Method and apparatus for characterizing breast tissue using multi-parameter MRI

Abstract

A method and apparatus for analyzing MRI data of breast tissue is described. The method comprises the steps of: acquiring a plurality of quantitative non-contrast MRI scan images of at least a portion of the breast region using different MR pulse sequences for each MRI scan image; determining a plurality of MRI metrics from the acquired quantitative MRI scans; generating a plurality of MRI metric maps using the determined metrics; calculating at least one of a breast complex map and a heterogeneity map from the metric map to display tissue characteristics of fat and non-fat tissue in the MRI image; determining a breast tissue heterogeneity characteristic from at least one of the breast complex map and the heterogeneity map; and identifying a data cluster from at least one of the breast complex map, the breast heterogeneity map, and the breast tissue heterogeneity characteristic.

Description

Technical Field

[0001] The present invention relates to the use of medical scan images, in particular MRI images, to analyze breast tissue. Background Art

[0002] Conventionally, the breast can be described as being composed of two main types of tissue: fatty tissue and fibrous functional tissue. The breast may also contain abnormalities, including cysts, fibroadenomas, and various types of cancer. These abnormalities are usually found in fibrous breast tissue. So-called "fatty" breasts have a low proportion of fibrous tissue compared to fatty tissue, while "dense" breasts have a high proportion of this fibrous tissue compared to breast fat. The term breast "fat" is a simplification of a large number of lipids including saturated, monounsaturated, and polyunsaturated types.

[0003] "Density" is a term used to refer to mechanically dense tissue and / or radiologically dense tissue (because it greatly attenuates X-rays, for example in the case of mammography or tomosynthesis). Mammography and, increasingly, tomosynthesis are the basis for screening asymptomatic people, who are usually women who are postmenopausal or undergoing menopause, collectively referred to as "elderly" women. Mammographic screening has been shown to enable early detection of tumors in older women. The physical basis of mammography is that fibrous tissue degenerates into "fat" during menopause, and fat is mostly "transparent" to X-rays (i.e., has a low attenuation rate). However, for a significant proportion of older women, their breasts are still (radiologically) dense, and for such women, mammography and tomosynthesis provide less information. For younger women, mammography and tomosynthesis provide much less information, because younger women's breasts tend to be mechanically dense, and this in turn makes the breasts radiologically dense. Breast (radiologically) density has been shown to be an important risk factor for predicting breast cancer.

[0004] Mammography and tomosynthesis volumes are often complex images that are challenging to evaluate and analyze, even for experienced radiologists. To this end, reporting and data standards have been developed and have been revised and improved over time. The most famous of these is BI-RADS (Breast Imaging Reporting and Data Standards) developed by the American College of Radiologists https: / / www.acr.org / Clinical-Resources / Reporting-and-Data-Systems / Bi-Rads. BI-RADS specifies a range of imaging modalities, including ultrasound, MRI, and mammography and tomosynthesis. Because mammography and tomosynthesis are based on population-based screening systems and are relatively low-cost, the mammography and tomosynthesis portions of the BI-RADS system are often used interchangeably with BI-RADS as a whole. In mammography and tomosynthesis, the breast is classified into one of a number of BI-RADS categories, where, for example, BI-RADS category 1 refers to a breast that is primarily composed of adipose tissue and is therefore essentially transparent to X-rays. This means that isolated dense areas in such a breast, such as possible tumors, can usually be directly detected. In contrast, BI-RADS Category 4 refers to a breast that is composed primarily of dense tissue, which makes detecting a tumor in the breast more difficult, like finding a snowball in a blizzard. Sometimes BI-RADS categories 1-4 are referred to as AD. Note that, in general, BI-RADS recognizes even more categories (e.g., 0-6); but the progression in the case of mammography and tomosynthesis is from fat-predominant, to scattered isolated areas of density, to predominantly dense, to dense-predominant. BI-RADS refers to the entire system of reporting and data standards, and an individual breast may be given a "score," which means being assigned to one of these categories. The best-known "scores" are for mammography and tomosynthesis, but scores for other imaging modalities also exist.

[0005] Breast parenchyma (functional tissue, not connective tissue) is generally highly heterogeneous. It is well known that regional differences reflect clinically important information; for example, some local areas may correspond to benign or malignant abnormalities. Similarly, regional differences may convey information about breast cancer risk. In addition, regional changes in breast tissue over time can indicate response to treatment, development of pathology, or recurrence of pathology. Finally, the amount and distribution of "dense" tissue can be used as a basis for assessing a woman's risk of breast cancer. Other bases for assessing risk include epidemiological factors (age, time of menarche, number of children, etc.) as well as genetic and epigenetic factors, such as certain gene mutations (e.g., BRCA 1, 2) and / or previous breast cancer in close relatives.

[0006] Genomic and circulating biomarker data continue to play an important role in the detection and diagnosis of breast disease. However, these data fail to convey information about regional characteristics, which is why imaging that can provide information about the phenotype of breast tissue is becoming increasingly important. Genetic profiling is often used to establish familial risk for breast disease. This risk calculation can be improved by combining genetic profiling and phenotypic classification.

[0007] Although academic groups have developed prototype CT systems, as mentioned above, X-ray imaging is overwhelmingly mammography (2D) and tomosynthesis (3D), which have fairly low X-ray doses at lower energies. These methods have been shown to be useful for postmenopausal women and are the basis of asymptomatic screening programs in many countries, including the UK. However, as mentioned above, the high X-ray attenuation of dense tissue means that mammography and tomosynthesis are of no clinical value for women with "dense" breasts. High breast density is the single greatest risk factor for breast cancer in postmenopausal women. This means that for women with dense breasts there is an increased risk not only of the presence of a tumor, but also of the risk of missing a tumor. Most premenopausal women and about half of postmenopausal women are classified as having "dense" breasts, so mammography is not appropriate for a large number of women. Ultrasound imaging is used for the breast, but the technology provides limited information and has a high false positive rate. Various other imaging methods have been explored, including infrared and microwave, but these have not been proven to be of sufficient value for widespread use.

[0008] For premenopausal women, as well as many postmenopausal women—especially, but not limited to, those with dense breasts—MRI offers many advantages. Unlike mammography / tomography, MRI enables visual examination of even the largest, densest breasts without the need for breast compression (which can be painful or uncomfortable for many women), and without the need for ionizing radiation. Unlike mammography and tomosynthesis, MRI can be used equally in premenopausal and postmenopausal women and provides clinically important soft tissue contrast. For these reasons, the use of MRI for breast imaging is increasing rapidly, and it is widely considered the most advanced breast imaging modality. Figure 1 (a) shows an MRI image 50 of a breast, with lymph nodes 51 highlighted in the image. Figure 1 (b) is an illustration of the area of ​​interest in the breast that will be imaged by MRI. The figure shows the breast tissue, sternum, lymph nodes, and secondarily areas on both sides of the chest wall including the axilla. Axillary imaging is primarily used to determine if the cancer has metastasized, as this has a significant impact on a woman's subsequent treatment.

[0009] To date, breast MRI has had two major limitations. First, and less important, MRI is costly, and imaging is more time-consuming than mammography / tomography. Innovations in image reconstruction (e.g., compressed sensing) and improved coil and magnet designs are helping to address MRI's cost issues, as is the development of "simplified" imaging protocols, which now typically require a 15-minute scan.

