate the standard deviation.
In each subject, the knee was divided into segments (A, B and C) and layers (deep, intermediate and superficial) where on analyse was done on the medial femoral condyle, one on the lateral femoral condyle and one in the trochlea area. In the result from the lateral femoral condyle, the segment A represent part of the trochlea area.
In the T2-mapping assessment, a cartilage lesion was found in three subjects (subject 1, 5 and 7). Fur-thermore, a possible small cartilage lesion was found in subject 3 and subject 8 also possibly has a cartilage lesion. The T2 map also showed indications for increased T2 values in the trochlea area in subject with healthy cartilage.
3.3.1
Subject 1
The T2-map of subject 1 showed varying T2 values in the cartilage and one lesion was identified on the medial femoral condyle. The medial and lateral condyle showed normal variations within the cartilage while the trochlea area showed higher T2 values, especially within the segment B. Furthermore, the color map showed irregular surfaces on the medial femoral condyle without such increasing values. The T2-map, the box-plot and the T2-values for each sub-region are shown in the figures and tables below.
Medial femoral condyle
Medial femoral condyle A B C Deep 33,3 (8,5) 39,8 (12,6) 47,5 (10,1) Intermediate 43,0 (7,3) 48,8 (8,5) 54,1 (6,4) Superficial 48,1 (13,4) 47,8 (14,6) 50,7 (12,1)
Lateral femoral condyle
Lateral femoral condyle A B C Deep 45,9 (12,69) 31,3 (8,1) 47,9 (17,9) Intermediate 50,8 (13,5) 39,7 (11,1) 44,6 (14,4) Superficial 53,7 (12,3) 45,8 (20,5) 38,4 (16,7)
Trochlea area Trochlea area A B C Deep 50,3 (12,3) 54,3 (14,2) 28,0 (7,3) Intermediate 53,6 (7,6) 65,3 (6,2) 36,5 (9,8) Superficial 58,1 (7,2) 74,9 (9,6) 51,2 (15,8)
3.3.2
Subject 2
The T2 map of subject 2 showed variations in the T2 values but no indication for a cartilage lesion. The medial femoral condyle and lateral femoral condyle showed normal T2 values, while the trochlea area, on the other hand, showed small changes and indications for higher T2 values. The T2-map, the box-plot and the T2-values for each sub-region are shown in the figures and tables below.
Medial femoral condyle
Medial femoral condyle A B C Deep 44,6 (10,3) 42,4 (6,8) 49,3 (8,7) Intermediate 51,3 (12,1) 41,9 (9,3) 52,7 (7,6) Superficial 47,9 (18,6) 34,2 (14,1) 44,8 (27,4)
Lateral femoral condyle
Lateral femoral condyle A B C Deep 56,3 (11,2) 46,3 (16,6) 29,5 (11,9) Intermediate 65,5 (13,3) 58,7 (16,4) 34,0 (13,1) Superficial 74,3 (16,5) 66,3 (27,2) 43,5 (27,4) Trochlea area Trochlea area A B C Deep 50,5 (10,8) 59,1 (7,5) 58,4 (12,3) Intermediate 59,6 (12,9) 71,2 (6,7) 68,5 (12,1) Superficial 66,9 (12,7) 78,4 (7,9) 87,7 (18,1)
3.3.3
Subject 3
The T2 map of subject 3 showed indication for a possibly small cartilage lesion in the trochlea area. One small change with increased T2 values, was observed. Moreover, no significant changes was observed in the knee. The medial femoral condyle and lateral femoral condyle showed both normal T2 values. The T2-map, the box-plot and the T2-values for each sub-region are shown in the figures and tables below.
Medial femoral condyle
Medial femoral condyle A B C Deep 31,5 (6,3) 41,2 (10,0) 47,2 (4,7) Intermediate 41,9 (5,6) 48,6 (7,5) 47,0 (6,1) Superficial 49,4 (10,4) 56,1 (30,7) 35,7 (10,3)
Lateral femoral condyle
Lateral femoral condyle A B C Deep 46,2 (9,0) 28,0 (5,5) 34,5 (8,3) Intermediate 58,7 (8,9) 36,4 (7,2) 43,4 (8,3) Superficial 70,0 (8,6) 42,4 (12,2) 47,6 (15,8)
Trochlea area Trochlea area A B C Deep 37,0 (4,1) 41,6 (5,4) 39,1 (7,0) Intermediate 40,2 (4,1) 50,8 (5,6) 47,3 (8,3) Superficial 43,4 (4,7) 58,9 (7,0) 54,4 (7,3)
3.3.4
Subject 4
The T2 map of subject 4 showed varying T2 values but with no indication for a cartilage lesion. The medial femoral condyle and lateral femoral condyle showed both low and high values, in addition, with high standard deviations. The trochlea area, showed small changes in the color map and indications for higher T2 values. The T2-map, the box-plot and the T2-values for each sub-region are shown in the figures and tables below. Medial femoral condyle
Medial femoral condyle A B C Deep 45,2 (6,4) 46,2 (10,2) 46,4 (8,3) Intermediate 56,7 (7,4) 54,2 (10,3) 47,0 (7,9) Superficial 67,2 (26,7) 46,8 (11,2) 38,3 (10,8)
Lateral femoral condyle
Lateral femoral condyle A B C Deep 53,9 (13,4) 56,6 (17,3) 29,2 (7,5) Intermediate 62,6 (14,7) 71,2 (21,1) 30,2 (7,3) Superficial 69,2 (16,9) 90,1 (51,7) 30,5 (13,1) Trochlea area Trochlea area A B C Deep 51,6 (6,4) 60,6 (8,4) 65,8 (6,5) Intermediate 59,6 (3,8) 71,5 (6,5) 74,5 (5,7) Superficial 62,8 (4,7) 79,4 (10,9) 84,7 (7,3)
3.3.5
Subject 5
The T2 map of subject 5 showed a large area with increased T2 values in the trochlea area, indicates a cartilage lesion. The T2 values on the medial femoral condyle and lateral femoral condyle showed normal variation. The T2-map, the box-plot and the T2-values for each sub-region are shown in the figures and tables below.
