• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    ADC map transformation procedure. The invention describes a method for transforming a first ADC mapto of a patient obtained by a first value b2a into a second ADC mapb of said patient similar to that obtained by a second value b2b, where the value b1 it is on both maps ADCa and ADCb is common. For each pixel of the first ADC mapto the procedure comprises: obtaining the ADC valueto of the pixel of the first ADC mapto ; determine what type of tissue the pixel on the first ADC mapto belongs to; and determine the ADCb value of the corresponding pixel of the second ADC mapb using one of the following formulas: (Machine-translation by Google Translate, not legally binding)
    公开号:ES2724599A1
    申请号:ES201830215
    申请日:2018-03-06
    公开日:2019-09-12
    发明作者:Guirado Félix Navarro；Cueto José Abelardo Viera；Garcia Beatriz Asenjo
    申请人:Servicio Andaluz de Salud；
    IPC主号:
    专利说明:

    [0001]
    [0002] ADC map transformation procedure
    [0003]
    [0004] OBJECT OF THE INVENTION
    [0005]
    [0006] The present invention belongs to the field of image visualization techniques based on the use of Magnetic Resonance Imaging (MRI).
    [0007]
    [0008] The object of the present invention is a new method that allows the transformation of a first ADC map obtained using a first b2a value during the image acquisition process into a second ADC map similar to that obtained using a second b2b value.
    [0009]
    [0010] BACKGROUND OF THE INVENTION
    [0011]
    [0012] The technique of diffusion-enhanced magnetic resonance imaging (DWI, Diffusion Weighted Images) allows to obtain representative parameters of the differences in the diffusion capacity of water molecules in the patient's tissues. Specifically, a representative parameter of said diffusion capacity that is obtained from this technique is the apparent diffusion coefficient (ADC, Apparent Diffusion Coefficient). Because the diffusion capacity of water molecules is inversely related to the number of cells per unit volume, the use of this parameter allows differentiating tissue types based on said water diffusion capacity. In fact, since, for example, tumors have a high concentration of cells due to aberrant growth, it can be interpreted that a tissue where water molecules have a low diffusion capacity corresponds to cancerous tissue. The information obtained by this technique is shown through the generation of the so-called ADC maps, where the brightness of each pixel corresponds to the value of the ADC apparent diffusion coefficient in the corresponding voxel.
    [0013]
    [0014] There are currently recommendations guides for the use of ADC maps for the detection of cancerous tissues. As an example, we can mention the article by Padhani AR, Liu G, Koh DM, and Chenevert TL, et al entitled "Diffusion-weighted magnetic resonance imaging as a biomarker cancer: consensus and recommendations", Neoplasia.
    [0015] 2009; 11 (2): 102-125. Fig. 1 shows an example of an ADC map where you can appreciate the different types of tissues depending on the diffusion capacity of water molecules.
    [0016]
    [0017] Obtaining these ADC maps is mainly done by emitting two opposite gradient pulses that vary with position and whose offset gradients are proportional to a certain coefficient b selected by the radiologist. Fig. 2 shows a semi-logarithmic scale curve (solid line) that represents, in a specific voxel of the patient's tissue, the response S (b) obtained as a function of the coefficient b used in the acquisition. As can be seen, the curve S (b) has two differentiated portions: for low b values, a first curved portion from greater to lesser slope; and for high b values, a second essentially straight portion of smaller slope than the first portion. The first portion is caused by the presence of perfusion in the voxel in question. The more important the perfusion mechanism in the movement of water molecules inside the voxel, the more steep the initial portion of the curve has. On the contrary, in the absence of perfusion in the voxel, the curve would approximately take the form of a line that passes through the origin of coordinates (dashed line).
