Combined pH and metabolic activity imaging for response assessment in immunotherapy

Abstract

A computerized system for monitoring cancer treatment and related methods. The system includes an input port (IN) for receiving two or more input images acquired of a subject, the two or more input images being capable of recording metabolic activity and extracellular pH levels. An evaluator (EVAL) of the system is configured to evaluate i) changes in metabolic activity over time and ii) changes in extracellular pH levels over time based on the input images. The evaluator (EVAL) is also configured to establish an indication of the following based on the evaluation: whether there is a response to a cancer treatment associated with a lesion of the subject or whether the evaluation is inconclusive. The indication is output via an output interface (OUT).

Description

Technical Field

[0001] The invention relates to a computerized system for monitoring a cancer therapy, a computer implemented method for monitoring a cancer therapy, a computer program element and a computer readable medium. Background Art

[0002] With the introduction of checkpoint blockade immunotherapy (CBIT), the treatment of modulating the immune system has become an important class of cancer treatments, for which the FDA has approved its use in many cancer types, even for first-line cancer treatment. These treatments can achieve durable responses in advanced patients. However, depending on the cancer type, for a single treatment, the percentage of responders is only between 15% and 30%. Another characteristic of checkpoint blockade therapy is the late time point at which responses begin to become apparent.

[0003] The basic mechanism of CBIT is to activate non-cytotoxic T cells that can recognize tumors as foreign bodies. By blocking the checkpoint molecules on the targeted cells, effector functions can be produced, and by infiltrating T cells, tumor cells can be killed. Due to the memory of the activated T cells, this response can be persistent. However, several mechanisms leading to resistance to treatment have been described. For example, see the review of P.Sharma et al. (Cell, Vol. 168, pp. 707-723, 2017). So far, reliable biomarkers that can predict the response of patients have not been identified. Therefore, early response measurement is very important clinically. Summary of the invention

[0004] Therefore, there is a need for better systems and / or methods for cancer treatment monitoring.

[0005] The objects of the invention are solved by the subject-matter of the independent claims, wherein further embodiments are included in the dependent claims. It should be noted that the following described aspects of the invention apply equally to the method of cancer therapy monitoring, to the computer program element and to the computer readable medium.

[0006] According to a first aspect of the present invention, there is provided a computerized system for monitoring cancer treatment, comprising:

[0007] an input port for receiving two or more input images acquired of a subject, the two or more input images being capable of recording metabolic activity and extracellular pH levels;

[0008] an evaluator configured to evaluate, based on the input image, i) changes in metabolic activity over time and ii) changes in extracellular pH levels over time, and to establish, based on the evaluation, an indication as to whether a response to a cancer treatment associated with a lesion in the subject exists or whether the evaluation is inconclusive; and

[0009] An output interface is used to output the indication.

[0010] According to one embodiment, the assessor is configured to establish an indication that there is a response to the cancer treatment if: i) there is an increase in metabolic activity and there is an increase in extracellular pH levels, or ii) there is a decrease in metabolic activity and an increase in extracellular pH levels.

[0011] According to one embodiment, the evaluator is configured to establish an indication of whether the response is a pseudo-progression or a real progress.

[0012] According to one embodiment, said evaluator comprises an image segmenter for segmenting said two or more images for a region of interest "ROI", wherein said evaluation of said metabolic activity and said changes in said extracellular pH level is based on image values ​​in said ROI.

[0013] According to one embodiment, said evaluation of said metabolic activity and said change of said extracellular pH level is based on image values ​​in a border portion of said ROI.

[0014] According to one embodiment, said evaluation of said changes in said metabolic activity and said extracellular pH level is based on calculating one or more ratios of said changes.

[0015] According to one embodiment, at least one image recording the metabolic activity is based on a PET image (signal), and / or, wherein at least one image recording the pH level is based on an MR (magnetic resonance) signal (e.g., an MR image) or at least can be derived from an MR image signal. In other words, the image recording the metabolic activity or pH will naturally have a suitable contrast or intensity according to the needs of converting the image or otherwise processing the image, for example, in PET, the radioactivity intensity recorded in the voxel position is converted into an SUV (standard uptake value) or so-called maxSUV to better represent a measure for metabolic activity. MR imaging as used in this article should be broadly understood to include magnetic resonance spectroscopy imaging.