Medial femoral condyle
Medial femoral condyle A B C Deep 31,1 (4,1) 38,1 (9,7) 53,7 (8,0) Intermediate 40,2 (5,1) 44,3 (7,6) 54,0 (6,9) Superficial 55,6 (24,2) 47,8 (10,4) 47,6 (9,3)
Lateral femoral condyle
Lateral femoral condyle A B C Deep 63,1 (15,2) 35,3 (6,3) 44,5 (7,4) Intermediate 66,1 (12,9) 45,2 (6,3) 53,6 (10,2) Superficial 75,3 (16,6) 55,3 (14,9) 55,7 (17,6)
Trochlea area Trochlea area A B C Deep 63,0 (12,1) 70,4 (11,7) 53,0 (12,5) Intermediate 69,6 (12,4) 86,1 (13,8) 65,0 (14,8) Superficial 70,2 (9,6) 91,0 (10,5) 72,3 (14,8)
3.3.6
Subject 6
The T2 map of subject 6 showed varying T2 values with increasing T2 values in the trochlea area. However, no indication for a cartilage lesion was identified despite the high values. The medial femoral condyle and lateral femoral condyle showed normal variations. The T2-map, the box-plot and the T2-values for each sub-region are shown in the figures and tables below.
Medial femoral condyle
Medial femoral condyle A B C Deep 34,8 (4,2) 41,9 (7,1) 52,2 (12,2) Intermediate 39,1 (9,2) 45,5 (6,3) 54,8 (10,7) Superficial 53,7 (19,4) 38,1 (12,4) 46,0 (13,5)
Lateral femoral condyle
Lateral femoral condyle A B C Deep 46,4 (9,3) 35,9 (11,7) 33,6 (11,1) Intermediate 53,3 (9,2) 45,4 (13,0) 32,0 (9,3) Superficial 56,6 (10,9) 65,0 (27,0) 33,1 (14,7) Trochlea area Trochlea area A B C Deep 72,5 (11,1) 79,4 (8,5) 79,8 (8,4) Intermediate 75,5 (9,7) 98,2 (8,9) 87,2 (9,3) Superficial 83,7 (13,6) 120,5 (15,8) 99,3 (10,7)
3.3.7
Subject 7
The T2 map of subject 7 showed a small cartilage lesion on the medial side of the trochlea. The hole knee showed varying T2 values with high standard deviation. The T2-map, the box-plot and the T2-values for each sub-region are shown in the figures and tables below.
Medial femoral condyle
Medial femoral condyle A B C Deep 22.2 (6.2) 27.0 (7.4) 44.3 (6.3) Intermediate 35.5 (7.9) 38.5 (12.6) 48.1 (7.5) Superficial 41.0 (11.1) 52.6 (26.0) 44.6 (8.7)
Lateral femoral condyle
Lateral femoral condyle A B C Deep 38.8 (9.3) 33.8 (15.2) 28.5 (18.2) Intermediate 48.4 (10.4) 49.1 (18.4) 35.8 (30.2) Superficial 54.7 (11.1) 67.2 (21.1) 44.4 (45.0)
Trochlea area Trochlea area A B C Deep 35.9 (6.6) 40.7 (8.2) 27.1 (9.4) Intermediate 36.6 (7.8) 56.9 (5.2) 42.5 (11.7) Superficial 37.1 (6.2) 62.0 (11.5) 53.1 (12.3)
3.3.8
Subject 8
The T2 map of subject 8 showed normal variations in the T2 values within the both conydle and no indication for a cartilage lesion was identified. The trochlea area, on the other hand, showed indications for higher T2 values and possibly indications for a cartilage lesion. The T2-map, the box-plot and the T2-values for each sub-region are shown in the figures and tables below.
Medial femoral condyle
Medial femoral condyle A B C Deep 27,0 (5,2) 35,5 (6,5) 47,7 (10,2) Intermediate 32,8 (7,7) 41,8 (4,6) 51,1 (8,1) Superficial 40,3 (18,4) 44,4 (14,7) 45,0 (11,4)
Lateral femoral condyle
Lateral femoral condyle A B C Deep 74,0 (20,0) 47,8 (16,8) 37,9 (17,6) Intermediate 75,6 (26,7) 58,8 (26,1) 36,0 (13,7) Superficial 95,9 (52,6) 79,4 (50,6) 39,1 (17,2) Trochlea area Trochlea area A B C Deep 65,1 (18,0) 83,2 (12,5) 74,2 (12,9) Intermediate 53,3 (9,7) 90,9 (11,1) 84,8 (12,6) Superficial 56,1 (9,8) 95,9 (12,9) 94,1 (11,8)
3.4
Comparison between T2 map and Standard MRI assessment
3.4.1
Subject 1
In both assessments, a cartilage lesion was identified in the same area. In the standard MRI assessment, the cartilage lesion was identified due to the changes within the cartilage and the underlying bone marrow lesion. In the T2 mapping, the T2 values were not increased, but when analysed the color map a cartilage
lesion could be identified due to the irregular surfaces. Furthermore, the anatomical image included in the map also had changes within the cartilage, and the thickness of the cartilage was reduced. In figure 3.4 a comparison between the result from the different methods is shown. The red arrow in each image points out the area where the cartilage lesion was found.
