A method to assign a grade to a tumour tissue specimen using field cycling relaxometry, corresponding system and computer program product

Abstract

A method to assign the grade to a tumour specimen resected during surgery / biopsy using by means of the Field Cycling Relaxometry technique, a corresponding system and computer program product.

Description

[0001]"Relaxometric method for ex vivo characterization of tumour tissue and system thereof" **** FIELD OF THE INVENTION The invention concerns a method for the assessment of the grade of a tumour mass resected during surgery / biopsy using relaxometric measurements. The invention also concerns a system for the assessment of the grade of a tumour mass (i.e., a relaxometer) and a corresponding computer program product for controlling the system. BACKGROUND OF THE INVENTION A tumour is classified on the basis of its stage and its grade. The stage reports about the size and its spread-out in the patient’s body whereas the grade deals with the level of differentiation and aggressiveness of the tumour cells. The lower the differentiation the greater is the difference in respect to the healthy cells and the potential to grow and to quickly disseminate in the body. The grade of a tumour is assigned on the basis of the histologic exam carried out by the anatomopathologist. The grade is expressed on a scale from 1 to 3 or 4, being 1 assigned to the more differentiated cells, i.e., cells that are morphologically more similar to the healthy ones with a low tendency to grow. Grade 3 and 4 refer to a very aggressive behavior with a great tendency to grow and diffuse in the patient’s body. The histological grade and the tumour stage are the parameters of choice to stratify oncological patients and to decide which is the most suitable therapeutic treatment to be undertaken. The assignment of the grade relies on the pathologist’s considerations relative to tumour architecture and cytology, such as tubule formation (a hallmark of the cells growth that bring them to form tubules), mitotic rate (number of cells under active division observed at the microscope under a 400X magnification) and nuclear pleomorphism (assessment of size and shape of the nucleus). Thus, it is evident that the assignment of the grade may be highly subjective as the histologic evaluation is, in principle, “operator-dependent”. In addition, different factors are used to determine the grade of different cancers, and some cancers have their own grading systems. Despite the efforts of international health organisations to periodically revise tumour classifications and develop universally accepted guidelines, there is still a lack of uniformity. There is therefore room for improvement in the definition of the grade of a tumour, in particular in providing an assessment based on a quantitative parameter that may integrate the currently ongoing subjective histological grade evaluation. OBJECT AND SUMMARY OF THE INVENTION An object of the invention is to provide a method for obtaining highly reliable information on a tumour tissue resected during surgery / biopsy in order to contribute in the definition of the proper grade using the Field Cycling Relaxometry. Another object of the invention is to provide a system for carrying out the method which enables such information to be obtained. Another object of the invention is to provide a computer program product to control the system operations. According to the invention, the above object is achieved thanks to the subject matter recalled specifically in the ensuing claims, which are understood as forming an integral part of this disclosure. The present invention concerns a method to assign a grade to a tumour specimen resected during surgery / biopsy by using the Field Cycling Relaxometry technique, the method comprising the following steps: a. obtaining a first sequence of at least m values of a measuring magnetic field strength B , wherein m ≥ 3, preferably m ≥ 5, a second sequence of at least n values of a time τ, a detection magnetic field strength B, a re-cycle delay time RD, a polarization magnetic field strength B and a polarization time t, a number Q corresponding to the maximum tumour grade; b. exposing the specimen to a polarization magnetic field having a polarization magnetic field strength B for a time t, thus allowing the magnetisation build-up, if the current measuring magnetic field strength B is ≤ 7 MHz; c. exposing the specimen to a measuring magnetic field having a measuring magnetic field strength B for a time τ, thus allowing the specimen magnetisation to relax towards a magnetisation equilibrium value determined by the actual B ; d. acquiring a value of magnetisation M of the specimen at time τ and at the measuring magnetic field strength B , the acquisition being carried out while exposing the specimen to a detection magnetic field having a detection magnetic field strength B after a 90° pulse has been applied to M; e. switching off the magnetic field for the re-cycle delay time RD; f. repeating steps b. to e. a number of cycles n, wherein the value of time τ varies at each cycle; g. fitting the values of magnetisation M acquired after n repetitions of steps b. to e. (magnetisation intensity vs. τ) according to a monoexponential Bloch equation to produce a respective magnetisation recovery / decay curve and calculate the proton longitudinal relaxation rate R at the respective magnetic field strength B ; h. repeating steps b. to g. a number of times m, wherein the value of the measuring magnetic field strength B varies each time, thereby calculating at least three proton longitudinal relaxation rates R , R , R at the at least three measuring magnetic field strengths B ; i. calculating: − a Ratio of two relaxation rates R ; − a Sum of at least two relaxation rates R ; − an extracellular water mole fraction p by fitting the at least three magnetisation recovery curves by means of the two-Site eXchange model; − an intracellular residence lifetime τ by fitting the at least three magnetisation recovery curves by the two- Site eXchange model; j. processing the parameters calculated in step g. and i. for assessing the grade of tumour tissue in the specimen by: 1. obtaining a sequence of Q-1 ordered cut-off values C, C, ..., C , calculating a parameter P = Sum * τ, wherein the specimen is grade Q if P has a second positioning with respect to a last cut-off value C , is grade q if P has a first positioning with respect to a cut-off value C and a second positioning with respect to a cut-off value C , and is grade 1 if P has a first positioning with respect to a first cut-off value C; or 2. calculating a score of Q discriminant functions f involving at least three parameters selected among the Ratio, the Sum, the extracellular water mole fraction p, the intracellular residence lifetime τ and the relaxation rate R obtained for a measuring magnetic field strength B , wherein q = 1 to Q, and wherein the specimen is grade q if f has the highest score. The present invention also concerns a system for carrying out the method and a corresponding computer program product for controlling the system operations. BRIEF DESCRIPTION OF THE DRAWINGS The invention will now be described in detail, purely by way of illustrative and non-limiting example, with reference to the attached figures, wherein: FIGURE 1. The Fast Field Cycling (FFC) experiment. A) Not-Polarized (NP) sequence (for high field strengths, i.e., usually applied when B > 0.17 T): the nuclear spin magnetisation is built up during the evolution period (τ) at relaxation field B, then the nuclear magnetic resonance (NMR) signal is detected at detection field B. The sequence is repeated, staggering τ each time. B) Pre-Polarized (PP) sequence (for measurements at low field strengths, i.e., usually at B < 0.17 T): the nuclear spin polarization is built up during the pre-polarization phase, at polarization field B. Relaxation occurs during the evolution period (τ) at relaxation field B, then the NMR signal is detected at detection field B. The sequence is repeated, staggering τ each time. FIGURE 2. An example of the magnetisation decay and recovery curves obtained by applying the pre-polarized (PP) and non-polarized (NP) sequences, respectively. FIGURE 3. A cartoon depicting the water compartmentalization in tissue. See text for symbol definitions. FIGURE 4. An example of application of the two-Site eXchange (2SX) model. A) Set of magnetisation decay data vs. (acquired with the PP sequence in the range 0.01 – 1 MHz of magnetic field strengths) of a breast tumour specimen (G3 grade). The lines correspond to the fits of the data according to the 2SX model as described in the text. B) For the data acquired at 0.02 MHz, the contribution of the apparent a and b relaxing components to the overall relaxation rate constant are shown. FIGURE 5. The longitudinal relaxation rate constant values as a function of the applied magnetic field strengths for Matrigel (open circle) and the R values (filled square) calculated by means of the fitting procedure applied to the data set shown in figure 4. The bars represent the standard deviations. FIGURE 6. Block diagram of an FFC NMR relaxometer. FIGURE 7. Comparison of the proton longitudinal relaxation rate R as a function of the applied magnetic field for three representative breast cancer tissue samples of different grades (G1, G2 and G3, as assigned by the anatomopathologist). FIGURE 8. Schematic representation of two protocols L1 and L2 according to which the method to evaluate the tumor grade of a tumor tissue herein disclosed is applied. FIGURE 9. Schematic representation of an embodiment of the two protocols L1 and L2 according to which the method to evaluate the tumor grade of a breast tumor tissue herein disclosed is applied. FIGURE 10. Flow chart of the method object of the present disclosure. FIGURE 11. Example of the application of the L1 protocol. The P value was calculated for a set of breast cancer tissue samples after obtaining the corresponding relaxometric data. The samples were classified as G2 or G3 according to the cut-off value with three misassignments when compared to the histopathological evaluation. FIGURE 12. Example of the application of the L2 protocol. The x-axis shows the grade of the samples according to the histopathological evaluation. On the y-axis, the samples were classified as G2 or G3 by applying the discriminant rule with two misclassifications, which are marked with an asterisk in the plot. DETAILED DESCRIPTION OF THE INVENTION In the following description, numerous specific details are given to provide a thorough understanding of embodiments. The embodiments can be practiced without one or more of the specific details, or with other methods, components, materials, etc. In other instances, well-known structures, materials, or operations are not shown or described in detail to avoid obscuring aspects of the embodiments. Reference throughout this specification to “one embodiment” or “an embodiment” means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Thus, the appearances of the phrases “in one embodiment” or “in an embodiment” in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. The headings provided herein are for convenience only and do not interpret the scope or meaning of the embodiments. The Field Cycling Relaxometry belongs to the field of the Nuclear Magnetic Resonance Techniques, as the more known Magnetic Resonance Imaging. The Field Cycling Relaxometry allows access to magnetisation decay / recovery curves by acquiring the Free Induction Decay (FID) signal (corresponding to the magnetisation intensity) from the sample after exposure to a given magnetic field (measuring magnetic field B, also indicated as relaxation magnetic field) for different intervals of time τ (see Figs. 1 and 2). By changing the intensity of the measuring magnetic field B, proton longitudinal relaxation rates R can be determined at all frequencies permitted by the instrument and correlated to the chemical composition and dynamics of the proton containing species, mainly water, in the investigated tissue specimen. Figure 1 shows the typical acquisition sequences: Not Polarized (NP) and Pre-Polarized (PP) sequences, for measurements at high (usually > 0.17 T) and low (usually < 0.17 T) field strengths, respectively. For the sake of clarity, 0.17 T corresponds to a proton Larmor frequency (= B, where is the proton gyromagnetic ratio constant and B the magnetic field strength in Tesla) of 7 MHz. The PP sequence differs from the NP sequence because of the application of a high magnetic field (polarization magnetic field B) before the application of the relaxation field B, in order to have a good signal intensity, otherwise not present at low fields. Therefore, in this type of experiment the magnetic field cycles between the polarization field (B) (if applied), several fields (B) at which the proton relaxation time has to be measured, and the detection field (B). The detailed description of the experiment is reported in many papers [i.e., Ferrante G. and Sykora S.: Technical aspects of Fast Field Cycling, Advances in Inorganic Chemistry, 57, 405-470]. Briefly, the basic field cycling experiment with the NP sequence consists of the following steps (see fig. 1 – panel A): 1. The magnetic field is switched to the value of a measuring magnetic field B for a time τ during which the magnetisation relaxes towards an equilibrium value, determined by the current B value. 2. The magnetic field is switched to the value of a fixed detection magnetic field B and the equilibrium magnetisation established at B is measured with a 90° pulse followed by acquisition of the magnetisation / recovery curve at B. 3. The magnetic field is switched off for a certain time RD (Re-cycle Delay) that is the time between repetition of the cycles, equal to 0.5 - 1.5 s. The FID acquired at B (step 2) is proportional to the magnetisation at time τ at the given B. The complete cycle should then be repeated a number n of cycles varying the value of τ (typically the number n of cycles of time τ is equal to 16 or 32, time τ ranging from 0.01 to 5 s in order to recover the entire magnetisation recovery curve). Briefly, the basic field cycling experiment with the PP sequence consists of the following steps (see fig. 1 – panel B): 1. The sample is exposed to a high polarization field B for a polarization time t with consequent building-up of the magnetisation of the sample. t is selected in order to intensity, i.e., with a signal to noise preferably (S / N) 10. 