[0010] Second, and more importantly, is the qualitative nature of conventional breast MRI. To appreciate this, consider briefly how MRI images are formed. First, the patient is placed in a strong magnetic field. In the case of breast MRI, the woman typically lies prone on the bed of the MRI machine with her breasts suspended in a specialized breast receiver coil. Women find lying prone with the breast coil pressing against their chest, surrounding the breast, very uncomfortable. They prefer lying supine, but typically the breasts fall on the surface of the chest, making imaging difficult.

[0011] Once the patient is in the MRI magnet in the appropriate position (prone or supine), radiofrequency energy of a specific frequency (RF) is radiated toward the patient. This causes some of the protons in the tissue being imaged to change their spin state. Soon after the RF energy stops, the protons in the imaged tissue relax back to their previous state. The way they do this is measurable and is a property of the tissue in which the protons were. There are several parameters associated with MRI, but the two most important have a time dimension and are expressed in terms of T. 1 and T 2 (or in many cases the variant T 2 *) indicates. MRI equipment manufacturers have developed many different pulse sequences, each of which produces (2D or 3D) images that emphasize the contrast between certain important tissue classes. For example, the brain is primarily composed of gray matter, white matter, and cerebrospinal fluid (CSF). Depending on the specific pulse sequence selected, the contrast between gray and white matter can be emphasized, or the CSF can be visualized. In current commercial practice, most pulse sequences produce images with pixels (or voxels) that confuse two basic measurements, T. 1 and T 2 , usually one of them (for example, T 1 ) dominates, in which case the image is called T 1 weighted. This means that T 1 Contrast dominates, but T 2 Contrast still has an impact.

[0012] Even for the same woman, this mix of contrast can vary from one radiographer to another and from one machine to another. These differences manifest as variations in the appearance of the image, so it requires the radiographer to use their judgement and experience. As a result, there is a large amount of inter- and intra-operator variability (estimated to be around 30%). In particular, the appearance of abnormalities such as cysts, fibroadenomas and cancers can vary greatly depending on the woman's breast and the operating parameters chosen by the radiographer. It would be a considerable advantage if breast MRI could be quantified, as this would allow the images to be standardized between different radiographers and different machines. This would mean, for example, developing a pulse sequence that produces images with voxels that are (for example) T 1 The accurate measurement of T 1 Weighted mixture.

[0013] The difference between a weighted MRI image and a quantitative map is that a weighted image is simply an image collected using specific scan parameters designed to provide a given type of image contrast, allowing the clinician to better visualize tissue. For a quantitative map, tissue properties, such as T1 or T2, are calculated from a series of raw data scans. For example, a T2-weighted image will have a range of signal intensities, where liquid will appear white and other tissues will appear darker. But a quantitative T2 map will show the calculated T2 values ​​for different tissues.

[0014] Current practice focuses only on detecting localized abnormal areas that may correspond to cancer. However, it has been found that the contrast provided by conventional weighted sequences often does not enable the detection of tumors, especially when the tumor is small. For this reason, it is well known that most cancers grow chaotic, leaky blood vessels through a process called neoangiogenesis. Contrast agents, which are usually administered by injection, are based on large molecular paramagnetic lanthanide compounds (usually gadolinium chelates) that leak from blood vessels into the surrounding extravascular space, locally enhancing the MRI signal. Although attempts have been made to quantitatively process MRI contrast images, this is rarely (if ever) possible because it is generally impossible to accurately measure the arterial input function generated by the administration of contrast agents. In addition, many women find the feeling of contrast agents being absorbed by the body and then excreted (usually within 24 hours) disturbing. Common side effects include nausea, headache, and dizziness. It has also been shown that gadolinium is deposited in the brain, liver, skin, and bones after the use of these contrast agents, which has raised concerns about the safety of contrast-enhanced imaging.

[0015] The present invention solves the above-mentioned problems with current MRI imaging of breast tissue. Summary of the invention

[0016] According to the present invention, a method for analyzing MRI data of breast tissue is provided, comprising: acquiring a plurality of quantitative non-contrast MRI scan images of at least a portion of a breast region using different MR pulse sequences for each MRI scan image; determining a plurality of MRI metrics from the acquired quantitative MRI scans; generating a plurality of MRI metric maps using the determined metrics; calculating at least one of the following: calculating a breast composite map and a heterogeneity map from the metric maps to display tissue characteristics of fat and non-fat tissue in the MRI images; determining a breast tissue heterogeneity characteristic from at least one of the breast composite map and the heterogeneity map; and identifying a data cluster from at least one of the breast composite map, the breast heterogeneity map, and the breast tissue heterogeneity characteristic.

[0017] Further preferably, the data clusters are continuous data clusters identified across one or more MR metrics.

[0018] Preferably, the breast composite image is compared with data from a patient database to identify clusters of consecutive data. Further preferably, the data comparison allows identification of abnormal areas of breast tissue.

[0019] In an embodiment of the present invention, the breast composite map is a 3D map, wherein consecutive data clusters represent similar breast tissues. Further preferably, the breast composite map provides a spatial distribution of dense breast tissues.

[0020] In an embodiment of the invention, determination of breast tissue heterogeneity properties comprises the following steps: determining one or more of: a mean value of the density of pixels or voxels in an MRI scan, a choropleth map showing the spatial distribution of the property, a texture map of the density.

[0021] Preferably, determining characteristics of breast tissue heterogeneity comprises determining one or more of: fat content and breast composition; breast inflammation; and changes in breast tissue architecture.

[0022] Further preferably, the determined metrics generate an MRI map for each determined metric.In an embodiment of the invention, at least two determined metrics are combined to generate a further metric map, and the further metric map is used to calculate a breast composite map.

[0023] Preferably, raw values ​​of at least two determined metrics are used to generate the further metric map.

[0024] In an embodiment of the present invention, the MRI metrics include one or more of T1, corrected T1, T2, T2*, PDFF and ADC.

[0025] In another preferred embodiment of the present invention, the MRI metric map is a PDFF map or a corrected T1 map.

[0026] Preferably, the MRI scan is a 2D scan producing a 2D image consisting of a set of pixels, or a 3D scan having a 3D image consisting of voxels.

[0027] It is further preferred that the at least one MRI metric is combined with one or more other MRI metrics before generating an additional MRI metric map, the additional MRI metric map providing an improved estimate of the initial value of the at least one MRI metric.

[0028] In an embodiment of the invention, at least one metric is a T1 metric and is combined with at least one of PDFF and T2*.

[0029] In an embodiment of the present invention, the method further comprises the step of generating at least one of the following MRI metric maps: T1 map, T2 map, T2* map, ADC map.

[0030] Preferably, the T1 metric is determined using a modified Look Locker inversion recovery sequence acquisition, a shortened modified Look Locker inversion recovery sequence acquisition, or a variable flip angle sequence acquisition. In an embodiment of the present invention, a multi-echo gradient echo sequence acquisition is used to determine the T2* metric and / or the PDFF metric. In a preferred embodiment of the present invention, a multi-contrast spin echo sequence acquisition is used to determine the T2 metric.

[0031] Preferably, the ADC metric is determined using single shot or multiple shot diffusion weighted echo planar imaging.

[0032] In a preferred embodiment of the invention, the MRI scan is obtained with the patient in a prone or supine position.

[0033] In an embodiment of the present invention, the breast density map and breast tissue characteristics are used to provide a disease risk score. Preferably, the disease risk score is a score associated with the BIRADS breast density classification.

[0034] In a preferred embodiment of the present invention, MRI images are acquired within a set time period to monitor changes in breast density maps and breast tissue characteristics. Preferably, the set time period is at least 1 week.