(a) Anatomical image from T2 map (b) T2 map (c) Anatomical image
Figure 3.4: Comparison between T2 map and anatomical image for subject 1
T2 values of the cartilage lesion and the surrounding cartilage ROI Cartilage lesion region Surrounding cartilage Deep 33,6 (7,9) 32,3 (6,8)
Intermediate 45,1 (5,3) 42,8 (7,9) Superficial 66,4 (38,5) 44,7 (10,4)
3.4.2
Subject 3
In the T2 mapping, an observation was made of a small change in the cartilage in the color map, possible indication for a small cartilage lesion. When compared this area with the anatomical images included in the standard MRI assessment, a certain change could be identified in the cartilage. In the standard MRI assessment, this corresponds, however, more to MRI signal changes than damaged cartilage. The figure 3.5 below shows a comparison between the different images. The red arrow in each image points out the area with increased T2 values.
(a) Anatomical image from T2 map (b) T2 map (c) Anatomical image
T2 values of the cartilage lesion and the surrounding cartilage ROI Cartilage lesion region Surrounding cartilage Deep 55,6 (10,5) 42,9 (8,4)
Intermediate 74,8 (9,8) 52,8 (7,0) Superficial 71,8 (7,0) 54,4 (9.0)
3.4.3
Subject 5
In both assessments, a cartilage lesion was clearly identified in the same area. In the standard MRI assess-ment, the cartilage lesion was identified due to the changes within the cartilage and the underlying bone defect. In the T2 mapping, the values were increased in the trochlea area and a distinct change was identified in the color map. In figure 3.6 below a comparison between the result from the different methods is shown. The red arrow in each image points out the area where the cartilage lesion was found.
(a) Anatomical image from T2 map (b) T2 map (c) Anatomical image
Figure 3.6: Comparison between T2 map and anatomical image for subject 5
T2 values of the cartilage lesion and the surrounding cartilage ROI Cartilage lesion region Surrounding cartilage Deep 79,5 (14,9) 48,6 (12,3)
Intermediate 97,6 (12,8) 59,8 (14,1) Superficial 104,4 (15,8) 69,4 (10,7)
3.4.4
Subject 6
On the T2 map, higher T2 values were observed in the trochlea area. When compared this area with the anatomical images included in the anatomical MRI assessment, no indication for any lesion was found. In figure 3.7 below a comparison between the result from the different methods is shown. The red arrow in each image points out the area where the cartilage lesion was found.
(a) Anatomical image from T2 map (b) T2 map (c) Anatomical image
Figure 3.7: Comparison between T2 map and anatomical image for subject 6
3.4.5
Subject 7
In both assessments, a cartilage lesion was identified in the same area. In the standard assessment, an additional small lesion was identified. Both lesions identified in the standard MRI assessment was adjacent to a bone defect and only small changes within the cartilage. When compared the additional lesions in the T2 map, no indication for changes were observed. The T2 map showed overall varying values in the knee with no significant increasing values. The figure 3.8 below shows a comparison between the result where the lesion was identified in the both methods and figure 3.9 shows the result where an additional lesion was identified in the anatomical image. The red arrow in each image points out the area where the cartilage lesions were found, either in the T2 map or in the anatomical image.
(a) Anatomical image T2 map (b) T2 map (T2 range 0-120) (c) T2 map (T2 range 1-81) (d) Anatomical image
Figure 3.8: Comparison between T2 map and anatomical image for subject 7 for the lesion identified in both methods
T2 values of the cartilage lesion and the surrounding cartilage ROI Cartilage lesion region Surrounding cartilage Deep 75,9 (21,3) 53,5 (7,1)
Intermediate 74,9 (20,1) 53,1 (3,3) Superficial 79,2 (16,9) 59,3 (7,0)
(a) Anatomical image T2 map (b) T2 map (T2 range 0-120) (c) T2 map (T2 range 1-81) (d) Anatomical image
Figure 3.9: Comparison between the T2 map and anatomical image for subject 7 for the lesion only identified in the standrad assessment
3.4.6
Subject 8
In the T2 mapping, higher T2 values were observed in the trochlea area, and in the color map changes could be identified. This possibly corresponds to a cartilage lesion. When compared this area with the anatomical images included in the standard MRI assessment, no indication for any lesion was found. In figure 3.10 and 3.11 below a comparison between the result from the different methods is shown. The red arrow in each image points out the area where the cartilage lesion was found.