2. The magnetic field is switched to the value of a measuring magnetic field B for a time τ during which the magnetisation relaxes towards an equilibrium value, determined by the value of B. 3. The magnetic field is switched to the value of a fixed detection magnetic field B and the equilibrium magnetisation established at B is measured with a 90° pulse followed by the acquisition of the magnetisation decay curve (FID) at B. 4. The magnetic field is switched off for a certain Re- cycle Delay time RD that is equal to 0.5 - 1.5 s. The FID acquired at B (step 3) is proportional to the magnetisation at time τ at the given B. The complete cycle should then be repeated a number n of cycles varying the value of τ (typically the number n of cycles of time τ is equal to 16 or 32 values, time τ ranging from 0.01 to 5 s in order to recover a large number of points covering the entire magnetisation decay curve). A graphical representation of a magnetisation decay / recovery curve (Intensity vs. τ) is provided in Figure 2. Each dot is obtained applying the steps previously described, which are repeated n = 32 times to obtain a decay / recovery curve. More precisely, the PP sequence was applied at a measuring magnetic field B = 0.01 MHz with 32 values of τ logarithmically distributed in the range 0.0035- 2.8 s; the NP sequence was applied at a measuring magnetic field B = 10 MHz with 32 values of τ logarithmically distributed in the range 0.01-4 s. The fit of a magnetisation decay / recovery curve to an exponential function (Bloch equation) provides the proton longitudinal relaxation rate R (the reciprocal of the proton longitudinal relaxation time T) at the field B. In general, the Bloch equations are a set of macroscopic equations that are used to calculate the nuclear spin magnetisation M = (M, M, M) as a function of time τ when relaxation times T and T are present. These are phenomenological equations that were introduced by Felix Bloch in 1946 as disclosed, i.e., in F. Bloch, "Nuclear Induction", Physical Review 70, 4604–73 (1946). In the case of NP and PP sequences, the magnetisation recovery / decay curve refers to the z component of the magnetisation, M, towards its thermal equilibrium value (M) at field B associated to the constant T (=1 / R), i.e., the longitudinal (or spin-lattice) relaxation time, and, according to the Bloch equation, is defined by the following equations: wherein ^^^^0^^^^and ^^^^0^^^^are parameters estimated from a fit procedure, where they are free to change as a function of the magnetic field strength B applied. ^^^^0^^^^corresponds to the highest measured intensity of the magnetisation M at each selected B . For the magnetisation recovery curve acquired by the NP sequence, ^^^^0^^^^is the difference between zero and the M value at τ = 0; for the magnetisation decay curveacquired by the PP sequence, ^^^^0^^^^is the difference between zero and the M value at τ → ∞ (see Fig. 2). The magnetisation decay / recovery curves may also be subjected to other mathematical treatment, such as applying a biexponential model when two spin populations are present (for example hydrogen protons associated with free water and macromolecules) or the Two-Site eXchange (2SX) model when the relaxation process is modulated by the water exchange across the cell membrane [Landis, C. S. et al. Magn. Reson. Med. 1999, 42, 467–47; Li, X, et al. Magn Reson Med. 2019, 82, 411–424; Hazlewood, C.; Chang, D.; Nichols, B.; Woessner, D. Nuclear Magnetic Resonance Transverse Relaxation Times of Water Protons in Skeletal Muscle. Biophys J. 1974, 14 (8), 583–606. https: / / doi.org / 10.1016 / S0006-3495(74)85937-0]. The 2SX model considers that the major water compartmentalization in tissue is intracellular and extracellular (“inside” / “outside”), being the vascular space small and negligible (see Fig. 3). In the 2SX model the above reported Bloch equations [1, 2] for the analysis of the experimental recovery / decay of magnetisation M(τ) are modified to describe this two- −^^ ⁄ compartments, substituting the term ^^^^ ^^^^^^^^^^^^1 = ^ −^^^^ ^1with the following equation: wherein a and R’ are the fraction and rate constant for the apparent a relaxing component, a and R’ are the fraction and rate constant for the apparent b relaxing component, and τ is the running time for recovery by relaxation, as before. Because a and a are related (a + a = 1), there are only three independent parameters: R’ , R’ , and a (or a), expressed as: where R is the intrinsic longitudinal relaxation rate constant of the extracellular water protons at the magnetisation field strength B , R is the intrinsic longitudinal relaxation rate constant of the intracellular water protons at the magnetisation field strength B , τ is the extracellular water lifetime and p is the tissue extracellular water mole fraction. p and τ values are related to the intracellular water lifetime τ and the tissue intracellular water mole fraction p by the equilibrium mass balance: = (p / p ) [7] This model can be applied when the condition │+ │ ~ │R - R │ [8] is met, as happens at low magnetic field strengths (comprised between 10 MHz and 0.01 MHz), where the extracellular compartment is characterized by lower R values with respect to the intracellular ones. In the fitting procedure (see Table 1, figure 4 and 5), the R are fixed to the relaxation rate measured for Matrigel (see figure 5), which mimics the extracellular compartment [Hughes, C.; Postovit, L.; Lajoie, G. Matrigel: A Complex Protein Mixture Required for Optimal Growth of Cell Culture. Proteomics 2010, 10 (9), 1886–1890. https: / / doi.org / 10.1002 / pmic.200900758; Ruggiero, M. R.; Baroni, S.; Pezzana, S.; Ferrante, G.; Geninatti Crich, S.; Aime, S. Evidence for the Role of Intracellular Water Lifetime as a Tumour Biomarker Obtained by In Vivo Field-Cycling Relaxometry. Angewandte Chemie - International Edition 2018, 57 (25), 7468–7472. https: / / doi.org / 10.1002 / anie.201713318]. Matrigel is the trade name for a gelatinous protein mixture secreted by Engelbreth-Holm-Swarm mouse sarcoma cells. It is considered a good model of the extracellular environment present in many tissues and used by cell biologists as a substrate (basement membrane matrix) for growing cells. The R , τ and p parameters were obtained from the 2SX model fitting, applying a simultaneous fitting of at least three magnetisation decay / recovery curves obtained over an extended range of magnetic fields comprised between 10 MHz and 0.01 MHz. The p is allowed to vary within a feasible range, in accordance with results already reported in the literature (0.09-0.5) [C. S. Landis, X. Lin, F. W. Telaqng, P. E. Molina, I. Palyka, G. Vetek, C. S. Springer Jr, Magn. Reson. Med. 1999, 42, 467–478; X. Li, W. D. Rooney, C. S. Springer Jr, PNAS 2008, 105, 17937-17942; S. L. Barnes, A. G. Sorace, M. E. Loveless, J. G. Whisenant, T. E. Yankeelov, NMR Biomed. 2015, 28, 1345–1356; E. Panagiotaki, S. Walker- Samuel, B. Siow, S. P. Johnson, V. Rajkumar, R. B. Pedley, M. F. Lythgoe, D. C. Alexander, Cancer Res. 2014, 74, 1902-1912]. The fitting procedure imposes that the values of τ and p parameters (which are independent of the magnetic field and intrinsic characteristics of the investigated specimen) are shared, i.e. they are the same for the three curves. Table 1 Independent variables of the 2SX model As example, the fitting results for the data acquired on a breast tumour sample (figure 4A) are reported in the following table, where the standard deviation of values is reported in bracket: Table 2 Variable values determined by applying the 2SX model to the data acquired on a breast tumour sample (in figure 4A) *From curve behavior; **From Matrigel; ***constrained value range from literature: 0.15-0.5 Discriminant Analysis (DA) may be applied for tumour grade assessment. DA developed historically along with multivariate statistics and can be traced back to R. A. Fisher (1890-1962) in the 1930s [Fisher, R. A. 1936. The use of multiple measurements in taxonomic problems. Annals of Eugenics 7: 179–188]. For a theoretical discussion of the subject, please refer to McLachlan, G. J. (1992) Discriminant analysis and statistical pattern recognition, New York, Wiley, doi: 10.1002 / 0471725293; CJ Huberty, S Olejnik (2005) Applied MANOVA and