[0035] In an embodiment of the present invention, the MRI scan is a scan of the entire breast area.

[0036] According to the present invention, there is also provided an apparatus for analyzing MRI data from breast tissue to determine breast tissue heterogeneity characteristics from a breast density map, the apparatus comprising at least one processing component arranged to perform the method as described above.

[0037] Further preferably, at least one processing component of the device for analyzing MRI data includes one or more of the following: one or more programmable components, which are arranged to execute computer program code to perform one or more steps of the above-mentioned method; and a hardware circuit system, which is arranged to perform one or more steps of the above-mentioned method.

[0038] Preferably, the apparatus further comprises at least one output component for outputting the breast density map or the determined breast tissue heterogeneity characteristics, the at least one output component comprising one or more of the following: a display device for displaying the breast density map or the determined breast tissue heterogeneity characteristics to a user; a data storage device for storing the breast density map or the determined breast tissue heterogeneity characteristics; and an interface component for transmitting the breast density map or the determined breast tissue heterogeneity characteristics to at least one external device. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Further details, aspects and embodiments of the present invention will be described by way of example only with reference to the accompanying drawings. In the accompanying drawings, the same reference numerals are used to identify the same or functionally similar elements. The elements in the various figures are shown for simplicity and clarity and are not necessarily drawn to scale.

[0040] Figure 1 (a) shows an example MRI image of breast tissue;

[0041] Figure 1 (b) is a schematic diagram of a breast showing different tissue types;

[0042] FIG. 2( a ) shows a flow chart of a method according to an embodiment of the present invention;

[0043] Figure 2(b) shows the details of the analysis steps in Figure 2(a);

[0044] Figure 3 Calculation of the proton density fat fraction (PDFF) of breast tissue according to an embodiment of the present invention is shown;

[0045] Figure 4 Calculation of quantitative breast density using a breast PDFF map according to an embodiment of the present invention is shown;

[0046] Figure 5 A flow chart showing the generation of T1 maps using MOLLI or ShMOLLI MRI pulse sequences;

[0047] Figure 6 A flow chart showing a T1 map calculated using a variable flip angle (VFA) pulse sequence;

[0048] Figure 7A flow chart showing the calculation of a T2* map using the multi-echo gradient method;

[0049] Figure 8 A flow chart showing a T2 map calculated using a multi-contrast spin echo pulse sequence;

[0050] Fig. 9 A flow chart showing diffusion weighted EPI (echo planar imaging);

[0051] Fig.10 Various breast density maps according to embodiments of the present invention are shown;

[0052] Fig.11 A hierarchical diagram of the brain is shown. DETAILED DESCRIPTION

[0053] The present invention will now be described with reference to the accompanying drawings, in which examples of methods and apparatus for characterizing breast tissue from medical imaging scan data are illustrated. The method uses quantitative MR imaging, preferably without the use of contrast agents. However, it should be appreciated that the present invention is not limited to the specific examples described herein and shown in the accompanying drawings.

[0054] Furthermore, since the illustrated embodiments of the present invention can for the most part be implemented using components known to those skilled in the art, the details will not be explained in greater detail, except as indicated below, as deemed necessary for understanding and appreciating the basic concepts of the present invention and in order not to confuse and obfuscate the teachings of the present invention.

[0055] By using quantitative MRI methods, different tissue metrics can be measured. Quantitative MRI methods have been used to study tissue and disease states for many years. In the early 1980s, the study of tissue properties such as T1 and T2 was important in helping to develop MRI methods that provided images with tissue contrast that highlighted pathology. The original gold standard quantitative MR methods tended to take a long time and were therefore not suitable for use in the clinical setting because patients could not tolerate lying still for hours on end.

[0056] Improvements in MRI hardware, computer systems, and pulse sequences mean that there are now multiple methods for quantitatively measuring tissue properties. The newer methods are fast, often meaning that images can be collected in a single breath hold (a breath-hold scan). It should also be noted that multiple methods have been developed for measuring different MR tissue metrics. For example, the gold standard method for measuring T1 is known as the inversion recovery spin echo pulse sequence. Collecting the information to calculate T1 using this pulse sequence can take several hours. Some of the newer methods for measuring T1 include MOLLI (modified Look-Locker inversion recovery) sh-MOLLI (shortened MOLLI), and variable flip angle. These methods take less than a minute to collect data, which makes them clinically feasible. It should also be noted that each of the faster methods has an error relative to the gold standard. For example, the MOLLI T1 value will always be lower than the gold standard, and the variable flip angle method will always be higher.

[0057] As with T1, there are a variety of methods to measure other quantitative measures. These methods are pulse sequences with defined scan settings that provide the data needed to calculate the desired measure. Calculating data for T2 typically requires the use of a pulse sequence that provides spin echo data. This may be one of many multi-contrast spin echo type pulse sequences. Conversely, calculating data for T2* typically requires the use of a multi-echo gradient echo pulse sequence. There are several variations of multi-echo gradient echo pulse sequences. A wide variety of these pulse sequences can be used to collect data to calculate T2*. Interestingly, the data needed to calculate the PDFF (proton density fat fraction) map is also typically collected with a multi-echo gradient echo pulse sequence. However, the specific settings for this pulse sequence will change between T2* data and PDFF data. These are separate MRI scans that use the same type of pulse sequence but are run with different scan settings.

[0058] The same considerations apply to apparent diffusion coefficient (ADC) maps, where the ADC parameter measures the magnitude of diffusion of water molecules in tissue. Changes in the ADC correlate with clinical defects in the tissue. The data required for the ADC map requires setting up a pulse sequence to collect diffusion-weighted data, but the specific pulse sequence may be single-shot or multiple-shot diffusion-weighted EPI (echo planar imaging), or even different types of diffusion-weighted scans that can provide the data needed to calculate the ADC. It is important to recognize that the data required to generate the quantitative map can be collected by a variety of different pulse sequences. The pulse sequence and associated scan settings must be appropriate for the desired quantitative map.

[0059] It is also known that newer, faster quantitative MRI scanning methods may be affected by confounding information from other tissue components. For example, by using MOLLI to measure T1 in an environment with a very short T2* characteristic (such as a liver with high iron content), the T1 value obtained will be lower than the value measured by the ground truth method. For short T2*, technology has been developed to correct this deviation in the MOLLIT1 value, see, for example, patent number GB2497668. Similarly, tissues with high fat content will also have an impact on MOLLI T1 measurements. Caution must be exercised when collecting quantitative MRI metrics to ensure that clinically useful methods are implemented and that correction methods are applied when necessary. Using existing rapid quantitative MRI imaging and correction tools for confounding, quantitative imaging can be used to characterize tissues in the clinic.

[0060] In normal practice, each quantitative MR measurement is performed on a pixel-by-pixel (2D scans) or voxel-by-voxel (3D scans) basis. If the data is regularized, the metric can be improved to reduce factors such as noise. Standard image analysis tools can be used to ensure that the quantitative parameter map provides the best estimate of the desired metric.