[0016] According to a second aspect, there is provided a computer-implemented method for monitoring cancer treatment, comprising the steps of:

[0017] receiving two or more input images acquired of a subject, the two or more input images being capable of recording metabolic activity and extracellular pH levels;

[0018] evaluating i) changes in metabolic activity over time and ii) changes in extracellular pH levels over time based on the input image;

[0019] establishing an indication based on the evaluation as to whether there is a response to a cancer treatment associated with the subject's lesion or whether the evaluation is inconclusive; and

[0020] The indication is output.

[0021] According to one embodiment, said establishing comprises establishing that there is an indication of a response to said cancer treatment when: i) there is an increase in metabolic activity and there is an increase in extracellular pH levels, or ii) there is a decrease in metabolic activity and an increase in extracellular pH levels.

[0022] According to one embodiment, the indication established is able to distinguish between false progress and real progress.

[0023] According to one embodiment, the at least one image recording the metabolic activity is based on a PET image (e.g. can be derived therefrom), and / or wherein the at least one image recording the extracellular pH level is based on an MR image (e.g. can be derived therefrom). Radiological modalities are advantageous because they provide information about lesions with minimal invasiveness compared to pathologies which would require serial biopsies and ideally biopsies from all lesions.

[0024] According to one embodiment, the establishing step comprises applying machine learning to the images.

[0025] According to a third aspect, a computer program element is provided which, when executed by at least one processing unit, is adapted to cause the processing unit to perform a method according to any of the above mentioned aspects or embodiments.

[0026] According to a fourth aspect, a computer readable medium is provided, having a program element stored thereon.

[0027] Proposed herein are systems and methods for improving clinical interpretation of the evaluation of tumor lesions by non-invasive data collection (particularly radiological imaging) using metabolic markers combined with a modality sensitive to changes in extracellular pH. By analyzing and combining the two (changes in metabolic activity over time and changes in extracellular pH levels over time), it is very possible to distinguish tumor cell activity from immune cell activity, and therefore infer whether the tumor is in a state of progression, stability, or response, or even regression with currently administered therapy.

[0028] In the proposed system, it is also possible to distinguish situations when metabolic evaluation by radiology is inconclusive, and in this way a correct interpretation of the radiological images is guaranteed.

[0029] definition

[0030] "Response" to cancer treatment is either regression or pseudoprogression.

[0031] "Remission" is a state in which cancer cells are dying in response to treatment.

[0032] The antonym of "response" is non-response including "progression" (cancer cell proliferation) or lesions remaining "stable".

[0033] "Pseudoprogression" is the activity of a lesion that looks or behaves like progression but is not actually progression. The term is particularly relevant in immunotherapy, where the metabolic activity of immune cells ("T cells") can be mistaken for that of the tumor.

[0034] As used herein, an "increase in pH levels" indicates lower acidity at higher numerical readings in a pH scale from 0 to 14. For example, a reading of "8" is at a higher pH level than a reading of "5."

[0035] "Tumor" or "(cancer) lesion" are used interchangeably herein.

[0036] "Metabolic activity" as used herein includes the chemical conversion (or processing) of certain compounds ("metabolites") by the cells of the human or animal body. An important but non-limiting example is the conversion of glucose to obtain energy.

[0037] "Image" as used herein should be interpreted broadly to include not only the entire image signal array, but also in extreme cases a single image signal for a single voxel location or such image signals selected for a subset of voxel locations in the entire field of view. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Exemplary embodiments of the present invention will now be described with reference to the following drawings, in which:

[0039] Figure 1 A system block diagram of a computerized system for cancer treatment monitoring is shown; and

[0040] Figure 2 A flow chart showing a method for monitoring response to cancer therapy is shown. DETAILED DESCRIPTION

[0041] Now first refer to Figure 1 The schematic block diagram in Figure 1A computerized system for monitoring a cancer treatment administered to an object (e.g., a human or animal patient) is shown. In particular, immunotherapy such as CBIT is contemplated herein in preferred embodiments, but this does not exclude other types of cancer treatments. In other words, although immunotherapy is primarily mentioned below, the teachings herein are also suitable for use in other cancer treatments (e.g., chemotherapy or targeted therapy or other treatments), and any of these other cancer treatments is contemplated herein in alternative embodiments.

[0042] Briefly, the proposed system allows to reliably and early (within weeks rather than months) determine whether a patient is responding to cancer treatment, i.e., whether the cancer lesion is in regression, stable, progressive or whether there is so-called pseudo-progression.

[0043] In case of progression, it is possible to switch earlier to a different treatment that may be more effective. This is because it has been found, as indicated in the background art, that in a large number of cases, patients do not actually respond to immunotherapy, i.e., although immunotherapy is applied, the cancer lesions actually keep proliferating.