Discriminant Analysis. Wiley Series in Probability and Statistics, John Wiley & Sons, Inc., ISBN:9780471468158; WK Härdle, L Simar (2015) Applied Multivariate Statistical Analysis. Springer Berlin, Heidelberg, eBook ISBN 978-3-662- 45171-7; MJ Adams (2005) Chemometrics and Statistics. Multivariate Classification Techniques. Editor(s): Paul Worsfold, Alan Townshend, Colin Poole, Encyclopedia of Analytical Science ISBN 9780123693976. DA is well-suited for classifying many aspects of medical diagnostics and is widely accessible and easy to perform using statistical packages such as OriginPro (OriginLab Corporation, Northampton, MA, USA); Stata Statistical Software (StataCorp LLC, College Station, TX); SAS software (SAS Institute Inc., Cary, NC, USA), IBM SPSS Statistics (IBM Corp., Armonk, N.Y., USA), MATLAB (The MathWorks Inc., Natick, Massachusetts, USA) and the free R Statistical Software (R Core Team, R Foundation for Statistical Computing, Vienna, Austria). DA allocates specimens to the grade groups using information from specimens whose group memberships are known (i.e., training data). More precisely, the objectives of a DA are: i) Discrimination: to construct a classification model based on the parameters estimates of the training data. ii) Classification: to distribute unlabelled samples into labelled groups with the classification model. DA can be described as a four-step process: 1) Preliminary phase / operations. Identify from all available input features the useful feature variables for the formulation of the discriminant function(s), i.e., identify the training data set. With regard to the training data in this context, the variables / features considered belong to the MR measurements of the decay / recovery curve of magnetisation, such as the derived R values (and their combinations such as ratio of two R values or sum of two or more R values acquired at different B, for example) and the extrapolated physical- chemical parameters (water intra- / extra-cellular residence times and mole fractions) acquired on representative specimens of the groups under study (i.e. breast tumour tissues of grade G1-3, as verified histologically). The data are analyzed to identify at which magnetic field strength the greatest differences in R are observed and which variables best discriminate between the groups. Thus, among all the possible ones, the variables to be considered for DA are those with significantly different mean values for the groups, with a small variance within the group and with a low correlation. The correspondence between the training data set and the assumptions (multivariate normality, equality or not of variance-covariance within group and low multicolinearity of the variables) of the discriminant model are verified with the suitable statistic test. The training data refer to tumour samples on Q possible groups with q = 1, 2, … Q where the maximum Q is usually equal to 4; the most common classification system for breast cancer involves just 3 possible grades, Q = 3. 2) Determination of the discriminant model. Discriminant analysis is designed to weigh and combine the variables of the training data set (discriminator variables) into discriminant functions, in such a manner as to force groups to be as mathematically distinct as possible. Several statistical criteria can be used. In practice, linear and quadratic discriminant models are frequently used, assuming an underlying multivariate normal data distribution. In this regard, please refer to A. Tharwat, “Linear vs. quadratic discriminant analysis classifier: a tutorial”, Int. J. Applied Pattern Recognition, Vol. 3, No. 2, 2016. Since the case of tumour grading is a problem of classification between Q possible tumour grades q with 1 ≤ q ≤ Q (usually Q = 4, indicated herein briefly as G1, G2, G3, G4), Q discriminant functions f are considered. Hence, the discriminant functions are used to determine the group label of the unknown sample based on comparing the d discriminant function scores and assigns the group label of the maximum score to the unknown sample (x) as shown in equation [9]: f (x)> f (x) with q1, q2 = 1, 2, …, Q, and q1 ≠ q2 [9] If the values of any two discriminant functions are equal (f (x) = f (x)), thus the unknown sample is on the boundary between the two classes. Discriminant functions are used to determine the decision boundary (in the form of a linear or quadratic function) between the groups and the region of each group in the data space. Each decision boundary (S ) separates two different regions, i.e., two groups q1 and q2, and it consists of two discriminant functions, f and f . Considering the case of two groups (i.e., Q = 2, group 1 for low grade G1 samples and group 2 for high grade G2 samples), thus there are two different discriminant functions (f and f) and the decision boundary (S (x) = f(x)-f(x)) is calculated as follows: where sgn indicates the sign of the function. It again means that each sample is assigned to the group that gives the highest value of the discriminant function. 3) Evaluation of the discriminant model. The efficacy of the discriminant model can be measured by the proportion of correct assignments. In practice, the model is applied on the training data set: the group labels of the training data set are predicted using the discriminant model and compared to the actual (determined by HE analysis) group labels. 4) Classification of an unknown sample. When the discriminant rule has been learned from the training data, it can then be used to predict the class label of new data points. In practice, in accordance with the previous schemes [9, 10], the values of the variable vector of the unknown sample are substituted in the N discriminant functions f(x) and the sample is assigned to the group at which corresponds the maximum value of the discriminant function. The current commercial Fast Field Cycling relaxometers operate with a solenoid magnet which may achieve magnetic fields from few kHz to a maximum of 42 MHz in terms of proton Larmor frequency (i.e., to about 1 T). The block diagram of a typical system (i.e., an NMR relaxometer) is shown in Fig. 6. The NMR system includes a resistive solenoid (indicated as “Magnet” in Fig. 6) connected to a suitable magnet power supply unit (controlled via a dedicated magnet interface) and a cooling system, specifically designed for Field Cycling purposes. The NMR system further includes an electronic control unit that in turn includes an interface to the magnet power supply unit, a Variable Temperature Controller (VTC), a RadioFrequency (RF) unit (transmitter and receiver, including a preamplifier) and an acquisition unit. The electronic modulation of the current flowing through the coil of the electromagnet permits fast variations of the magnetic field induction and the acquisition of very short T, at present, down to fractions of a millisecond. The probe, the RadioFrequency (RF) unit (transmitter and receiver, including a preamplifier), the sample Variable Temperature Controller (VTC) and the Acquisition Unit are conventional systems for NMR instruments, and a corresponding detailed description will not be provided herein for the sake of brevity. In particular, the acquisition unit detects the NMR signal from the sample and digitizes that signal for processing by a host computer system (or, more generally, a processing unit), e.g., a workstation or a computer system embedded in the relaxometer apparatus. A computer program (e.g., software package) running on the host computer system supports all experimental procedures (e.g., data acquisition, visualization and