    [0018]
    [0019] Currently, the curve is modeled assuming a Brownian movement of water molecules inside the voxel. That is, it is based on the assumption that the movement of water molecules within the voxel is isotropic and that the only cause of movement is diffusion. The equation that governs the amplitude of the image in the current ADC maps is thus obtained:
    [0020]
    [0021] S ( b) = S0e ~ bADC (1)
    [0022]
    [0023] where:
    [0024] So is the amplitude of the signal in case of not applying diffusion gradients, b the parameter that defines the power of the diffusion gradient, and
    [0025] ADC the average diffusion coefficient of water molecules within the voxel, that is, the Apparent Diffusion Coefficient (ADC).
    [0026]
    [0027] More specifically, to obtain ADC maps in which the intensity of each pixel reflects the ADC apparent diffusion coefficient value in the corresponding voxel, two diffusion-enhanced images with two different diffusion gradients are used, with powers b1 and b2 , and it operates with the relationship between them. For each pixel the following operation:
    [0028]
    [0029]
    [0030]
    [0031]
    [0032] That is, the value of the ADC apparent diffusion coefficient in a voxel is defined as the slope of the line that joins s (b1) with s (b2). It is evident, therefore, that the selection of the values of b has an important impact on the ADC maps obtained. For historical reasons, b1 = 0 s / mm2 is currently used in clinical practice protocols. For its part, the value of b2 depends on the type of tissue and should be as large as possible. At the same time, it must be low enough that the ratio between the signal and the noise (SNR) of the echo returned for the intensity gradient b2 is acceptable for the calculation of ADC.
    [0033]
    [0034] In short, the curve S (b), with diffusion and perfusion effects, is approaching along a line that joins S (0) and S (b2) whose slope is the apparent diffusion coefficient ADC. This approach, therefore, ignores the existence of perfusion.
    [0035]
    [0036] Fig. 3 graphically shows how the ADC apparent diffusion coefficient is obtained for two different b2 values in a particular voxel. Specifically, Fig. 3 shows the lines resulting from applying equation (1) to values of (b1, b2) respectively of (b1a, b2a) = (0 s / mm2, 500 s / mm2) and of (b1b, b2b) = (0 s / mm2, 1000 s / mm2). The slope of the dashed line of Fig. 3 corresponds to the ADC value calculated with (b1a, b2a) = (0 s / mm2, 500 s / mm2), and the slope of the dotted line of Fig. 3 corresponds to ADC value calculated with (b1b, b2b) = (0 s / mm2, 1000 s / mm2). As you can see, the slope of these two lines is not the same. The reason for the appearance of this difference is the assumption made in equation (1) that diffusion is the only cause of movement when, as mentioned, the reality is that movements within a voxel are due so much to diffusion as to infusion. For these reasons, the curve S (b) is not a line that passes through the origin of coordinates, and as a consequence there are considerable differences between ADC maps obtained using different values of b2 that, in each pixel, can reach up to 25%.
    [0037]
    [0038] Currently, each hospital is acquiring ADC maps using a value of b2 different from that used in other hospitals. Although this fact is not particularly important in the qualitative interpretation of the ADC maps of a particular patient In a private hospital, it is critical when conducting quantitative comparative studies. Data relating to ADC maps obtained with a first value of b2 in a given hospital cannot be compared with data related to ADC maps obtained with a second value of b2 in a different hospital. This constitutes a major drawback in the context of the investigation and a burden for the use of quantitative data in clinical practice.
    [0039]
    [0040] DESCRIPTION OF THE INVENTION
    [0041]
    [0042] The present invention describes a method that allows the transformation of a first ADC map obtained using a first determined b2a value into a second ADC map similar to that obtained using a second b2b value. For this, a model called IVIM (Intra Voxel Incoherent Motion) is used in which the S (b) curve is approached more precisely through two lines: a first line with a greater slope approximates the initial portion of the curve corresponding to the perfusion mechanism; and a second straight with a smaller slope approximates the rest of the curve. It is a model that is closer to the real form that the S (b) curve adopts when there is perfusion than the model described above in relation to equation (1). The mathematical development described below allows, through the IVIM model, to obtain an expression by which the ADC value of each ADC map pixel can be transformed from an initial value corresponding to the first b2a value into a value similar to that which is would have obtained using a second b2b value.