[0044] In one embodiment, the proposed computerized immunotherapy monitoring system runs in software on a data processing unit PU (e.g., a computer device), or runs on multiple computers connected in a network, or is implemented in hardware (e.g., a programmed microchip, such as an FPGA), or is implemented on a hardwired chip or other circuit.

[0045] The operation of the system is based on images. In particular, images of the patient are acquired by a suitable imaging modality. Images (referred to herein as input images) are fed into the system via one or more input ports IN. The input images include X The input images also include a second set of images (referred to as control images IM_1), and this is a baseline image IM_0 acquired at time T0 before starting cancer treatment at time T1. X The set of one or more control images IM_1 may be acquired after or during administration of the treatment. When the same or similar imaging modality has been used for IM_0, they are used for IM_1.

[0046] Applicants have found that the study of both changes in extracellular pH levels and changes in metabolic activity at or near the lesion during the course of treatment provides highly reliable biomarkers for specifically evaluating one or more immunotherapy responses to other cancer treatment regimens. The present invention then proposes to measure changes in these two factors (extracellular pH levels and metabolic activity) based on appropriate imaging that is sensitive to these two factors.

[0047] In one embodiment, suitable imaging modalities include MR (magnetic resonance) imaging and PET (positron emission tomography) imaging. More particularly, MRI imager MRI is used to collect MR images with suitable contrast agents for imaging of extracellular pH levels. In particular, in one embodiment, for pH level imaging, CEST-MRI is used. Advantageously, imaging techniques that can distinguish extracellular pH from intracellular pH are used herein. Suitable CEST MRI markers for extracellular pH imaging contemplated herein include appropriate CEST reagents, for example, 3-aminopropyl phosphate (3-APP) or 2-imidazole-1-yl-3-ethoxycarbonyl propionic acid (IEPA). It has been found that CEST measurements not only vary with pH, ​​but also vary with CEST reagent concentrations. However, it is possible to remove confusing concentration factors by analyzing the relative CEST effects ("ratio determination CEST") from different interchangeable molecules.

[0048] Specific types of CEST are also contemplated herein (eg, APT imaging "amide proton transfer"). Further details on APT can be found in Zhou J et al. (Nature Medicine, September 2003, p. 1085).

[0049] MR spectroscopy (MRS) or MR spectroscopy imaging (MRSI) is also conceived to measure pH levels. In particular, intravenous injection of hyperpolarized 13C-labeled bicarbonate results in rapid production of hyperpolarized 13CO2 in a reaction catalyzed by carbonic anhydrase, which depends on the local pH. The ratio of the 13C signal intensity from H13CO3 and 13CO2 measured using MRS can be used (primarily) to calculate extracellular pH.

[0050] A review of in vivo pH measurement techniques can be found in the article by X. Zheng et al. (Journal of Nuclear Medicine, Vol. 51, No. 8, pp. 1167-1170, August 2010).

[0051] PET Imager PET is used to image metabolic activity based on suitable tracers such as FDG (fluorodeoxyglucose, preferably 18F), but imaging of other compounds besides sugars / glucose (e.g., lactate) is also contemplated. Suitable tracers contemplated herein for imaging metabolic activity include 18 F-FDG and 64 Cu-DOTA-pHLIP and other tracers.

[0052] Metabolic activity involves the chemical processing of certain compounds ("metabolites") by cells of the human or animal body. An example is the processing of glucose for energy. This activity or similar activities can be measured in functional radiological imaging (e.g., PET or SPECT) by injecting a radioactive tracer prior to imaging that resembles the metabolites and is therefore "taken up" by the cells for processing like real metabolites (radioisotope tracing). This creates an image contrast mechanism in which high activity corresponds to high radioactivity concentration and therefore high image intensity / contrast. For example, as contemplated herein 18 F-FDG is a type of glucose that is known to be consumed in large quantities by tumors (Warburg effect). However, other tracers may be used instead, depending on the metabolic activity or metabolic pathway for which one wishes to radiolabel and image the isotope.