evaluation). The present invention relies on the observation of the present inventors that a tailored comparative procedure between magnetisation recovery curves acquired at given B values allows us to assess the grade of a tumour. The relaxation rate R of biological tissues is affected by the field strength at which the measurement is carried out. In particular, R shows a dramatic increase on going from high to low fields. At the low fields the intracellular R is high (very short T), whereas R of the extracellular compartment maintains a relatively low value. As water protons are the main source of the detected signal, the observed relaxation rates R are weighted to values that reflect either the ratio between the volumes of the two compartments and the exchange rate between them. They contain relevant information on the cell types present in the investigated tissue specimen and on the physio-pathological transformations occurring inside the cells and at the level of membrane water permeability. The exchange of water molecules across the tumour cellular membrane reflects either an enhanced metabolism (the higher production of metabolites implies an increase of the osmotic pressure) and the over- expression / up / down-regulation of transporters at the cellular membrane. In Fig. 7 the R values obtained from tumour breast tissues of different grade, in the field range corresponding to Larmor Frequencies (= B, where is the proton gyromagnetic constant) of 0.01 to 1 MHz, are reported. It is possible to observe a decrease in R values with increasing tumour grade, as the phenomena related to relaxation mechanisms are affected by the metabolic and structural changes associated with cellular differentiation. In one embodiment, the present invention concerns a method to assign a grade to a tumour specimen resected during surgery / biopsy using the Field Cycling Relaxometry technique (the flow diagram of the method being shown in Figure 10), the method comprising the following steps: a. obtaining a first sequence of at least m values of a measuring magnetic field strength B , wherein m ≥ 3, preferably m ≥ 5, a second sequence of at least n values of a time τ, a detection magnetic field strength B, a re-cycle delay time RD, a polarization magnetic field strength B and a polarization time t, a number Q corresponding to the maximum tumour grade; b. exposing the specimen to a polarization magnetic field characterized by a magnetic field strength B for a time t, thus allowing the magnetisation build-up, if the current measuring magnetic field strength B is 7 MHz; c. exposing the specimen to a measuring magnetic field having a magnetic field strength B , for a time τ, thus allowing the magnetisation of the specimen to relax towards a magnetisation equilibrium value determined by the actual B ; d. acquiring a value of magnetisation M of the specimen at time τ and at the measuring magnetic field strength B , the acquisition being carried out while exposing the specimen to a detection magnetic field having a detection magnetic field strength B after the application of a 90° pulse; e. switching off the magnetic field for the re-cycle delay time RD; f. repeating steps b. to e. a number of cycles n, wherein the value of time τ varies at each cycle; g. fitting the values of magnetisation M acquired after n repetitions of steps b. to e. (magnetisation intensity vs. τ) according to a monoexponential Bloch equation to produce the corresponding magnetisation recovery / decay curve and calculate the proton longitudinal relaxation rate R at the respective magnetic field strength B ; h. repeating steps b. to g. a number of times m, wherein the value of the measuring magnetic field strength B varies each time, thereby calculating at least three proton longitudinal relaxation rates R , R , R at the at least three measuring magnetic field strengths B ; i. calculating: − a Ratio of two relaxation rates R ; − a Sum of at least two relaxation rates R ; − an extracellular water mole fraction p by fitting the at least three magnetisation recovery curves by the two- Site eXchange model; − an intracellular residence lifetime τ by fitting the at least three magnetisation recovery curves by the two- Site eXchange model; j. processing the parameters calculated in step g. and i. for assessing the presence of tumour tissue in the specimen by: 1. obtaining a sequence of Q-1 ordered cut-off values C, C, ..., C , calculating a parameter P = Sum * τ, wherein the specimen is grade Q if P has a second positioning with respect to a last cut-off value C , is grade q if P has a first positioning with respect to a cut-off value C and a second positioning with respect to a cut-off value C , and is grade 1 if P has a first positioning with respect to a first cut-off value C; or 2. calculating a score of Q discriminant functions f involving at least three parameters selected among the Ratio, the Sum, the extracellular water mole fraction p, the intracellular residence lifetime τ and the relaxation rate R obtained for a measuring magnetic field strength B , wherein q = 1 to Q, and wherein the specimen is grade q if f has the highest score. The method for determining the tumour grade herein disclosed can be carried out either during or after the surgery. In one embodiment, the present invention concerns an NMR relaxometer system that includes: - a solenoid connected to a cooling system and a power supply unit, where the solenoid defines a space for hosting the probe in which the sample is settled; - a radio-frequency unit; - a magnet interface controlling the magnet power supply and thus the magnetic field generated by the solenoid; - a sample temperature control unit; - an acquisition module configured to acquire magnetization data from the biological specimen; - a computer host equipped with a computer program (e.g., software package) for controlling the acquisition unit and controlling various steps of the experimental procedure (e.g., data acquisition, visualization and evaluation). In one embodiment, the present invention concerns a computer program product comprising instructions to operate an NMR relaxometer system according to the invention to execute the steps of the method according to the invention. The assessment of the optimal conditions of the Fast Field Cycling relaxometer will be acquired with an auto-test carried out before the start of the measurement session using a properly designed phantom endowed with relaxation properties in the range of those expected for the investigated specimen. The phantom is made by a cross-linked protein in a buffered solution with a preservative agent and it is sealed under inert gas atmosphere. For example, the phantom consists of a solution of Bovine Serum Albumin (BSA) 9% in phosphate buffer pH 7.4 plus sodium azide 0.3 %, cross-linked by heat (20’ at 80°C) or by glutaraldehyde. The skilled man, with his common general knowledge, knows how to produce a phantom endowed with the relaxation properties in the range of those expected for the investigated biological tissue under analysis. The check of relaxometer performance consists of the acquisition of the R values at two magnetisation field strengths twice with the defined protocol. Both the signal intensity and the obtained R values are evaluated. The limits of acceptability for signal intensity, average and standard deviation values of the repeated R measurements have to be within the 5% of the values certified by the phantom manufacturer. The main parameters to be set for carrying out the method of the present invention are: - the number of times m and the values of the measuring