    [0043]
    [0044] Hereinafter, reference will be made to a first ADCa map calculated using a first b2a value, and a second ADCb map that one wishes to obtain and corresponding to a second b2b value. In both cases, the corresponding b1 (that is, both b1a and b1b) is assumed to be common and usually equal to 0.
    [0045]
    [0046] IVIM model
    [0047]
    [0048] Next, the IVIM model used in this invention for the transformation of ADC maps is briefly described. As mentioned, this model takes into account the presence in the voxel of two very different movements: diffusion in the extracellular space and microvascular perfusion. The IVIM model is defined by the following equation:
    [0049] S ( b) = S 0 ( a - f) e - b D f - e ~ bD *) (3)
    [0050]
    [0051] where:
    [0052] f is the perfusion fraction (in as much as one),
    [0053] D is the actual diffusion coefficient, and
    [0054] D * is the pseudo broadcast coefficient.
    [0055]
    [0056] The parameter f is interpreted as the importance of perfusion in the movement of water molecules of a voxel. For very well irrigated tissues f is close to 0.3, while for poorly vascularized tissues f is close to 0. On the other hand, the parameter D * represents perfusion within the vessels. This movement is much faster than that of extracellular and extravasal diffusion. Therefore, the values of D * are usually of an order of magnitude greater than the values of D.
    [0057]
    [0058] Fig. 4 graphically shows, in a continuous line, the shape taken by a curve obtained according to the IVIM model defined by equation (3). This curve has two distinct portions: for low b values, a first straight portion where perfusion has a more pronounced effect; and for high b values, a second straight portion with a smaller slope where perfusion has a lower effect. As can be seen with the naked eye by comparing this curve with the S (b) curve, in a broken line, the IVIM model defined by equation (3) is much more precise than the ADC model defined by equation (1). The first portion of the IVIM model curve approaches the first portion of the S (b) curve quite well, and the second portion of the IVIM model curve is almost coincident with the second portion of the S (b) curve.
    [0059]
    [0060] Proposed Procedure
    [0061]
    [0062] The invention proposes to use the IVIM model for, starting from the knowledge of the slope of a line that passes through the origin of coordinates and a point of the curve S (b) corresponding to a certain b2a (that is, starting from the apparent diffusion coefficient ADCa obtained for said b2a), determine the slope of a line that passes through the origin of coordinates and another point of the curve S (b) corresponding to a certain different b2b (that is, determine the apparent diffusion coefficient ADCb that would be obtained for said value of b2b different from b2a). By carrying out this procedure for each pixel of the first ADCa map, it is possible to obtain an ADCb map similar to the one obtained. using the b2b value during the acquisition process. Therefore, the use of this procedure would allow comparative studies using ADC map data obtained in different hospitals with different values of b2.
    [0063]
    [0064] The present invention describes a method for the transformation of ADC maps, which allows transforming a first ADCa map of a patient obtained by a first b2a value into a second ADCb map of said patient similar to that obtained by a second b2b value. The value b1 on both ADCa and ADCb maps is assumed to be common and, usually, equal to 0. This procedure comprises, for each pixel of the first ADCa map, perform the following steps:
    [0065]
    [0066] 1. Obtain the ADCa value of the pixel of the first ADCa map.
    [0067]
    [0068] 2. Determine what type of tissue the pixel of the first ADCa map belongs to.
    [0069]
    [0070] 3. Determine the ADCb value of the corresponding pixel of the second ADCb map using one of the following formulas:
    [0071]
    [0072]
    [0073]
    [0074]
    [0075] O well
    [0076]
    [0077] ADCb = Dtej gone {1 - ) ADcaí $ (5)
    [0078]
    [0079] where
    [0080] ftejido is the fraction of perfusion in the type of tissue to which the pixel belongs; and Dtissing is the actual diffusion coefficient in the type of tissue to which the pixel belongs.
    [0081]
    [0082] The determination of the type of tissue to which the analyzed pixel belongs is carried out through tools currently known in this field. For example, various procedures based on image analysis are known that allow differentiation of tissues through segmentation techniques based on pixel brightness levels, seed growth techniques, etc.