[0061] The present invention discloses a method for breast MRI imaging, which can be used to determine breast tissue characteristics. MRI imaging is quantitative MRI imaging. In a preferred embodiment of the present invention, the MRI scan does not require the use of a contrast agent. A non-contrast MRI scan will be performed, and the MR data will be used to generate one or more quantitative MR metric maps. These may include PDFF maps, T1 maps, and corrected T1 maps. In another embodiment of the present invention, one or more of T1, T2 maps, T2* maps, and / or ADC maps may also be generated. These MR metric maps are used to calculate one or more of a breast density map, a breast composite map, and a heterogeneity map, which will show the tissue characteristics of fat and non-fat tissue in the MRI image. These quantitative MRI maps can then be used to characterize all or part of the imaged breast tissue. Preferably, a breast tissue heterogeneity characteristic is determined from at least one of the breast composite map and the heterogeneity map, and a data cluster is identified from at least one of the breast composite map, the breast heterogeneity map, and the breast tissue heterogeneity characteristic. In a preferred embodiment of the present invention, the data cluster is a continuous cluster identified across one or more MR metrics.

[0062] The method of the present invention can be used to provide an assessment of breast tissue across the breast, identify abnormal areas in the substance, and measure changes in the breast over time from the various maps and metrics generated. In a preferred embodiment of the present invention, MRI images are acquired over a set period of time to monitor changes in the breast composite map and breast tissue properties. For example, in order to review changes in response to chemotherapy or radiotherapy, the breast can be imaged weekly, but the breast can also be monitored for a longer period of time, wherein the breast is imaged at longer intervals, such as 3 months or longer periods. The assessment can include an estimate of the risk of a woman subsequently developing breast cancer.

[0063] An overview of the method 100 is shown in Figure 2(a). In a preferred embodiment of the present invention, the steps in the method include:

[0064] At step 102, an MRI scan is completed to provide multi-parameter data, preferably full breast coverage of both breasts, but the coverage can be unilateral breast or a region of a single breast. The MRI scan is a quantitative scan, and in a preferred embodiment of the present invention, the scan is performed without the use of a contrast agent. In an embodiment of the present invention, a plurality of quantitative MRI scan images of at least a portion of the breast region are acquired using a different MR pulse sequence for each MRI scan image. The acquired MRI scan is either a 2D scan that produces a set of 2D images composed of pixels, or a 3D scan with a 3D image composed of voxels. A 2D scan will produce a 2D map, and a 3D scan will produce a 3D map.

[0065] In step 102, an MRI scan may be performed in the prone position (the patient lies prone and the breasts are suspended in a breast receiver coil) or in the supine position (the patient lies supine and a flexible coil is placed over the breasts). Further details of both scanning methods are provided below. The methods described below detail the process of obtaining a scan of the entire breast area, but in some embodiments of the invention, only a portion of the breast area may be imaged, or only one of the breasts may be imaged, or only a portion of one of the breasts may be imaged.

[0066] Prone scan

[0067] i. Position the patient in the breast RF coil, ensuring that the breast tissue to be imaged is completely contained within the coil. Comfortable positioning also means ensuring that the forehead is resting against the manufacturer's support device and that the patient's arms are resting in a "superman" position, with the arms extending to either side of the patient's head, or resting at the patient's side.

[0068] ii. All standard safety procedures will be followed and patients will be provided with equipment to notify the control room if a problem arises.

[0069] iii. Advance the patient into the magnet to the scanning position.

[0070] iv. Collecting localizer scans, these scans allow the radiologist to position the patient on the scanning table so that a more accurate data acquisition scan can be performed.

[0071] v. Collect non-quantitative volume scans.

[0072] vi. Collect multiple scans to generate a quantitative map, corresponding to step 102 above.

[0073] Supine scan

[0074] i.Have the patient lie supine on the scanning table with their head resting on pillows. If any breast immobilization device is used, make sure it is worn and comfortable. The patient's arms should be in a comfortable position, usually at their sides.

[0075] ii. Place the flexible RF coil over the breast (upper chest).

[0076] iii. All standard safety procedures will be followed and patients will be provided with equipment to notify the control room if a problem arises.

[0077] iv. Advance the patient into the magnet to the scanning position.

[0078] v. Collect locator scans.

[0079] vi. Collect non-quantitative volume scans.

[0080] vii. Collect multiple scans to generate a quantitative map, corresponding to step 102 above.

[0081] Although the multi-parametric MRI scans may be obtained with the patient in different positions, the analysis of the multi-parametric MRI scan data acquired for the different positions to determine the MRI metric maps, breast density maps, breast composite maps, heterogeneity maps, and breast tissue heterogeneity characteristics will be the same.

[0082] Then, step 104 generates one or more MR parameter maps from one or more MRI metrics from the MR scan data by using an appropriate fitting algorithm. Preferably, the determined metrics are used to generate an MRI map for each determined metric. The MR parameter map may include one or more of T1, corrected T1, T2*, T2, PDFF, and ADC. Step 110 is to analyze at least one MR metric map to generate one or more of a breast density map, a breast composite map, and a heterogeneity map to display tissue characteristics of fat and non-fat tissue in the MRI image. As explained in more detail below, the analysis is used to extract multi-parameter MR values ​​of breast tissue density and parenchyma, which can be used to determine breast tissue heterogeneity characteristics from at least one of the heterogeneity map or the breast composite map, which is shown in more detail in Figure 2(b). In addition, heterogeneity measurements of breast density and MR parameters can be calculated at this stage.

[0083] At step 110, the different MR parameter maps generated in step 104 are analyzed in various ways to generate different maps describing various tissue qualities, such as a breast composite map and a heterogeneity map, which display tissue characteristics of fat and non-fat tissue in the MRI image. An example is a breast density map (such as Fig.10), which shows tissues of different densities in the imaged breast. At step 112, breast density maps are generated from the MR parameter maps, and these maps are then used to generate heterogeneity maps from the breast density maps (step 118). A heterogeneity MR map is a map that shows how a given metric varies within a given region. There can be a variety of ways to measure heterogeneity, such as: the mean of the density of pixels or voxels in an MRI scan, a hierarchical map showing the spatial distribution of an attribute, a texture map of density. It should also be noted that a single heterogeneity measure can be assigned to a spatial map. This single heterogeneity measure can be calculated from the heterogeneity map. As an example, the interquartile range of the values ​​within the map may prove to be a useful measure of heterogeneity. An alternative step at 114 is to generate a heterogeneity map from multiple MR parameter maps. These maps are then used in step 120 to generate a composite breast map based on the heterogeneity map derived from the MR metric map. A composite MR parameter map is generated by a combination of 2 or more maps. An example of a composite map is a corrected T1 map (where T1 is corrected for the iron or fat content of the tissue). Thus, the composite map displays new MR parameter measurements. A large number of composite maps can be used to characterize tissue. In the case of a composite heterogeneity map (such as will be generated in step 120), the new metric can describe a measure of the combined heterogeneity of the PDFF and T1 maps, which will provide a combined view of tissues with high water content (from the T1 map) and high fat content (from the PDFF map). This may help identify patterns of cysts (high water content structures) in fat structures. In an embodiment of the present invention, before generating additional MRI metric maps, at least one MRI metric is combined with one or more other MRI metrics to provide an improved estimate of the initial value of at least one MRI metric. Preferably, at least one metric is a T1 metric and is combined with at least one of PDFF and T2*. Alternatively, from the MR parameter map generated in step 104, parameters can be used to generate a new composite map (step 116), for example, a corrected T1 map, where the T1 values ​​are corrected for fat or iron content in the tissue. These composite maps can then be used to generate a heterogeneity map at 122, or a breast density map of the breast can be generated from the composite map 124. In general, the ability to combine various measures or combined measures of tissue properties is a result of step 110. All of the graphs will help further characterize breast tissue properties.