magnetic field strengths B used for the acquisition of the magnetisation recovery / decay curves; - the value of the polarization magnetic field strength B and the polarization time t; - the number of cycles n and the values of the time intervals τ for the acquisition of the magnetization relaxation curves at each B value; - the waiting time before the starting of the new acquisition cycle (recycle delay, RD). In case the magnetisation M measured by the relaxometer is lower than a threshold value fixed by the instrument manufacturer, the method steps b. to h. are repeated at least one more time and the magnetizations M determined in the first scan are summed to the magnetizations M determined in the second scan. For determining the grade of the specimen, the following parameters are calculated in steps g. and i.: - the relaxation rate R at a given B , obtained by the fit of the magnetisation recovery curves acquired in step c. (Intensity vs. τ) according to the monoexponential Bloch equation; - the Ratio that is the ratio of two relaxation rates R / R measured at two different measuring magnetic field strengths B , wherein B is in a range of 0.01 - 0.15 MHz, and B is in a range of 0.15 - 10 MHz; - the Sum that is the sum of two or more (n=3-7) relaxation rates (2R and nR, respectively) measured at different B ; - the intracellular residence lifetime τ obtained by the simultaneous fit of the at least three magnetization recovery curves (Intensity vs. τ) according to the 2SX model; - the tissue extracellular water mole fraction p obtained by the simultaneous fit of at least three magnetization recovery curves (Intensity vs. τ) according to the 2SX model. The above parameters are then subjected to a further elaboration: - calculation of a parameter P = Sum * τ, and - calculating a score of Q discriminant functions f involving at least three parameters selected among the Ratio, the Sum, the extracellular water mole fraction p, the intracellular residence lifetime τ and the relaxation rate R obtained for a measuring magnetic field strength B , wherein q = 1 to Q, and wherein the specimen is grade q if f has the highest score. The specimen is classified as follows: - tumour of grade Q if P has a second positioning with respect to a last cut-off value C , grade q if P has a first positioning with respect to a cut-off value C and a second positioning with respect to a cut-off value C , and grade 1 if P has a first positioning with respect to a first cut-off value C; or - tumour of grade q if fq has the maximum value among the Q discriminant functions. The cut-off(s) C, C, ..., C is (are) determined in a training phase during which several resected tumour tissues are subjected to the method disclosed above and then to the histological analysis provided by a surgical pathologist. The histological analysis (comprising tissue formalin-fixation, processing, sectioning, and staining with Hematoxylin & Eosin) is the reference method for tumour assessment and allows the correct distribution of the analyzed specimen in the grade classes. The number s of cut-off (C) depends on the Q groups being s = Q-1 (i.e. in the case of Q = 3, the cut-off C separates the G1 and G2 groups and the cutoff C separates the G2 and G3 groups). In this context, a True positive (TP) is an outcome where the herein disclosed relaxometric method correctly predicts the grade class. Similarly, a true negative is an outcome where the method incorrectly predicts the grade class. The cut-off values to assign the sample to the correct class are evaluated according to the Multiclass Receiver Operating Characteristic (ROC) curve. The ROC curve is constructed per each of the n classes. In each step, a given class is regarded as the positive class and the remaining classes are regarded as the negative class as a bulk. The ROC curve is constructed by plotting the true positive rate (TPR) against the false positive rate (FPR) for the different possible cut-off values. The true positive rate is the proportion of samples that are correctly predicted to be positive out of all positive samples (TP / (TP + FN)), where FN are the False Negative samples. Similarly, the false positive rate is the proportion of samples that are incorrectly predicted to be positive out of all negative samples (FP / (TN + FP), where FP are the False Positive samples. The TPR and FPR correspond to the sensitivity and the (1 - specificity) of the method, respectively. Each point on the ROC curve represents a sensitivity / specificity pair. The cut-off value is the point which classifies most of the samples correctly, i.e., it has the best sensitivity / specificity pair. The scores of the discriminant functions f are calculated involving at least three parameters selected from the Ratio, the Sum, the extracellular water mole fraction p, the intracellular residence lifetime τ and the relaxation rate R . The discriminant functions f are determined by performing a discriminant analysis of a training data set, involving at least three parameters of the parameters set forth above. In one embodiment, the acquisition is carried out at a detection magnetic field strength B comprised between 5 MHz and 20 MHz expressed in terms of proton Larmor frequency. In one embodiment, the measuring magnetic field strength B is comprised in the range 0.01 to 10 MHz in terms of proton Larmor frequency. In one embodiment, the relaxation rates R (=1 / T ) are determined by applying the following Bloch equations: ^^^^0^^^^is the highest measured magnetisation M at B ; ^^^^^^^^0 is the magnetisation M at τ = 0 if B > 7 MHz(i.e., the NP sequence) or the magnetisation M at τ → ∞ if 7 MHz (i.e., the PP sequence); and equation [1] is applied for B > 7 MHz (i.e., the NP sequence), equation [2] is applied for B 7 MHz (i.e., the PP sequence). In one embodiment, at least two values of B are comprised between 0.01 and 0.15 MHz, and at least one value of B is comprised between 0.15 and 10 MHz. In one embodiment, at least three values of B are comprised between 0.01 and 0.15 MHz, and at least two values of B are comprised between 0.15 and 10 MHz. In one embodiment, at least four values of B are comprised between 0.01 and 0.15 MHz, and at least three values of B are comprised between 0.15 and 10 MHz. In one embodiment, the Ratio is calculated by dividing two relaxation rates R at two different measuring magnetic field strengths B , wherein the relaxation rate R in the numerator is the relaxation rate R determined to a value of the measuring magnetic field strength B comprised between 0.01 and 0.15 MHz, and the relaxation rate R in the denominator is the relaxation rate R determined to a value of the measuring magnetic field strength B comprised between 0.15 and 10 MHz. In one embodiment, the Sum is calculated by summing two relaxation rates R at two different measuring magnetic field strengths B , wherein the two relaxation rates R are determined at two values of the measuring magnetic field strength B comprised between 0.01 and 10 MHz. In one embodiment, the intracellular residence lifetime τ and the tissue extracellular water mole fraction p are calculated by means of the following equation(s): and wherein R is the longitudinal relaxation rate constant of the extracellular water protons of Matrigel at the measuring magnetic field strength B ; R is the longitudinal relaxation rate constant of the intracellular water protons of the specimen at the measuring magnetic field strength B ; ^^^^0^^^^and ^^^^0^^^^have the meaning set forth above; τ = τ (p / p) [7], and equation [3a] is resolved for B > 7 MHz (i.e., the NP sequence), equation [3b] is