    [0083] Additionally, it is necessary to know in advance the value of the parameters set or corresponding to each type of fabric. Indeed, for each pixel of the first ADCa map, the value of the woven or woven in that pixel will depend on the type of tissue to which the corresponding voxel belongs, since not all tissues have the same characteristics in terms of diffusion capacity and / or perfusion of water molecules. Note that the value of the parameters set or Dyed is independent of the value of b2 used in the acquisition of the images, since it is a characteristic of the patient's tissue.
    [0084]
    [0085] The value of the parameters woven or woven for each type of tissue present in the area of the patient being analyzed can be known in advance. For example, it is possible to obtain them from known databases or existing bibliography. Alternatively, it is possible to carry out a study prior to the transformation of the ADC map of a patient to determine average values of the parameters woven, Dyed and D * tissue for each type of tissue in a given sample of patients.
    [0086]
    [0087] Therefore, according to a preferred embodiment of the invention, the method further comprises the previous step of determining the value of the parameters woven or woven of each type of tissue from data from a sample of patients.
    [0088]
    [0089] More preferably, said previous step of determining the value of the woven or woven parameters of each type of tissue from data from a sample of patients comprises the following steps:
    [0090]
    [0091] i) For each patient in the patient sample, from several DWI images acquired using different values of parameter b2, with common b1, obtain at least four pairs of data (S (b2), b2) of pixels corresponding to one type of tissue.
    [0092]
    [0093] ii) For each patient in the patient sample, determine the value of the parameters patient tissue, patient tissue, and D * patient tissue of each pixel of said tissue type by adjusting equation (3) described above:
    [0094]
    [0095]
    [0096] iii) Averaging, among all the patients in the patient sample and the pixels of said type of tissue, the values of the parameters patient tissue, patient tissue, and D * patient tissue to obtain average values of tissue, tissue, and D * tissue of said type of fabric
    [0097]
    [0098] That is, for each patient at least four images are obtained, for example respectively corresponding to values of b2 = 0, b2 low, b2 moderately high, and b2 high. For the pixels corresponding to a certain type of tissue, the parameters of patient tissue, patient tissue, and D * patient tissue of that particular patient are then determined. By averaging the values of the parameters obtained among all the pixels of that type of tissue and among all the patients of a sufficiently large patient sample, an average value of the parameters set, Dyed, and D * tissue for that type of tissue is obtained .
    [0099]
    [0100] BRIEF DESCRIPTION OF THE FIGURES
    [0101]
    [0102] Fig. 1 shows an example of an ADC map according to the prior art.
    [0103]
    [0104] Fig. 2 graphically shows an example of curve S (b) corresponding to a given voxel where the effect of perfusion can be seen.
    [0105]
    [0106] Fig. 3 graphically shows the geometric interpretation of the ADC value for a given voxel.
    [0107]
    [0108] Fig. 4 graphically shows the shape of the IVIM model used in the present invention.
    [0109]
    [0110] PREFERRED EMBODIMENT OF THE INVENTION
    [0111]
    [0112] The mathematical development from which the invention arises is described, as well as a particular example of calculating an ADCb value from a known ADCa value using the method of the present invention.
    [0113]
    [0114] Mathematical justification
    [0115]
    [0116] As previously mentioned, equation (3) corresponds to the IVIM model:
    [0117]
    [0118] S ( b) = S0 ((1 - f) e ~ b-Dw *> f te ..do . E -bD * tejld0 ^
    [0119] However, this equation can be simplified taking into account that, for high b2 values, D * tissue >> Dyed. In fact, a b1 equal to 0 s / mm2 and a b2 that can adopt values between 650 and 3000 s / mm2 are normally used for the acquisition of ADC maps. For values of b2 within this range, the actual diffusion coefficient Dtissed is much smaller than the pseudodiffusion coefficient D * tissue. The reason is that the speed at which blood circulates through the microvessels present in the voxel is much greater than that at which the water molecules diffuse through the extravasal and extracellular tissue.