[0053] Sometimes, the image values ​​according to the PET images are converted into SUV (standard uptake values) for better comparison, wherein the measured radioactivity is standardized relative to the patient's weight and the initial injection dose of the tracer. "SUVmax", which is the maximum value of the SUV in a specific ROI lesion, is often used in clinical decision making. Although the mean SUV within the ROI can be used instead, SUVmax is preferably used in this article because it has been found that SUVmax can better estimate the true SUV than the mean SUV. It has also been found that SUVmax has significantly improved reproducibility. See "PET / CT Standardized Uptake Values ​​(SUVs) in Clinical Practice and Assessing Response to Therapy" by Paul E Kinahan et al. (Semin Ultrasound CT MR, December 2010, Vol. 31, No. 6, pp. 496-505).

[0054] Herein, the MRI and PET baseline images acquired at time point T0 before the start of cancer treatment are designated as M0 for pH level-sensitive imaging, and "P0" is used as a designator for metabolic activity-sensitive PET imaging.

[0055] At one or more time points T1 after or during treatment, control images IM_1 are obtained using MRI and PET imaging modalities, and these control images are indicated herein as M1 and P1 , respectively.

[0056] In summary, at least four images are then acquired, but the corresponding imaging pairs (M0, M1) and (P0, P1) can be fused into single images, respectively, in which case the input image comprises two images. There may also be imaging modalities that are capable of producing images with contrast for both pH levels and metabolism, in which case the input image also comprises only two images at any one monitoring moment. In addition to this functional imaging, there may be further images (baseline and control images) for morphological imaging, such as CT or radiographic images, phase contrast images or ultrasound (US) images or other images. In particular, CT morphological information can be used for registration with MRI images that also have a morphological component with high resolution. Therefore, depending on the embodiment, there may be 6 or more input images, depending on how the MR is analyzed.

[0057] After input images comprising two pairs M0 / M1 and P0 / P1 are input into the system at input port IN, a registration module REG is used to register the corresponding image pairs voxel by voxel with each other, that is to say, i) M0 is registered to M1 and P0 is registered to P1, or ii) M0 is registered to P0 and M1 is registered to P1. However, the latter registration is optional, in particular if the images are naturally registered and no specific registration step is required, which may be the case, for example, when a combined PET-MR tandem imager is used.

[0058] The image segmenter unit SEG then processes the input image to segment the image for a portion of the image including a region of interest (ROI), in particular a cancer lesion. Background image information can be cut out. Using the segmenter SEG is a preferred case, but not necessarily for all embodiments. Alternatively, the segmentation is done first and then the registration. The registration can then be focused only on the ROI, which is generally easier and consumes less CPU time than in the case of whole body registration.

[0059] After an optional registration and / or an optional segmentation, the input image is forwarded to an evaluation unit EVAL which is configured to evaluate extracellular pH level changes and metabolic changes at or near the lesion based on the input image.

[0060] The evaluation unit EVAL is further configured to establish an indication based on the evaluation of the pH level and metabolic changes indicating whether the lesion is responding or whether the evaluation is inconclusive.

[0061] The indication may be provided in graphical form and / or in alphanumeric / textual form. The indication may be visual or tactile, or may be an audio signal or other signal. If the indication is visual, the indication may be drawn on the display device MT. The indication may be stored in a memory MEM (e.g. a medical database or other database), or may be processed in other ways.

[0062] The indication may be characterized in a probabilistic manner, so that the indication does not indicate a lesion response, but rather may indicate the likelihood of such a response, preferably listing the probability quantification result as a number between 1 and 0 or as a percentage.

[0063] The evaluation operation of the evaluation unit EVAL will now be described in more detail. In one embodiment, the evaluation unit EVAL is operable to form a response difference image based on the baseline image and the control image. In particular, based on the two image pairs, the differences that follow can be formed as |M1-M0| and |P1-P0|, which quantify the local changes of the two indicators (pH level and metabolic activity), respectively. The order of the voxel-by-voxel subtraction "-" is not important, because the above-mentioned local changes are indicated by the absolute value operator |*|.

[0064] Now, each voxel in the difference image represents the extracellular pH level change and the metabolic activity change, respectively. In one embodiment, the evaluation includes thresholding (one for the pH level factor and one for the metabolic activity factor) to compare the difference with the corresponding threshold value. With respect to establishing an indication, in one embodiment, if both the pH level change and the metabolic activity change exceed the corresponding threshold value, this will indicate that there is a response to the cancer treatment (particularly pseudo-progression) at least for the considered location. However, this is only an exemplary strategy for establishing the desired indication for treatment monitoring based on thresholding, and more strategies will be further discussed below at Table 1.