resolved for B 7 MHz (i.e., the PP sequence). In one embodiment, the specimen is a resected lymph node(s) having a weight comprised between 15 mg and 115 mg, preferably between 30 mg and 60 mg, as obtained by the surgeon in the surgery room. In one embodiment, the specimen is placed in a glass tube (having generally an outer diameter of 5 mm but other sizes are possible), the glass tube being introduced in the Field Cycling Relaxometer. In one embodiment, the steps d. to f. are carried out at a temperature 25 °C, preferably at about 10 °C. In one embodiment, when the steps d. to f. are carried out at a temperature lower than the room temperature (i.e., lower than 25 °C), the test tube containing the tissue specimen is brought to the selected temperature through the passage in a sample cooling device, wherein the sample cooling device is designed to host the test tube to allow the specimen to reach the desired temperature. In one embodiment, the polarization magnetic field strength B of the polarization magnetic field is comprised between 8 and 25 MHz. In one embodiment, the polarization step b. is carried out for a polarization time t, wherein t is comprised in the range 0.5 to 1.5 s. In one embodiment, the switching off of the magnetic field in step e. allows the specimen magnetization to completely decay. In one embodiment, the switching off of the magnetic field in step e. lasts for at least 0.5 sec. In one embodiment, the number of cycles n ranges from 10 to 32. In one embodiment, the time value τ ranges from 0.01 to 4 sec. In one embodiment, in each cycle n, in which steps b. to e. are repeated, the time value τ is increased, preferably on a logarithmic basis. In one embodiment, the process steps b. to h. are carried out once employing a Matrigel specimen for determining the relaxation rate constants R at the magnetic field strengths B applied according to the monoexponential Bloch equation. The obtained R are used as the longitudinal relaxation rate constants of the extracellular water protons R in the application of the two-Site eXchange model. In the following some embodiments of the present invention are disclosed. Assessment of the tumour grade In order to assess the tumour grade, two protocols L1 and L2 have been developed as shown in Figure 8. The two protocols imply to apply the following steps as reported in the Figure 8: − Acquisition step: acquisition of the magnetization recovery curves at a temperature ≤ 25°C at: 2-4 values of Larmor frequencies in the range 0.01-0.15 MHz using the protocol AP3 or AP4; 1-3 values of Larmor frequencies in the range 0.15-10 MHz using the protocol AP1 or AP2. Protocols AP1 to AP4 are disclosed below. In the case of protocol L1 (first panel of Figure 8): − Data treatment: the fitting of the magnetisation recovery / decay curves obtained in the acquisition step (Intensity vs. τ) is performed according to the monoexponential Bloch equation [1-2] to calculate the R values at the given B. The obtained R data are analyzed in terms of the above reported Sum. The same data are also analyzed in terms of the 2SX model resolving equation [3a-b] to extract the intracellular residence time . In the case of protocol L2 (second panel of Figure 8): - Data treatment: the fitting of the magnetization recovery curves obtained in the acquisition step (Intensity vs. τ) is performed according to the monoexponential Bloch equation [1-2] to calculate the R values at the given B . The obtained R data are analyzed in terms of the above reported Ratio and Sum methodology. The same data are also analyzed in terms of the 2SX model resolving equation [3a-b] to extract the intracellular residence lifetime τ and the extracellular water mole fraction p. At least three of the previous parameter values (Ratio, Sum, τ, p and the R value at Bri) are used to calculate the value of the N discriminant functions f and to apply the sample classification rule, i.e., equations [9,10]. Examples of settings for the R acquisitions at B between 0.15 and 10 MHz are: − [AP1]: B = 9.5 MHz, t = RD = 0.9 s, n = 14 τ distributed on a logarithmic basis in the interval 0.01-1.8 s; − [AP2]: B = 9.5 MHz, t = RD = 0.9 s, n = 32 τ distributed on a logarithmic basis in the interval 0.01-4.0 s. Examples of settings for the R acquisitions at B between 0.01 and 0.15 MHz are: − [AP3]: B = 9.5 MHz, t = RD = 0.9 s, n =14 τ distributed on a logarithmic basis in the interval 0.0035-1.0 s; [AP4]: B = 9.5 MHz, t = RD = 0.9 s, n = 32 τ distributed on a logarithmic basis in the interval 0.035-2.8 s. Application of the L protocols to assess the grade of the examined breast tumours 29 samples of breast tumour tissue were examined. The results from the anatomopathological laboratory yielded 14 G2 and 15 G3 samples. For data acquisition, the protocols reported in Figure 9, a specification of Figure 8, were followed. In particular, the acquisitions for both protocols were carried out at 10 °C and at the following Larmor frequencies: 0.01, 0.02, 0.037 and 0.07 MHz using the acquisition protocol AP2; 0.15, 0.39 and 1 MHz using the protocol AP1 (overall required time: ca. 20’). The protocol was made explicit for the classification of breast cancer samples, so P of G3 < P of G2 and Q = 3. Both the protocols consider the Sum parameters, calculated as follows: Sum = 2R = R + R in protocol L1; Sum = 2R = R + R in protocol L2. The M decay curves acquired at the previous 7 listed magnetic field strengths were subjected to the 2SX model treatment, obtaining the and p values. In the protocol L1, the P parameter is calculated as follows: P = Sum * = 2R * = (R + R ) * In this case, it was found that when the parameter P is less than or equal to 1.4 (the cut-off C), the sample is classified in G3 group. When P is greater than 1.4 the specimen is classified in the G2 group (see Figure 11). By comparing the classification obtained with protocol L1 and the anatomopathologist analysis, three misassignments were found, thus yielding a sensitivity of 93% and specificity of 86% and accuracy (in terms of AUC, i.e., area under the curve) of 92%. The protocol L2 applies the discriminant rule learned from the training data set, developed by using the OriginPro software (Version 2023, OriginLab Corporation, Northampton, MA, USA) according to the QDA model and equal prior probabilities. The variables involved in the determined discriminant functions are R , Sum (2R = R + R ), p and , f (R , 2R1, p ) and f (R , 2R1, p ) where 1 indicates the G2 group and 2 the G3 group. The values of R , 2R, p of each sample are substituted in the discriminant function f(x) and f(x) and the sample is assigned to the class which correspond the maximum score according to the classification scheme [9, 10]. By comparing the classification obtained with protocol L2 and the anatomopathologist analysis (Figure 12), two misassignments were found, thus yielding a sensitivity of 93%, specificity of 93% and accuracy (in terms of AUC, i.e., area under the curve) of 93%. The histological evaluation The histological analysis is the gold-standard reference and the results from the relaxometric determinations have been referred to the histological evaluation for all the investigated specimens. The typical procedure is as follows: Hematoxylin & Eosin (H&E) staining was conducted using standard methods on formalin-fixed, paraffin-embedded tissues. Samples were fixed in 10% neutral buffered formalin and embedded in paraffin. For histopathological analysis, samples were serially sectioned (5 μm) and every section stained with H&E. Histopathological scoring of tumour was performed by an expert anatomopathologist.