(a) Anatomical image from T2 map (b) T2 map (c) Anatomical image
(a) Anatomical image from T2 map (b) T2 map (c) Anatomical image
Chapter 4
Discussion and Conclusion
The main goal of this master thesis was to identify cartilage lesion on the distal femur with to different meth-ods, Episurfs standard MRI assessment included anatomical images and the newer T2 mapping technique and compared the result with each other. The assessment was made by estimating the articular cartilage on the distal femur. Both methods showed indications for cartilage lesions and results in similar outcomes. T2 mapping values
The distribution of T2 values in the femoral cartilage of each subject showed varying values. The mapping showed both variations in the cartilage within each subject and compared between each other. Furthermore, the different region on distal femur (medial femoral condyle, lateral femoral condyle, and trochlea), as well as the different layers, showed also varied result in T2 values. These findings, are however, in accordance with results from previous studies by Yoon, Min A et al. [9], Shiomi et al. [10], and Hannila et al. [11], also suggesting that there is a normal variation of T2 values within the knee. The result also showed for some subject, high standard deviation, which also showed high variation within the cartilage. These results, together with the nonspecific T2 values for healthy and abnormal cartilage, make the analysis of healthiness of the cartilage difficult. It also makes it difficult to compare the different subject with each other to get a perception of how the T2 values should be.
When looking at T2 values of other tissues within the field of view (FoV), we can see that muscle appeared as red in the color map, represent low T2 values, fat as yellow/orange, represent intermediate values and bone appeared as blue, represent high T2 values. When comparing these values with the literature, to see how accurate the T2 mapping method was, it seems to be consistent with expected, and the conclusion is that the mapping show relevant outcomes.
The subjects were scanned at four different times and therefore, different TE times was acquired, see table B.1 - B.4 in Appendix B. When comparing the different TE time, it is only small differences. Subject 5-8 is acquired with the same sets and have the same TE times. The same is for subject 1 and 2. The different between subjects 5-8 and 1-2 is the TR time. It is, however, the TE time that affects the result with T2-mapping and therefore, the different TR times should not affect the computations and the comparison between the different subjects. Subject 3 and 4 is, on the other hand, acquired with different set and different TE times was achieved. Anyway, the TE time for subject 4 is basically exactly the same as the other, and for subject 3 only 0.5 ms differ. This small differences should not affect the computations.
Outcomes from the assessment
The trochlea area showed indications of increased T2 values in each subject. More or less changes were observed and by only looking at the T2 map, it could be interpreted as damage. However, when compared to the specific region with the anatomical images from the standard assessment, there were no indications for abnormal cartilage.
The differences between the map of cartilage with and without a cartilage lesion are the extent and uniformity of the color within the map. The map for subject 5, which actually has a lesion, had significant differences in T2 values in the cartilage at the damaged area compared to the other regions. The color of
the map also shifts more markedly for subject 5 than for example subjects 6 and 8 that have high T2 values but no indications for abnormal cartilage in the anatomical images. In these subjects, the increasing values are furthest by the superficial layer and the edge of the cartilage, as well as the thickness of the cartilage is very thin, which make it difficult to assess the actual structure. As reported in the previous study ([9], [10], [11], [12]), the mean T2 values is higher in the trochlea area as well as in the superficial layer. Based on this, one can conclude that the increasing values could be normal and thus not corresponds to any lesion. Furthermore, both subject 6 and 8 were scanned with a magnetic field strength of 1.5 T. Lower magnetic field strength indicates higher T2 values, and this should also be one reason for the higher T2 values. Sub-ject 6, also had moving artefacts within the images, which also can be one additional reason for the outcomes. When analyzing the T2 map for subject 3, the cartilage had very uniform color and low T2 values, which corresponds to normal and healthy cartilage, except on one slice, where a significant increase in T2 values was observed. This was observed by a change of the color scale in a small region within the cartilage (figure 3.5). When compared it with the anatomical images, some small changes could be distinguished, probably MRI signal changes. In this case, it was evident in the T2 map that there was some change in the cartilage, due to the specific changes in color. In the anatomical images, on the other hand, it was not as clear and would not be marked as damage. So the question here is whether it is the beginning of a cartilage lesion or if it is only MRI signal changes.
Also when comparing these findings with those subjects who had higher T2 values in the trochlea area without any indication for abnormal cartilage, this change was more sharp and more inside the cartilage, unlike where the changes occurred to the edge of the cartilage and adjacent to structure with higher T2 values. The T2 map of the subject 7 showed only one indication for a cartilage lesion while the standard assessment showed signs of an additional lesion in the trochlea area. When comparing the findings in the anatomical images from the standard assessment with the T2 map, no indication of increased values was shown in the T2 map. In this case, the additional cartilage lesion corresponds more to a bone marrow lesion, which also can be observed in the T2 map.
What can be seen from having analyzed the various cartilage lesions included in this thesis is that when it comes to T2 mapping, the best results are obtained if the lesion is in the cartilage and does not correspond to a bone marrow edema where the cartilage is more displaced, i.e. the cartilage is replaced by bone. If the structure is replaced with another structure, like bone, that also has different T2 values, the result will differ. In order to get the best results, the damage should therefore correspond to changes in the cartilage. Also, thinner cartilage will result in thinner ROIs and dividing thinned cartilage into superficial and deep ROIs will not correspond to the same tissue as for healthy subjects.