[0084] The output 126 from the analysis 110 may be one or more of: a breast density map, a composite map, a heterogeneity map, or a measure of heterogeneity or a characteristic of breast tissue heterogeneity. Subsequently, data clusters are identified from at least one of the breast composite map, the breast heterogeneity map, and the characteristic of breast tissue heterogeneity. In an embodiment of the present invention, the breast composite map is a 3D map in which consecutive data clusters represent similar breast tissue. Preferably, the breast composite map provides a spatial distribution of dense breast tissue.

[0085] In step 128, the analysis results from the analysis step 110 (i.e., one or more of the breast density map, composite map, heterogeneity map, or heterogeneity measure) are compared to a pre-existing database. This will identify the location of the tissue characteristics in the parameter space / database. Preferably, the database comparison will allow for the identification of continuous data clusters. The comparison also allows for the identification of abnormal areas of breast tissue. Step 132 is used to detect lesions in the breast tissue based on this comparison, step 134 is used to calculate the risk of breast cancer, step 136 correlates the risk with a current qualitative clinical classification, and step 130 is used to measure the response of the breast tissue to treatment / breast disease progression or recurrence.

[0086] In a preferred embodiment of the present invention, an MRI scan volume (multi-slice 2D or 3D scan, producing 2D and 3D maps, respectively) acquired at step 102 that fully covers both breasts is collected to quantify one or more of the following MRI parameters or metrics according to step 104:

[0087] • T1 metric using MR pulse sequences such as modified Look-Locker Inversion Recovery (MOLLI), sh-MOLLI and / or variable flip angle methods.

[0088] Corrected T1 parameters may also be calculated from the raw T1 measurements.

[0089] • T2* metric using MR pulse sequences such as multi-echo gradient echo.

[0090] • T2 measurement using MR pulse sequences such as multi-contrast spin echo.

[0091] • Proton density fat fraction (PDFF) metric using MR pulse sequences such as multi-echo gradient echo.

[0092] • The apparent diffusion coefficient (ADC) metric, which is a parameter that measures the magnitude of diffusion of water molecules in tissue, using MR pulse sequences such as single-shot or multiple-shot diffusion-weighted echo planar imaging (EPI).

[0093] The MRI metrics determined from the different pulse sequences are then used to generate a corresponding MRI metric map for each determined MRI metric. In a preferred embodiment of the present invention, the MRI metric map is used to calculate at least one of: a breast composite map and a heterogeneity map for displaying tissue characteristics of adipose and non-adipose tissue in the MRI image. Thereafter, a breast tissue heterogeneity characteristic is determined from at least one of the breast composite map and the heterogeneity map; and data clusters are identified from at least one of the breast composite map, the breast heterogeneity map, and the breast tissue heterogeneity characteristic.

[0094] As mentioned in the scan description above, different MR pulse sequences may be used to obtain different MR metrics for different reasons. This may be to optimize tissue contrast while maintaining a clinically relevant scan time. The different MR pulse sequences may be performed in any order, and there may be specific operating protocols for different MRI scanner manufacturers, or for other reasons, for example, but the order of the MR pulse sequences is not important, what is important is that multiple different MR pulse sequences are used. In embodiments of the present invention, one or more MR pulse sequences may be used, and in further embodiments of the present invention, all different pulse sequences may be used to provide MR metrics from each different pulse sequence.

[0095] Here is a brief overview of the different MR pulse sequences mentioned above:

[0096] · Figure 5 Determination of a T1 map from a MOLLI pulse sequence (also called ShMOLLI) is shown in FIG.

[0097] MOLLI (Modified Look-Locker Inversion Recovery) is an inversion recovery pulse sequence 500 that collects data along the T1 recovery curve over 2 or 3 acquisition blocks. MR data is collected at step 501. The collected data is fit to the inversion recovery T1 recovery curve at step 502. Finally at 503, a T1 metric map is generated. ShMOLLI (shortened) modified Look-Locker inversion recovery follows the same general steps, but on a shorter time scale. Both sequences can be used for breath-hold scanning, thereby reducing motion artifacts. They are often used for cardiac T1 parametric scanning.

[0098] · Figure 6 Determination of a T1 map from a VFA pulse sequence 600 is shown in FIG.

[0099] A VFA (variable flip angle) MRI scan is typically a series of short 3D gradient echo scans, where each MRI scan has a different flip angle. At step 601, data is collected by changing only the flip angle of multiple acquisitions. At step 602, the collected data is fitted to the MR signal equation to solve for T1. Finally, at step 603, a T1 metric map is generated. Gradient echo scans are typically short (e.g., a typical scan time is 10-40 seconds), which means that the VFA method provides data to generate a 3D T1 map within a clinically useful time frame.

[0100] · Figure 7 Determination of a T2* map using a multi-echo gradient echo pulse sequence 700 is shown in FIG.

[0101] A multi-echo gradient echo sequence is a gradient echo sequence that is set to acquire different echo times during a single MRI scan - rather than running multiple gradient echo scans, each with a different echo time. Using multi-echo gradient echo provides a time-efficient method to collect scans with different echo times required to generate parameter maps. Gradient echo sequences can be designed so that the information they collect provides data for calculating T2* maps, which take into account the magnetic field around the tissue structure. At step 701, data is collected using a gradient echo pulse sequence, acquiring multiple echoes at each acquisition, followed by step 702, in which the acquired data is fitted to an exponential decay curve. Finally, at step 703, a T2* map is generated. In an embodiment of the present invention, PDFF parameters and metric maps are determined using a multi-echo gradient echo pulse sequence.

[0102] · Figure 8 Determination of a T2 map using a multi-contrast spin echo pulse sequence 800 is shown in FIG.

[0103] A multi-contrast spin echo sequence is a spin echo sequence that is set to acquire different echo times during a single MRI scan. Like multi-echo gradient echo, this provides a time-efficient method for collecting spin echo data with different echo times. Spin echo sequences are designed so that the information they collect provides data for calculating T2 maps. The magnetic field does not affect these scans. Wherein T2 is a characteristic of the tissue, and T2star is a characteristic of the tissue within its physical environment. At step 801, data is collected using a spin echo pulse sequence, and multiple echoes are acquired each time. This is followed by step 802, in which the acquired data is fitted to an exponential decay curve. Finally, at step 803, a T2 map is generated.

[0104] · Fig. 9 Determination of the apparent diffusion coefficient (ADC) using a diffusion-weighted EPI (echo planar imaging) pulse sequence 900 is shown in FIG.

[0105] At step 901, data is collected using a diffusion-weighted EPI pulse sequence, with the amount of diffusion weighting (designated "b") being varied at each acquisition. This is followed by step 902, where the following is calculated:

[0106] ADC=-In(S / S0) / b

[0107] Where S is the signal at a given b value, and S0 is the signal where b=0. Finally, in step 903, an ADC map is generated. Preferably, single-shot or multiple-shot diffusion-weighted echo planar imaging is used to determine the ADC parameters. The EPI pulse sequence is an extremely fast scan that can be acquired in less than a second. Adding diffusion weighting to a pulse sequence is a very time-consuming option, so diffusion weighting on a spin echo scan may mean that the scan time is too long to be used clinically. Therefore, diffusion-weighted scans are often based on EPI pulse sequences. The added diffusion weight provides information related to the way water molecules diffuse in tissues, i.e., the value ADC (apparent diffusion coefficient).