Claims

CLAIMS 1. A method to assign a grade to a tumour specimen resected during surgery / biopsy by using the Field Cycling Relaxometry technique, the method comprising the following steps: a. obtaining a first sequence of at least m values of a measuring magnetic field strength B , wherein m ≥ 3, preferably m ≥ 5, a second sequence of at least n values of a time τ, a detection magnetic field strength B, a re-cycle delay time RD, a polarization magnetic field strength B and a polarization time t, a number Q corresponding to the maximum tumour grade; b. exposing the specimen to a polarization magnetic field having a polarization magnetic field strength B for a time t, thus allowing the magnetisation build-up, if the current measuring magnetic field strength B is ≤ 7 MHz; c. exposing the specimen to a measuring magnetic field having a measuring magnetic field strength B for a time τ, thus allowing the specimen magnetisation to relax towards a magnetisation equilibrium value determined by the actual B ; d. acquiring a value of magnetisation M of the specimen at time τ and at the measuring magnetic field strength B , the acquisition being carried out while exposing the specimen to a detection magnetic field having a detection magnetic field strength B after a 90° pulse has been applied to M; e. switching off the magnetic field for the re-cycle delay time RD; f. repeating steps b. to e. a number of cycles n, wherein the value of time τ varies at each cycle; g. fitting the values of magnetisation M acquired after n repetitions of steps b. to e. (magnetisation intensity vs. τ) according to a monoexponential Bloch equation to produce a respective magnetisation recovery / decay curve andcalculate the proton longitudinal relaxation rate R at the respective magnetic field strength B ; h. repeating steps b. to g. a number of times m, wherein the value of the measuring magnetic field strength B varies each time, thereby calculating at least three proton longitudinal relaxation rates R , R , R at the at least three measuring magnetic field strengths B ; i. calculating: − a Ratio of two relaxation rates R ; − a Sum of at least two relaxation rates R ; − an extracellular water mole fraction p by fitting the at least three magnetisation recovery curves by means of the two-Site eXchange model; − an intracellular residence lifetime τ by fitting the at least three magnetisation recovery curves by the two- Site eXchange model; j. processing the parameters calculated in step g. and i. for assessing the grade of tumour tissue in the specimen by:

1. obtaining a sequence of Q-1 ordered cut-off values C, C, ..., C , calculating a parameter P = Sum * τ, wherein the specimen is grade Q if P has a second positioning with respect to a last cut-off value C , is grade q if P has a first positioning with respect to a cut-off value C and a second positioning with respect to a cut-off value C , and is grade 1 if P has a first positioning with respect to a first cut-off value C; or 2. calculating a score of Q discriminant functions f involving at least three parameters selected among the Ratio, the Sum, the extracellular water mole fraction p, the intracellular residence lifetime τ and the relaxation rate R obtained for a measuring magnetic field strength B , wherein q = 1 to Q, and wherein the specimen is grade q if f has the highest score.

2. The method according to claim 1, wherein the acquisition is carried out at a detection magnetic field strength B comprised between 5 MHz and 20 MHz expressed in terms of proton Larmor frequency.

3. The method according to any one of the preceding claims, wherein the measuring magnetic field strength B is comprised in the range 0.01 to 20 MHz in terms of proton Larmor frequency.

4. The method according to any one of the preceding claims, wherein the relaxation rates R (=1 / T ) are determined by applying the following Bloch equations:wherein ^^^^0^^^^is the highest measured magnetization M at B ; ^^^^^^^^0 is the magnetization M at τ = 0 if B > 7 MHz or; and equation [1] is resolved for B > 7 MHz, equation [2] is resolved for B5. The method according to any one of the preceding claims, wherein the Ratio is calculated by ratioing two relaxation rates R at two different measuring magnetic field strengths B , wherein the relaxation rate R in the numerator is the relaxation rate R determined to a value of the measuring magnetic field strength B comprised between0.01 and 0.15 MHz, and the relaxation rate R in the denominator is the relaxation rate R determined to a value of the measuring magnetic field strength B comprised between 0.15 and 10 MHz.

6. The method according to any one of the preceding claims, wherein the Sum is calculated by summing two relaxation rates R at two different measuring magnetic field strengths B 7. The method according to any one of the preceding claims, wherein the intracellular residence lifetime τ and the tissue extracellular water mole fraction p are calculated by means of the following equation(s):R is the longitudinal relaxation rate constant of the extracellular water protons of Matrigel at the measuring magnetic field strength B ; R is the longitudinal relaxation rate constant of the intracellular water protons of the specimen at the measuring magnetic field strength B ; ^^^^0^^^^is the highest measured magnetization M at B ; ^^^^^^^^0 is the magnetization M at τ = 0 if B > 7 MHz orthe magnetization M atτ = τ (p / p) [7]; and equation [3a] is resolved for B > 7 MHz, equation [3b] is resolved for B8. The method according to any one of the preceding claims, wherein the specimen is a tumour tissue(s) having a weight comprised between 15 mg and 115 mg as obtained by the surgeon in the operative room.

9. The method according to any one of the preceding claims, wherein the steps d. to f. are carried out at a temperature25 °C, preferably about 10 °C.

10. The method according to any one of the preceding claims, wherein the polarization magnetic field strength B of the polarization magnetic field is comprised between 8 and 25 MHz.

11. The method according to any one of the preceding claims, wherein the polarization step b. is carried out for a polarization time t, wherein t is comprised in the range 0.5 to 1.5 s.

12. The method according to any one of the preceding claims, wherein the switching off of the magnetic field in step e. lasts for at least 0.5 sec.

13. The method according to any one of the preceding claims, wherein the number of cycles n ranges from 10 to 32.

14. The method according to any one of the preceding claims, wherein the time value τ ranges from 0.01 to 4 sec.

15. The method according to any one of the preceding claims, wherein the process steps b. to h. are carried out once employing a Matrigel specimen for determining the relaxation rate constants R at the magnetic field strengths B applied according to the monoexponential Bloch equation and the obtained R are used as the longitudinal relaxation rate constants of the extracellular water protons R in the application of the two-Site eXchange model.

16. An NMR relaxometer system comprising: - a solenoid connected to a magnet power supply unit and to a cooling system, wherein the solenoid defines a space for hosting a probe in which a biological sample can be inserted; - an electronic control unit that includes an interface coupled to the magnet power supply unit for controlling the power supply unit, a temperature controller coupled to the probe for controlling the temperature of the probe, a radiofrequency unit, and a data acquisition unit configured to acquire magnetization data from the biological sample; and - a processing unit configured to control the dataacquisition unit to execute the steps of the method according to any one of claims 1 to 15.

17. A computer program product loadable in the memory of a processing unit of an NMR relaxometer system according to claim 16 and comprising instructions to allow the NMR relaxometer system to execute the steps of the method according to any one of claims 1 to 15.

Images