T2 mapping ROI
An important and a major part of the T2 mapping is the segmentation of the ROI, which are manually drawn to delineate cartilage areas on the T2 maps by the operator. This will have a major impact on the result. Firstly, the reproducibility will be weak due to the small differences between different tissue within the knee. It is small margins between what belongs to the cartilage and what belongs to other structures. Failure to segment will include more or less values and also include values that not belongs to cartilage. Secondly, the T2 map is created from a multi-echo spin echo series included several anatomical images. The ROIs are drawing either on the map or on the source image (the anatomical image) where the source image is to prefer because it is easier to distinguish the different structures. In order to provide a good result, good image quality is needed. The image quality will impact the ability to distinguish the cartilage from other structures and make the segmentation of the ROI more or less challenging. The better the image quality, the easier to segment, and the greater the likelihood of analyzing the correct structures. With poor image quality, it would be difficult to determine the cartilage-bone interface and the cartilage surface. Also, if other factors affect the image quality, such as MRI signal changes or motion artefacts, the segment of the ROI will be challenging. Finally, the operator must also have a knowledge of the anatomy of the knee, knowing what corresponds to cartilage and what corresponds to other tissues in the knee. Without this knowledge, it would be difficult to assess the right region. The result is also highly reflected in how carefully the ROIs are
made. The drawing of ROIs is a very precise task and therefore requires some skills. All these aspects would have a high impact on the quantitative results and may be a disadvantage that complicates the assessment using T2 mapping and affecting the results negatively.
Comparison between the T2 mapping and the standard MRI assessment
In the T2 mapping, only one sequence is included and used for the assessment. The advantage of this is that it is possible to assess the knee from one sequence. The outcomes from the T2 map showed similar findings as to the standard MRI assessment, which shows that this method also is useful to estimate damage in the knee cartilage. Furthermore, an additional advantage of the T2 map comparing to the standard MRI assessment is that T2 map also can be analyzed quantitatively. T2 mapping provides both a visual evaluation with high image quality and quantitative analysis, by drawing ROIs on the map.
The disadvantage of the T2 map is that you only have one orientation of the knee; in this study, a sagittal was included. The extent and size of the lesion are therefore difficult to determine, and you cannot compare the findings with other sequences and orientation as in the standard MRI assessment. In the T2 map, you only can measure the length of the lesion in one slide, and this does not give any information of the entire lesion. The variability of the averages T2 values is also a disadvantage. This makes it difficult to find standard values for different tissues and what is normal and abnormal. It also makes it difficult to compare different subjects with each other. Furthermore, T2 values vary with magnetic field strengths and depend on the type of used sequences, which also complicate the standardization of T2 values.
In the standard assessment, there are instead several sequences included and all is needed for a full damage assessment, based on Episurfs method. The advantages of these are that the different sequences visualize the knee in different orientation, which results in a better overview of the knee. Furthermore, several sequences make it possible to compare the findings with each other and enable a better assessment. The damages marking process used in the standard MRI assessment to assess the cartilage makes it possible to mark out the lesion and together with all different sequences, the extent of the lesion is accessible. The software used in the standard MRI assessment also provides the ability to create and visualize the segmented structures of the knee, which gives a clear overview of the damage, as shown in section 3.2.
The scan time for the multi-echo sequence used for creating the T2 map in this thesis was approximately 8-15 minutes. From this scan, one can get one PD-weighed image and one T2-weighted image. The scan time for the standard MRI knee protocol used in this thesis is approximately 15-20 minutes, where one 2D sequence takes around 4 minutes.
To perform a T2 mapping, a further MRI sequence is needed in addition to the sequences already included in the standard MRI protocol. As mention above, the scan time for this sequence is quite long and will extend the total scan time. Extended scan time is, however, not to perform. It leads to higher cost, fewer examinations per day, and is an increased risk of artefacts, especially motion artefacts. In order to reduce the scan time, the number of different TE times can be reduced. This will, however, reduce the spatial resolution in the images. The scan time versus the resolution is as always a trade-off, and a compromise is needed for the best outcome.
Conclusion
The conclusion of this master thesis is that both methods visualize cartilage lesion in the knee joint, and both are useful to estimate damage in the knee cartilage. From an Episurf point of view, it is challenging to assess cartilage lesion only based on the T2 mapping. The T2 map shows a large variation in T2 values and the extent and size of the lesion was difficult to assess. Also, if the cartilage lesion corresponds to a bone marrow lesion, which is common, or if the cartilage is thinned and replaced with example a bone overgrowth, the assessment of the map becomes more difficult. Furthermore, T2-mapping requires an additional sequence, which also has a slightly longer scan time, and result in an extra step in the process, since it is not just to do a T2 mapping of the available sequences.
The T2 map is also user-dependent because you manually draw the ROI and some skill is required. The ROI was necessary for the quantitative analysis, to calculate the mean value and standard deviation and it was also necessary to distinguish the cartilage from other structures. The T2 map could be analyzed without drawing any ROIs, but for the more accurate analysis of the cartilage, the ROIs was a good complement.
Based on this study, the best results were found when using T2 mapping together with the standard anatom-ical images. The best scenario should be to use the T2 mapping together with some of the sequences from the standard MRI assessment. T2 mapping is an easy way to identified cartilage changes but not useful to estimate the size of the lesion, for which the anatomic images are helpful. T2 mapping is a useful supplement to use with other methods or to quickly analyze the cartilage and see differences without deeper analyzes. The software IntelliSpace portal is available on some MRI machines and can be used in connection with MR samplings. It can, therefore, be useful for quick analysis.