[0108] Some "raw values" of these MR parameters obtained by the above-mentioned one or more MR pulse sequences may be preferably combined with other values ​​to more closely estimate the "true" values ​​of the parameters. In an embodiment of the present invention, at least two determined metrics are combined to generate another metric map, and the other metric map is used to calculate a breast composite map. For example, T calculated using (sh)MOLLI or VFA 1 The raw values ​​of can be compared with (proton density) fat (fraction) and / or T 2 *Estimated value combination. Figure 5 A MOLLI / ShMOLLI flow chart is shown. Data is collected at step 601, and in a preferred embodiment of the invention, the pulse sequence samples the T1 curve by collecting at least three sets of data on the T1 curve. ShMOLLI samples the T1 curve in a similar manner, but on a significantly shorter time scale and with a reduced number of data points, thereby enabling a shortened breath-hold MRI scan. An example scan time for a MOLLI scan is 18 seconds, and an example scan time for a shMOLLI scan is 9 seconds. At step 502, the acquired data is fit to an inversion recovery T1 recovery curve, and at step 503, a T1 map is generated.

[0109] cT1 is a T1 measurement normalized for iron content in the liver and normalized for donor and field strength and is described in GB2497668. Breast corrected T1 will be normalized for donor and field strength and corrected for fat content. In an embodiment of the present invention, we will establish a normalization method by establishing a "ground truth" scanner system and then performing quantitative scans for scanner manufacturers (typically GE, Philips and Siemens) to generate a normalization table to map the phantom data back to the values ​​measured on the "ground truth" scanner.

[0110] In an embodiment of the present invention, using Figure 3The steps shown in 200 calculate the proton density fat fraction (PDFF) from the scan data of the breast. In step 202, a fat spectrum suitable for breast use is used, and in a preferred embodiment of the present invention, a multi-echo gradient echo MRI scan is acquired at step 204, but other pulse sequences may be used for other embodiments of the present invention. The multi-echo gradient echo data and the appropriate fat spectrum of the breast are provided as inputs to step 206, where an algorithm is applied to the input data. Various different algorithms can be used for this, including the IDEAL method (as described in US7176683) and the MAGO method (as described in GB2576886). The PDFF measurement obtained at step 208 provides an estimate of the amount of "fat" in each voxel of the MRI image. The calculation of PDFF generally requires an MR spectrum of the fat contained in the imaged tissue, but other methods of determining PDFF may be used in alternative embodiments of the present invention. The "fat" in the MR spectrum may be a specific type of lipid, or may be represented as a summary of all different lipid types. The MR spectrum adapted for the breast is used in conjunction with an existing IDEAL or MAGO algorithm (described above) to produce a PDFF measurement in each voxel. These algorithms use two substances (such as fat and water) within a voxel to determine the relative signal contribution within that voxel, preferably with the aid of a cost function analysis approach, although other approaches may also be used.

[0111] Figure 4 FIG. 3 shows the calculation of quantitative breast density using a breast PDFF map 300. At step 302, using the above Figure 3The method provides a breast density PDFF map showing the fat content f(x) of each voxel. The density of the breast is based on how much fibroglandular tissue is present in the breast. By describing the breast tissue as either fat or non-fat (which will be composed of fibroglandular tissue and any lesions), we can then think of the density as the amount of non-fat tissue present. Therefore, if a voxel in the MR image represents 40% fat, then the remaining 60% of that voxel will be considered to represent non-fat or dense tissue. Therefore, a breast density map can be calculated from the PDFF map at step 304, where for voxel x, the amount of fat tissue f(x) is expressed as a percentage, and then the density (non-fat amount) of the voxel can be estimated as 1-f(x). To calculate the density map, we will take the PDFF map, which expresses the fat content in each voxel as a percentage, and by calculating 1 minus the fat percentage, step 304 will calculate the density map. The density of the entire breast will be calculated at step 306 as the average of all the density voxels in the breast. In a preferred embodiment of the present invention, breast density will be measured over the entire breast and regions to provide a determination or measurement of breast heterogeneity at step 308. In a preferred embodiment of the invention, determining the breast tissue heterogeneity characteristic comprises the steps of determining one or more of the following: a mean value of the density of pixels or voxels in the MRI scan, a grading map showing the spatial distribution of the characteristic, such as the spatial distribution of dense breast tissue, a texture map of density. In another embodiment of the invention, determining the breast tissue heterogeneity characteristic comprises determining one or more of the following: fat content and breast composition; breast inflammation; and changes in breast tissue architecture.

[0112] However, in some cases, breast density as determined from a PDFF map will be measured for only one breast or only a portion of a breast.

[0113] In an alternative embodiment of the present invention, breast density can also be calculated by segmenting fibrous and fatty tissue according to their corrected T1 values ​​and calculating the relative amount of each. In this case, we will know the total volume of the breast, segment the entire breast tissue, and then from the corrected T1 map, we will be able to distinguish between fatty and non-fatty tissue using the corrected T1 values. In another embodiment of the present invention, these two alternatives, namely, first estimating the fat fraction and then estimating the dense tissue first, will be combined to get the best compromise estimate of both.

[0114] For example, the corrected T1 parameter for breast fat is about 250ms, while the T1 parameter for non-fat tissue will be greater than 500ms, as determined from preliminary experiments and also known from the literature in this field. After segmenting fat from non-fat tissue, we can calculate the ratio of the volume of non-fat tissue to the volume of the entire breast. There are many freely available software packages that can be used for this calculation, such as ITK-Snap. This program uses a level set segmentation algorithm, which is widely used in medical imaging. The two breast density measures can then be cross-validated by comparing the entire breast density calculated using the PDFF map method with the entire breast density calculated using the corrected T1 volume method.

[0115] Fig.10 Various breast density maps are shown according to embodiments of the present invention. Fig.10 (a) shows very dense breast tissue, where the calculated density is 80%, Fig.10 (b) shows a mixed density breast, where the calculated density is 25%, and Fig.10 (c) shows breast tissue classified as non-dense, with a calculated density of 12%. We can correlate these images back to Figure 1 , Figure 1 The location of the axilla in the breast is shown, as well as the location of the lymph nodes within the axilla on an MRI scan. This demonstrates the usefulness of MRI scanning, which covers the entire chest wall and thus provides visualization of lesions associated with breast cancer, but not physical lesions in the breast.

[0116] From the breast density map that has been generated according to the present invention (described above and in Fig.10 In the example shown in Figure 2, additional measures of breast tissue heterogeneity characteristics of breast density will be calculated. These include, but are not limited to: the mean of the density of all pixels / voxels in the breast, a grading map, and a texture map of density. Fig.11An example grading map of the brain is shown in . This is a map that shows the spatial distribution of a property. An example of the invention could be a map that shows all densities above 70% dense tissue in one color, and densities from 0% to 30% dense in another color and above 30% but below 70% density in a third color. Grading maps are typically generated by grouping together pixels / voxels with similar properties to show the spatial distribution. An example of a property to be mapped is all voxels with a density above 70%. Similarly, spatial statistics will be generated from dense breast data, for example, to estimate the "scattered distribution" of dense tissue areas mentioned in some formulas for BI-RADS 2 and 3 categories. Continuous areas of dense tissue can be calculated from images using known methods (including superpixels) and well-known spatial statistics methods for calculating measurements of the spatial distribution of dense tissue within the breast. These measurements make the imprecise discussion of spatial distribution used in the BI-RADS 2 and 3 descriptions precise.