    [0120]
    [0121] Therefore, taking into account that D * tissue >> Dyed, the above equation can be simplified:
    [0122]
    [0123] Now, taking into account the ADC definition of equation (1):
    [0124]
    [0125] S ( b ) = S0e ~ bADC
    [0126]
    [0127] If we match the two previous expressions:
    [0128]
    [0129] S0e ~ b ADC = S0 (1 - woven) e ~ b'Dte¡ldo
    [0130]
    [0131] Operating, you get:
    [0132]
    [0133] ADC (b) = ^ tej id o + ^ ln ^ 7 ^)
    [0134]
    [0135] Therefore, for two ADC maps obtained using different values of b2 (b2a, b2b), you get:
    [0136]
    [0137]
    [0138]
    [0139]
    [0140] By operating with these two equations, it is possible to determine ADC (b2b) based on the parameter set or depending on the parameter set. Equations (4) and (5) mentioned above are obtained:
    [0141]
    [0142]
    [0143]
    [0144] Description of a particular example
    [0145]
    [0146] For a test subject, anatomical reference images enhanced in high resolution T1 are obtained and subcortical structures are segmented. To calculate the ADCa and ADCb maps, the images enhanced in diffusion are obtained using ba = (0 s / mm2, 1500 s / mm2) and bb = (0 s / mm2, 1000 s / mm2), respectively. In order to obtain the parameters f, D and D * of the different structures, diffusion-enhanced images are obtained with values b = (0 s / mm2, 15 s / mm2, 30 s / mm2, 65 s / mm2, 300 s / mm2, 685 s / mm2, 1000 s / mm2, 1175 s / mm2, 1350 s / mm2, 1500 s / mm2).
    [0147]
    [0148] The IVIM parameters obtained for the putamen are f = 0.072, D = 690 mm2 / s and D * = 18373 mm2 / s while the average ADC values are 732 mm2 / s for b = 1500 s / mm2 and 770 mm2 / s for b = 1000 s / mm2:
    [0149]
    [0150] Applying equations (4) and (5) on the ADC map pixels obtained with b = 1500 s / mm2 to correct the value of b and make it equivalent to that obtained with b = 1000 s / mm2, an average value is obtained in the putamen 756 mm2 / s and 784 mm2 / s, respectively.
    [0151]
    [0152] Therefore, applying the proposed method, the difference between the maps obtained with b = 1500 s / mm2 and b = 1000 s / mm2 goes from -4.94% to -1.7% and 1.8% if using the equations (4) and (5), respectively.
    权利要求:
    Claims (3)
    [1]
    1. ADC map transformation procedure, to transform a first ADCa map of a patient obtained by a first b2a value into a second ADCb map of said patient similar to that obtained by a second b2b value, where the b1 value is in Both ADCa and ADCb maps are common, characterized in that they comprise, for each pixel of the first ADCa map, performing the following steps:
    - obtain the ADCa value of the pixel of the first ADCa map;
    - determine what type of tissue the pixel of the first ADCa map belongs to; Y
    - determine the ADCb value of the corresponding pixel of the second ADCb map using one of the following formulas:

    [2]
    2. The ADC map transformation method according to claim 1, further comprising the previous step of determining the value of the woven or woven parameters of each type of tissue from data from a patient sample.
    [3]
    3. ADC map transformation method according to claim 2, wherein the step of determining the value of the parameters woven and Dyed of each type of tissue from data from a sample of patients comprises the following steps:
    - for each patient in the patient sample, from several DWI images acquired using different values of parameter b2, with common b1, obtain at least four pairs of data (S (b2), b2) of pixels corresponding to a type of tissue;
    - for each patient in the patient sample, determine the value of the parameters patient tissue, patient tissue, and D * patient tissue of each pixel of said tissue type by adjusting the equation:
    S ( b) = S0 ( a - f) e - b D f - e ~ bD *)
    - and averaging, among all the patients in the patient sample and all the pixels of said type of tissue, the values of the parameters patient, patient tissue, and patient * D to obtain average values of tissue, tissue, and D * tissue of said type of tissue.