Chapter 5
Future work
Today, T2 mapping is not commonly used in the diagnostics, and there are no specific uses yet. Many studies have, however, now been done on T2 mapping and it is now an upcoming technique. The technique is an interesting method, and the studies have shown the usefulness of assessing the healthiness of cartilage. T2 mapping will probably be more useful with more applications areas in the near future. It is therefore exciting to keep an eye on the technology and its development.
In this study, it was limited cases having a cartilage lesion, only three subjects and only one with a cartilage lesion not affected by a bone marrow lesion. This resulted in limited analyzes when it comes to comparing the different methods to analyze the usefulness to identified cartilage lesion. It would, therefore, be interesting to do more analyzes with T2 mapping of specific cartilage lesions, such as focal damage, OCD, ON, and mild osteoarthritis, to see and analyze the different outcomes and its differences.
An interesting area with T2 mapping would also be to analyze cases that are difficult to assess, for ex-ample, where the anatomical images only show vague indications for a cartilage lesion. T2 mapping should identify cartilage lesion in an earlier stage, and when it is vague indications in the anatomical images, it might be the beginning of damage. Furthermore, when it is only vague indications in the anatomical images, it is difficult to know if it is within normal variations or if it is abnormal variations. In these cases, T2 mapping would be an interesting method to use. For the first, it would be easier to find small changes due to the color map and the numerical values but also the fact that T2 mapping should have the possibility to show changes within the cartilage on an earlier stage. Also, it would also be interesting to analyze the cases where the MR sequences show less damage than reality.
Appendix A
State of the art
A.1
Magnetic resonance imaging
Magnetic resonance imaging (MRI) is a non-invasive medical imaging technology used in radiology to produce detailed anatomical images of the human body. MRI uses a strong magnetic field and radio frequencies to generate images. This technique makes better images of organ and soft tissue and has a higher resolution than other medical imaging technologies without using ionizing radiation.
A.1.1
MRI principle
MRI is based on the principles of nuclear magnetic resonance (NMR), which is concerned with the magnetic properties of certain nuclei. The NMR phenomenon was first described by Bloch and Purcell in 1946.
Magnetic resonance imaging detects the magnetic moment created by single protons in hydrogen atoms. The hydrogen atoms exist naturally in the human body, especially in fat and water, and so provides the best MR signals. Each proton has a spin and when it is placed in a static magnetic field, B0, the protons will
turn and align with the externally applied field. The proton spins will align either parallel or anti-parallel to B0. These states correspond to a low energy state (parallel) and a high energy state (anti-parallel) and
are often called spin-up and spin-down[13].
The static field, B0 also causes each spinning proton or its magnetic moment to ”wobble”. The direction
of the spin tilts and rotates around the direction of the magnetic field. This is called the precession and is rotated with a fixed frequency, called the Larmor frequency [13]. The Larmor frequency depends on the magnetic field strength,B0, and the gyromagnetic ratio,γ, a constant property of the nucleus A.1. The tilting
of the precessing photon splits its magnetic vector, m, into a longitudinal component, mz pointing in the
z-direction and a transverse component, mxy pointing in the xy-direction. The concept of precession is
fundamental in MRI [14].
ω0= γ × B0 (A.1)
Where ω0 is the angular frequency (Larmor frequency), B0 the magnetic field strength and γ the
gyro-magnetic ratio constant (a constant for each nucleus). [15].
To obtain an MRI signal from the spinning protons, their energy state must be increased. It can be achieved by applying a radio frequency (RF) pulse into the tissue. The frequency of the RF pulse much match the Larmor frequency of the protons in order to affect the spins [14]. The application of an RF pulse causes the hydrogen vector to flip from the lower energy state, (longitudinal magnetization, Mz) into a
higher energy state (transverse magnetization, Mxy). The length of the RF pulse will determine the angle
the spins will be flipped, called the flip angle. By applying an RF pulse with flip angle 90◦ the spins will be pushed into the xy-plane and causes them to move into the same phase and precess together. This produces a transverse magnetism, Mxy, which rotates at the Larmor frequency ω0. After the RF pulse, when the
radio frequency source is switched off, the energy emitted and the protons return to their original equilibrium orientation, some earlier than others [13] [15]. M recovers while M decays. The spin in the xy-plane will
be effected by the local variation of the static magnetic field since it is not uniformed and to the magnetic field of surrounding neighbours. This causes the spin to precess with slightly different frequency, some faster than others, and hence be out of phase. To re-phase the spins a 180◦PR pulse can be applied to flipping the
spins in the xy-plane. The fast spins will be caught up with the slower spins and be in phase after a further time, called the echo time, TE. This causes a signal, which is used to create the MR images. It is only Mxy
that produces an MR signal, but since Mxy is produced by tipping Mzthe MR signal depends on the value
of Mz directly before the applied RF pulse [14].
A.1.2
Relaxation processes
An important aspect of the MRI phenomenon is the relaxation, the return of an excited spin to its equilibrium state. Relaxation is the process by which the protons release the energy that they absorbed from the RF pulse. There are two separate processes that work along the longitudinal axis, T1-relaxation and transverse axis, T2-relaxation. This process is time-dependent and provides the main mechanism for image contrast. Both T1 and T2 reflect the physical and chemical properties of the tissue and depend on the structure of the environment and the motion of protons [13] [15].