[0117] The present invention also allows for the quantitative breast density determined from the breast density map generated from the MRI metric map described above to be correlated with current qualitative guidelines currently used by clinicians, such as the BI-RADS density score. As we described above, clinicians use the BI-RADS score to classify breast density into one of four categories (BI-RADS1 to BI-RADS 4). These categories are used to influence a woman's clinical pathway, determining which diagnostic procedures she may need to reliably identify lesions. Unfortunately, the classification of women into these four categories has high inter-operator variability, particularly for the two intermediate classifications (BI-RADS-2 and BI-RADS-3). By comparing to our patient database, we correlate the calculated quantitative breast density metric with the current BI-RADS density definition to produce a BI-RADS density score that is not operator dependent.

[0118] Embodiments of the present invention also allow for the use of multi-parameter MRI to quantify breast parenchyma. In addition to corrected T1, PDFF, and breast density, additional parameters in the breast will be quantified, including T2, T2*, and apparent diffusion coefficient (ADC). These additional parameters are calculated by processing the MRI scan data using standard appropriate fitting algorithms, such as Figures 5 to 9 These parameters together create a unique dataset for each patient. We then compare the patient's dataset with the parameter space of our full patient database to identify data clusters. Preferably, the clusters are continuous clusters of data.

[0119] The present invention also allows one or more of the patient's multi-parameter measurements obtained from the MRI data to be compared with the locations of these continuous clusters in parameter space to detect and classify abnormal areas (lesions) in the breast.

[0120] Multiparametric MRI measurements of the patient are compared to the locations of these clusters in parameter space to characterize breast tissue and provide an assessment of breast health.

[0121] Another advantage of the present invention is the use of multi-parametric MRI measurements to measure changes in breast parenchyma to chart breast response to treatment, progression of breast disease, and recurrence of breast disease.

[0122] As described above, a multi-echo gradient echo pulse sequence is used to collect data suitable for an MRI metric map, preferably a PDFF map of the breast.

[0123] Using the data collected for the various pulse sequences as described above, an MRI metric map of the breast, such as a pixel / voxel level PDFF map, is calculated. There are a variety of methods available for this calculation, including the IDEAL method (described in US7176683) and the MAGO method (described in GB2576886). An MR spectrum adapted for the breast is used in conjunction with the existing MAGO algorithm (described in GB 2576886) to produce a PDFF measurement in each voxel.

[0124] The breast density map is calculated from the PDFF map, where for voxel / pixel x, the amount of fat tissue f(x) is known and the density (non-fat mass) can then be estimated as 1-f(x). Figure 4 .

[0125] MRI data acquired using the methods described above can be used to calculate measures of breast tissue density and associated properties.

[0126] The average breast tissue density will be a specific value assigned to each breast or portion of a breast (if only a portion of the breast is imaged). The breast tissue density is calculated as the average of the density of all voxels (for 3D scan images) or pixels (for 2D scan images) within the breast. A texture map for each breast (or portion of a breast) will be generated using currently available standard texture analysis methods (e.g., entropy, standard deviation density functions as can be found in MATLAB).

[0127] From the breast tissue heterogeneity measures and MRI metric maps generated as described above, characteristics of breast tissue density will be calculated, such as breast tissue heterogeneity characteristics including fat content, breast composition, and changes in breast tissue architecture.

[0128] From the collected breast tissue density and breast tissue heterogeneity measures, a score will be generated that correlates to the standard BIRADS breast density classification. In some embodiments of the invention, the score will be a disease risk score.

[0129] In the case of a 3D breast tissue density map (generated as described above), the breast tissue density map provides a measure of breast volume density heterogeneity, rather than just a 3D representation of breast density.

[0130] As described with respect to FIG. 2 , the quantitative MRI data (such as T1, corrected T1, PDFF, T2, T2*, and ADC) required to generate one or more multi-parameter maps of the breast are collected. In addition, localizer scans and non-quantitative volume scans may also be collected for structural investigation and data overlay. The data acquired from the quantitative MRI scans are then processed as follows:

[0131] Using the quantitative MRI data collected in the previous steps, pixel / voxel-by-voxel MR metric maps (such as T1, corrected T1, PDFF, T2, T2*, and ADC) are calculated. In an embodiment of the present invention, the PDFF map is calculated by using an MR spectrum suitable for the breast and the MAGO or IDEAL algorithm. The T1, T2, T2*, and ADC metrics are calculated by processing the scan data using a standard appropriate fitting algorithm. The corrected T1 will be standardized across scanner manufacturers and field strengths, and the fat content in T1 will be taken into account. These metrics can then be used to generate the corresponding MRI metric maps.

[0132] Quantitative MRI metrics will be used to determine breast tissue characteristics, which in embodiments of the invention will include one or more of fat content and composition, inflammation, and breast tissue structural changes.

[0133] In embodiments of the present invention, quantitative MRI metric maps will also be used to help identify lesions and characterize breast tissue by highlighting areas with abnormal MRI metric values.

[0134] By using a combination of one or two or more of these MRI breast metric maps (such as T1, Corrected T1, PDFF, T2, T2*, ADC), measures of breast health as well as breast tissue heterogeneity will be calculated.

[0135] In an embodiment of the invention, breast health and tissue heterogeneity metrics may be based on a dictionary of normal and abnormal tissue parameter values.

[0136] As described above, the present invention provides a method for analyzing MRI data from breast tissue to determine breast tissue heterogeneity. In another embodiment of the present invention, there is also provided an apparatus for analyzing MRI data from breast tissue to determine breast tissue heterogeneity characteristics based on a breast density map, the apparatus comprising at least one processing component, the processing component being arranged to perform the method of the present invention. Preferably, the at least one processing component comprises one or more of the following: one or more programmable components, which are arranged to execute computer program code to perform one or more steps of the method as described above; and a hardware circuit system, which is arranged to perform one or more steps of the method as described above. In another preferred embodiment of the present invention, the apparatus may also include at least one output component for outputting a breast density map or a determined breast tissue heterogeneity characteristic, the at least one output component comprising one or more of the following: a display device for displaying a breast density map or a determined breast tissue heterogeneity characteristic to a user; a data storage device for storing a breast density map or a determined breast tissue heterogeneity characteristic; and an interface component for transmitting a breast density map or a determined breast tissue heterogeneity characteristic to at least one external device.

[0137] As described above, the present invention can be implemented in a computer program for running on an image processing system, comprising at least a code portion for executing the steps of the method according to the present invention when running on a programmable device such as an image processing system or a code portion that enables the programmable device to perform the functions of the device or system according to the present invention.

[0138] A computer program is a list of instructions such as a specific application and / or operating system. A computer program may, for example, include one or more of the following: a subroutine, a function, a procedure, an object method, an object implementation, an executable application, an applet, a server program (servlet), source code, object code, a shared library / dynamically loaded library and / or other instruction sequences designed to be executed on a computer system.

[0139] The computer program may be stored internally on a tangible and non-transitory computer readable storage medium or transmitted to a computer system via a computer readable transmission medium. All or some of the computer program may be provided on a computer readable medium that is permanently, removably or remotely coupled to an information processing system. Tangible and non-transitory computer readable media may include, for example but not limited to, any number of the following media: magnetic storage media including disks and tape storage media; optical storage media such as optical disk media (e.g., CD ROM, CD-R, etc.) and digital video disk storage media; non-volatile memory storage media including semiconductor-based memory cells such as FLASH memory, EEPROM, EPROM, ROM; ferromagnetic digital memory; MRAM; volatile storage media including registers, buffers or caches, main memory, RAM, etc.

[0140] A computer process typically consists of an executing (running) program or part of a program, current program values ​​and state information, and the resources used by the operating system to manage the execution of the process. An operating system (OS) is software that manages the sharing of computer resources and provides programmers with an interface for accessing those resources. An operating system processes system data and user input and responds by allocating and managing tasks and internal system resources as services to the system's users and programs.