    类似技术:
    公开号 | 公开日 | 专利标题
    O’Donnell et al.2002|New approaches to estimation of white matter connectivity in diffusion tensor MRI: Elliptic PDEs and geodesics in a tensor-warped space
    JP6851359B2|2021-03-31|Volume measurement Comprehensive cardiovascular analysis by phase contrast MRI
    Cline et al.1987|3D reconstruction of the brain from magnetic resonance images using a connectivity algorithm
    Roebroeck et al.2008|High-resolution diffusion tensor imaging and tractography of the human optic chiasm at 9.4 T
    CN105488804B|2018-06-12|Brain ASL, SPECT and the method and system of MRI image registration fusion Conjoint Analysis
    Kreher et al.2008|Connecting and merging fibres: pathway extraction by combining probability maps
    CN104116523B|2016-08-03|A kind of ultrasonic image analyzes system and the method for analysis thereof
    Zarinabad et al.2012|Voxel‐wise quantification of myocardial perfusion by cardiac magnetic resonance. Feasibility and methods comparison
    US20190178968A1|2019-06-13|Resolution enhancement of diffusion imaging biomarkers in magnetic resonance imaging
    US10765349B2|2020-09-08|Magnetic resonance imaging device and method for calculating oxygen extraction fractions
    US20090324052A1|2009-12-31|Detection and localization of vascular occlusion from angiography data
    Barbieri et al.2011|Segmentation of fiber tracts based on an accuracy analysis on diffusion tensor software phantoms
    ES2896954T3|2022-02-28|Noise reduction in image data
    ES2724599B2|2020-08-21|ADC map transformation procedure
    Ehricke et al.2006|Visualizing MR diffusion tensor fields by dynamic fiber tracking and uncertainty mapping
    US9513354B2|2016-12-06|Determining electrical properties of tissue using complex magnetic resonance images
    JP6805335B2|2020-12-23|Volume presentation for planning the location of injection points
    WO2019022733A1|2019-01-31|Systems and methods for strategically acquired gradient echo imaging
    Irfanoglu et al.2014|DR-BUDDI: diffeomorphic registration for blip up-down diffusion imaging
    Zhang et al.2005|Approximating anatomical brain connectivity with diffusion tensor MRI using kernel-based diffusion simulations
    Zhang et al.2014|Multimodality neurological data visualization with multi-voi-based DTI fiber dynamic integration
    Dellani et al.2007|White matter fiber tracking computation based on diffusion tensor imaging for clinical applications
    US20190025393A1|2019-01-24|System and Method for Producing High-Resolution Magnetic Relaxation Parameter Maps
    US10429478B2|2019-10-01|Push-button vessel wall MRI with 3D scout scan
    Nimsky et al.2006|Visualization strategies for major white matter tracts for intraoperative use
    同族专利:
    公开号 | 公开日
    ES2724599B2|2020-08-21|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    US20110043206A1|2009-08-20|2011-02-24|Tokunori Kimura|Magnetic resonance imaging apparatus and rf coil unit|
    US20110085722A1|2009-10-14|2011-04-14|Thorsten Feiweier|Correction of distortions in diffusion-weighted magnetic resonance imaging|
    US20160084929A1|2014-09-18|2016-03-24|Siemens Aktiengesellschaft|Method and apparatus to correct noise effects in quantitative techniques in magnetic resonance imaging|
    法律状态:
    2019-09-12| BA2A| Patent application published|Ref document number: 2724599 Country of ref document: ES Kind code of ref document: A1 Effective date: 20190912 |
    2020-08-21| FG2A| Definitive protection|Ref document number: 2724599 Country of ref document: ES Kind code of ref document: B2 Effective date: 20200821 |
    优先权:
    申请号 | 申请日 | 专利标题
    ES201830215A|ES2724599B2|2018-03-06|2018-03-06|ADC map transformation procedure|ES201830215A| ES2724599B2|2018-03-06|2018-03-06|ADC map transformation procedure|
    [返回顶部]