T1-relaxation time
T1-relaxation is the relaxation process that describes the recovery of the Mz. The T1-relaxation time is the
time required for the z component of the net magnetization vector, Mzto return to 63% of its equilibrium.
T1 reflects the exchanges of energy between the excited proton and its surroundings, meaning that individual spins or protons change from a higher energy state to a lower [15]. The recovery is described by the equation:
Mz= |M0| 1 − exp−t T1 (A.2) T2-relaxation time
T2-relaxation is the relaxation process that describes the decays of the Mxy. The T2-relaxation time is the
time required for the transverse components of the net magnetization vector, Mxy to decay to 37% of its
equilibrium. T2 refers to an energy transfer from an excited proton to another nearby proton. It occurs since the protons are in constant motion due to the spinning and each proton experiences an additional magnetic field produced by each neighbouring protons [15]. The T2-relaxation time is described by the equation:
|Mxy| = |M0|exp
−t T2
(A.3)
A.1.3
MRI image
To reconstruct an MR image, it is necessary to determine spatial encoding. This is achieved by applying magnetic field gradients in three slightly different ways; slice selection (a selective excitation), frequency encoding and phase encoding. This gives each signal a unique combination of a slice, phase, and frequency encoding and making it possible to determine its original spatial location in the object [16].
Image contrast
The MR image contrast depends on three main components in the tissue, the proton density (PD), T1 relaxation time (T1) and T2 relaxation time (T2) and is controlled by two parameters, TR (repetition time) and TE (echo time). TR is defined as the time to repeat the sequence and TE the time to echo. By applying different TR and TE, different weighting of the images can be achieved [14]. Table A.1 shows the choice of TR and TE to achieve the different weightings and the effects of the different weighting. By considering the individual terms in Table A.1, it is clear that T1 effects are connected to TR and T2 effects to TE.
Table A.1: T1-, T2- and PD-weighted effects Weighting Properties Result
T1-weighting TR and TE short Short T1 = bright T2-weighting TR and TE long Long T2 = bright PD-weighting Short TE and long TR High PD = bright
To achieve high contrast, the signal differences between the tissue and its surroundings should be as large as possible. The signals are affected by the different T1 and T2 times of the tissue. Tissue with short T1 or long T2 will have a large signal and appear bright on the MR image. Tissue with long T1 and short T2 will instead weak signal and appear dark. On a T1-weighted image, fat appears bright and water dark. On a T2-weighted image, water appears bright and fat dark [14]. T1-weighted images are useful for the anatomic structure of the body and T2-weighted images for identifying pathological processes such as degenerative changes, edema, infection and inflammation. The T2-weighted images provide good contrast between the different tissues or structures in the knee [17].
(a) A sagittal T1-weighted image (b) A sagittal T2-weighted image
Figure A.1: T1- and T2-weighted images of the knee
Another way to improve the contrast is to saturate signals from spins that are visualized on an image, typically spins from fat or water. By using a fat suppression pulse sequence or a fat saturation (FS), an additional RF pulse is applied with the same resonant frequency as fat, to suppress signal from fat protons. The fat signal will then be removed and making the signal from water sharper [15]. Figure A.10c show an PD-weighted fat saturation (FS) MR images.
Figure A.2: A sagittal PD-weighted images of the knee with fat saturation
2D and 3D sequences
In MRI both two-dimensional (2D) and three-dimensional (3D) techniques are used to obtain images. The 2D scan provides images with high resolution in only one plane or direction, while 3D scan provides high resolution images in all three planes and have the possibility to reformat into any direction.
The 2D sequences provide good contrast between tissues and clear delineation of lesions in knee cartilage, which gives the 2D images high diagnostic value. The 3D images are useful for reconstruction such as segmentation. However, 3D images are time consuming and are unreliable for assessing other joint structures such as the menisci, ligaments, and bone. In order to make the best assessment, both 2D and 3D images are included in the MRI protocols [18].
Imaging Planes
The images in MRI are obtained in three different planes, sagittal plane (Sag), coronal plane (Cor) and axial/transverse plane (Ax/Tra). For the assessment of the articular cartilage, the sagittal sequences are the most important and provide the best overview of the knee and the cartilage. The coronal sequences visualized the condyles very well and the edge of the cartilage, which gives a good overview of the knee. The transversal sequences are important for the assessment of the trochlea area [1].
(a) Sagittal view (b) Coronal view (c) Transverse view
Figure A.4: Images of the knee joint in the three different imaging planes
A.2
Articular Cartilage
A.2.1
Structure of articular cartilage
Articular cartilage is a form of hyaline cartilage and covers the end of the bone on any joint. It is composed of the extracellular matrix and includes water (65-85%), type II collagen (15-20%) and proteoglycans (5-10%). The function is to absorb shock and provide a smooth surface to facilitate motion. The cartilage layer is normally 2-4 mm thick and varying across the areas of the joint. Articular cartilage has no ability to regenerate and heal after injury because it does not contain any blood vessels [19]. In the knee, articular cartilage covers the ends of the femur, the top of the tibia, and the back of the patella (Figure A.5).