[0141] In the foregoing specification, the invention has been described with reference to specific examples of embodiments of the invention. However, it is clear that various modifications and changes may be made therein without departing from the scope of the invention as set forth in the appended claims, and the claims are not limited to the specific examples described above.

[0142] Those skilled in the art will recognize that the boundaries between logic blocks are merely illustrative, and that alternative embodiments may merge logic blocks or circuit elements, or impose alternative functional decompositions on various logic blocks or circuit elements. Therefore, it should be understood that the architectures described herein are merely exemplary, and that in fact many other architectures that achieve the same functionality may be implemented.

[0143] Any arrangement of components that achieve the same functionality is effectively "associated" so that the desired functionality is achieved. Thus, any two components combined herein to achieve a particular functionality may be considered "associated" with each other so that the desired functionality is achieved, regardless of the architecture or intermediate components. Likewise, any two components so associated may also be considered "operably connected" or "operably coupled" to each other so as to achieve the desired functionality.

[0144] In addition, those skilled in the art will appreciate that the boundaries between the above operations are merely illustrative. Multiple operations may be combined into a single operation, a single operation may be distributed in additional operations, and each operation may be performed at least partially overlapping in time. Moreover, alternative embodiments may include multiple instances of a particular operation, and the order of the operations may be changed in various other embodiments.

[0145] However, other modifications, changes, and substitutions are possible. Accordingly, the specification and drawings are to be regarded as illustrative rather than restrictive.

[0146] In the claims, any reference symbol in brackets shall not be interpreted as limiting the claims. The word "comprising" does not exclude the presence of other elements or steps other than those listed in the claims. In addition, the term "a" as used herein is defined as one or more than one. Moreover, the use of introduction phrases such as "at least one" and "one or more" in the claims should not be interpreted as implying that any particular claim containing such introduced claim elements by the indefinite article "a" is limited to an invention containing only one such element, even if the same claim includes the introduction phrase "one or more" or "at least one" and indefinite articles such as "a". The same is true for the use of definite articles. Unless otherwise specified, terms such as "first" and "second" are used to arbitrarily distinguish the elements described by these terms. Therefore, these terms are not necessarily intended to indicate the time or other priority of these elements. The mere fact that certain measurements are described in mutually different claims does not indicate that the combination of these measurements cannot be used to benefit.

Claims

1. A method for analyzing MRI data of breast tissue, comprising: acquiring a plurality of quantitative non-contrast MRI scans of at least a portion of the breast region using a different MR pulse sequence for each MRI scan; determining a plurality of MRI metrics from the acquired quantitative MRI scans; generating a plurality of MRI metric maps using the determined metrics; calculating at least one of a breast composite map and a heterogeneity map from the metric map to display tissue characteristics of fat and non-fat tissue in the MRI image; determining a breast tissue heterogeneity characteristic from at least one of the breast composite map and the heterogeneity map; as well as Data clusters are identified from at least one of a breast composite map, a breast heterogeneity map, and a breast tissue heterogeneity characteristic. 2 . The method of claim 1 , wherein the data clusters are continuous clusters identified across one or more MR metrics.

3. The method of claim 2, wherein the breast composite map is compared with data from a patient database to identify consecutive data clusters.

4. The method of claim 3, wherein the data comparison allows identification of abnormal areas of breast tissue.

5. The method of claim 3 or claim 4, wherein the breast composite map is a 3D map in which consecutive clusters of data represent similar breast tissue.

6. The method of claim 5, wherein the breast composite map provides a spatial distribution of dense breast tissue.

7. The method of any preceding claim, wherein determining the heterogeneity characteristics of breast tissue comprises the following steps: One or more of the following is determined: a mean value of the density of pixels or voxels in the MRI scan, a gradation map showing the spatial distribution of a property, a texture map of the density.

8. The method of any preceding claim, wherein determining characteristics of breast tissue heterogeneity comprises determining one or more of: fat content and breast composition; breast inflammation; and changes in breast tissue architecture.

9. A method as claimed in any preceding claim, wherein the determined metrics generate an MRI map for each determined metric.

10. The method of any of claims 1 to 8, wherein at least two determined metrics are combined to generate a further metric map, and the further metric map is used to calculate a breast composite map.

11. The method of claim 10, wherein raw values ​​of the at least two determined metrics are used to generate the further metric map.

12. The method of any preceding claim, wherein the MRI metrics include one or more of T1, corrected T1, T2, T2*, PDFF, and ADC.

13. The method of any preceding claim, wherein the MRI metric map is a PDFF map or a corrected T1 map.

14. A method as claimed in any preceding claim, wherein the MRI scan is a 2D scan producing a 2D image consisting of a set of pixels, or a 3D scan having a 3D image consisting of voxels.

15. The method of any preceding claim, wherein at least one MRI metric is combined with one or more other MRI metrics before generating an additional MRI metric map, the additional MRI metric map providing an improved estimate of an initial value of the at least one MRI metric.

16. A method as claimed in claim 15 when dependent on claim 8, wherein the at least one metric is a T1 metric and is combined with at least one of PDFF and T2*.

17. The method of any preceding claim, further comprising the step of generating at least one of the following MRI metric maps: T1 map, T2 map, T2* map, ADC map.

18. The method of any preceding claim, wherein the T1 metric is determined using a modified Look Locker inversion recovery sequence acquisition, a shortened modified Look Locker inversion recovery sequence acquisition, or a variable flip angle sequence acquisition.

19. A method as claimed in any preceding claim, wherein the T2* metric and / or the PDFF metric is determined using a multi-echo gradient echo sequence acquisition.

20. A method as claimed in any preceding claim, wherein the T2 metric is determined using a multi-contrast spin echo sequence acquisition.

21. A method as claimed in any preceding claim, wherein the ADC metric is determined using single shot or multiple shot diffusion weighted echo planar imaging.

22. A method as claimed in any preceding claim, wherein the MRI scan is obtained with the patient in a prone or supine position.

23. The method of any preceding claim, wherein the breast composite image and breast tissue characteristics are used to provide a disease risk score.

24. The method of claim 23, wherein the disease risk score is a score associated with the BIRADS breast density classification.

25. A method as claimed in any preceding claim, wherein MRI images are acquired over a set period of time to monitor changes in breast composite images and breast tissue properties.

26. The method of claim 25, wherein the set time period is at least 1 week.

27. A method as claimed in any preceding claim, wherein the MRI scan is a scan of the entire breast area.

28. An apparatus for analysing MRI data from breast tissue to determine breast tissue heterogeneity characteristics from a breast composite map, the apparatus comprising at least one processing component arranged to perform the method of any preceding claim.

29. The apparatus of claim 28, wherein the at least one processing component comprises one or more of: one or more programmable components arranged to execute computer program code to perform one or more steps of the method of any of claims 1 to 28; as well as Hardware circuitry arranged to perform one or more steps of the method of any one of claims 1 to 28.

30. The apparatus of claim 28 or 29, wherein the apparatus further comprises at least one output component for outputting the breast composite map or the determined breast tissue heterogeneity characteristics, the at least one output component comprising one or more of the following: a display device for displaying the composite image of the breast or the determined heterogeneity characteristics of the breast tissue to a user; a data storage device for storing a composite map of the breast or a determined heterogeneity characteristic of breast tissue; as well as An interface component is configured to transmit the breast composite map or the determined breast tissue heterogeneity characteristics to at least one external device.

Images