Figure A.5: A side view of the knee joint.Reprinted with permission from Myers Sports Medicine & Or-thopoedic Center
Articular cartilage has an organized layered structure that is structured into four zones, from the articu-lating surface down to the subchondral bone (Figure A.6). The superficial, or tangential, zone is the upper layer and makes up 10%- 20% of the articular cartilage thickness. The collagen fibers are tightly packed and oriented parallel to the articular surface. This zone is responsible for the protection of shear stress, abrasions, and fractures of deeper layers. The middle, transitional, zone makes up 40%-60% of the total thickness and consists of spherical cells, proteoglycans, and thicker collagen fibrils. The collagen fibers are loosely packed
and randomly oriented. The middle zone provides an anatomic and functional bridge between the superfi-cial and deep zones. The deep zone represents approximately 30% of the articular cartilage thickness. It contains of collagen fibers that are well organized in a perpendicular orientation to the articular surface and anchored in the underlying subchondral bone. This gives it an ability to resist compressive force which is the deep zones functionality. The deepest cartilage layer that distinguishes the articular cartilage from the subchondral bone, is the calcified zone. This zone minimizes the stiffness gradient between the rigid bone and the cartilage [19] [20].
Figure A.6: Schematic, cross-sectional diagram of healthy articular cartilage: A, cellular organization in the zones of articular cartilage; B, collagen fiber architecture (Copyright American Academy of Orthopaedic Surgeons. Reprinted with permission from the Journal of the American Academy of Orthopaedic Surgeons, 1994;2:192-201)
A.2.2
Pathology of articular cartilage
Pain in the knee joint is a common condition that affects people of all ages. Since cartilage is avascular and does not contain nerves the pain is not connected to the cartilage. Therefore, the most cartilage lesion is accompanied by a bone lesion or a reaction in the bone. Due to this, a large percentage of the population probably has, to some extent, damaged cartilage and remain undiagnosed until the damaged joint progresses to posttraumatic osteoarthritis [21].
An articular cartilage injury or a chondral injury may occur as the result of a sudden injury, an overuse injury, an underlying condition such as arthritis or wear and tear over many years. The cartilage lesion can extend partly through the cartilage or all the way to the underlying bone. There are two distinct chondral injury, focal lesions and degenerative lesions. Focal lesions are well-delineated defects, usually caused by trauma, osteochondritis dissecans or osteonecrosis. Degenerative defects are typically poorly demarcated and usually caused as a result of ligament instability, meniscal injuries, malalignment or osteoarthritis. The International Cartilage Repair Society (ICRS) has developed a classification system for grading the cartilage lesions based on the depth of the lesion, see table A.2 [22].
Table A.2: The International Cartilage Repair Society (ICRS) Cartilage Lesion Classification System. ICRS Grade 0 Normal
ICRS Grade 1 Nearly Normal (soft indentation and/or superficial fissures and cracks) ICRS Grade 2 Abnormal (lesions extending down to <50% of cartilage depth) ICRS Grade 3 Severely Abnormal (cartilage defects >50% of cartilage depth) ICRS Grade 4 Severely Abnormal (through the subchondral bone)
Osteoarthritis
Osteoarthritis (OA) is the most common chronic condition of the joints and affect millions of people world-wide. It occurs people of all ages but is most common in people older than 65. OA characterized by a breakdown of the cartilage and the underlying bone. In the final stage of OA, the cartilage breaks down completely and causes the bones within the joint to rub together. The knee joint osteoarthritis is classified into five grades, from space narrowing to bone attritions. This disease causes pain, swelling, and problems moving the joint [23]. Figure A.7 shows two different grades of OA damage.
(a) Lower grades OA (b) Higher grades OA
Figure A.7: Two coronal PD FS MR images with different grades of OA damages.
Osteochondritis Dissecans
Osteochondritis dissecans (OCD) is a lesion of the articular cartilage and the underlying bone. It occurs when a small fragment of bone begins to separate from its surrounding region. This can result in a loose body [23]. Figure A.8 shows an MRI image of an OCD damage whit a loose body that has been identified on the medial side of the knee [23].
Osteonecrosis
Osteonecrosis (ON) is a joint disease caused by decreased blood flow. Due to this insufficient blood flow, the bone cells start to die and break down. As a result of the collapses of bone, the articular cartilage covering the bone also collapse and may lead to osteoarthritis [23]. Figure A.9 shows an MRI of an ON with underlying cartilage lesion [23].
Figure A.9: Osteonecrosis, Coronal PD FS-image
Images sequences
To assess the articular cartilage, it is preferred to use proton density (PD), figure A.10a or T2-weighted sequences, figure A.10b. These sequences are good for the knee anatomy and to distinguish rifts in the cartilage. In these sequences, there are large differences in grey value between cartilage and fluid which makes it clear where synovial fluid has penetrated. The best distinguished between tissues within the knee gives with fat saturation. With these sequences, cartilage will appear in intermediate signal (grey), water in hypersignal (white) and subchondral bone in hypo signal (black), figure A.10c [1].
(a) Sagittal PD image (b) Sagittal T2 image (c) Sagittal PD image with FS
Figure A.10: Different between a PD MRI image, a T2 MRI image and a PD MRI image with fat saturation
A.3
T2 mapping
Progress in MR imaging for articular cartilage assessment has developed to parametric mapping techniques. These parametric techniques visualize the biochemical and biophysical changes of articular cartilage. One of these tissue-specific time parameter used to distinguish normal and abnormal tissue is the T2-relaxation