[0065] In order to form a more robust evaluation formula and to establish a more reliable indication of responsiveness or non-responsiveness, it is preferred not to rely solely on a single voxel position (although this may still accomplish the task in some exceptional cases). In a preferred embodiment, the evaluation of both pH level changes and metabolic activity changes as described above is evaluated for multiple (particularly all) voxels in the segmented region corresponding to the cancer lesion ROI.

[0066] In one embodiment, only voxels inside the cancer lesion segmentation result are considered, while other voxels outside the ROI are ignored. In yet another embodiment, a boundary margin is defined at the edge and within the ROI segmentation result, and only voxels within the margin are considered in the described threshold-based evaluation, because the applicant observes that the response (if any) may first occur in the boundary area. In other embodiments, a boundary margin at the edge outside the ROI is established with a suitable width, and only voxels with the external margin are considered for the above-mentioned threshold-based evaluation. In a third embodiment, the margin is partially inside the segmentation edge and partially outside the segmentation edge.

[0067] As an alternative or additional scheme to the above-mentioned "double" thresholding of the difference of pH level and metabolic activity, for each voxel, a ratio is formed based on the corresponding pH level change and metabolic activity change according to the difference image. For each voxel, there is an associated corresponding ratio pH / ma (or its reciprocal), where "pH" is the local pH change and "ma" is the local metabolic change. These ratios are then compared with a single threshold instead of two thresholds for each voxel. It has been found that an increased pH / ma ratio is an indication of responsiveness. Again, as described above, for double thresholding, these ratios can be considered to be evaluation results only within the segmented cancer lesion ROI or in its corresponding boundary portion. Alternatively, the ratios for relative changes can be used as (P1-P0) / P0 and / or (M1-M0) / M0, respectively.

[0068] As a further alternative, ratios are first formed and then the ratios are differentiated to assess changes. In particular, a difference image is formed based on a voxel-by-voxel ratio image of the baseline image and the control image according to |P1 / M1-P0 / M0 or (P1-P0) / P0 / (M1-M0) / M0|, where " / " indicates voxel-by-voxel division.

[0069] In all of the above embodiments, while the entire image could be evaluated voxel by voxel as described above, this would likely increase computational time, and it is preferable to restrict pixel evaluation to in or near the segmented cancer lesion locations.

[0070] In all the above threshold-based embodiments, image noise may corrupt the image information as recorded in the input image. Therefore, it is preferred to take this factor of image noise into account in the thresholding to make the computerized evaluation more robust. For example, changes within the noise margin Δ are interpreted as noise-induced fluctuations only and not as real changes, and are therefore ignored. In other words, the threshold TR is an addition result and comprises a part representing the real change c and a part representing the noise margin Δ, respectively: TR=c+Δ. Alternatively or additionally, the input image is passed through a noise filtering stage (not shown) to eliminate or at least reduce the noise contribution by means of a smoothing scheme or other schemes. In the case of filtering, it may not be necessary to modify the threshold by means of Δ as described in the previous sentence.

[0071] As an alternative or extension of all of the above, local changes in pH level and metabolism can be mathematically combined into an expression other than a ratio (e.g., a weighted sum or other mathematical combination) to thereby formulate different metrics for response evaluation. For example, a formula for changes in metabolic activity (relative or absolute) can be formed based on SUVmax or other metrics.

[0072] The operation of establishing an index for treatment monitoring will now be described in more detail, with reference now to the following table (Table 1), wherein "+" indicates an increase in the corresponding channel (pH channel or metabolic channel), "-" indicates a decrease, and "0" indicates no change (it should be understood that the noise level factor has been taken into account by the noise margin Δ mentioned above, or the image has been noise-filtered):

[0073] Table 1

[0074] Metabolism (FDG) pH(CEST) in conclusion + + Pseudo progress (response) + 0 Inconclusive, most likely progress + － Tumor progression 0 + Inconclusive, most likely response 0 0 Stable tumor 0 － Tumor progression － + Fading (Response) － 0 Inconclusive, most likely response － － No conclusion

[0075] In the above table, metabolic assessment was based on PET / CT (18F-FDG) images, and extracellular pH was based on images acquired by MRI (CEST).

[0076] In summary, the evaluator EVAL as contemplated herein is configured to establish and distinguish any one or more (in particular all) of the following indications:

[0077] i. If the FDG signal increases over time and the pH decreases over time, this indicates tumor progression.

[0078] ii. If the FDG signal increases over time and the pH value increases over time, this indicates pseudo-progression of the tumor.

[0079] iii. If the FDG signal increases over time and the pH is stable over time, this indicates that the metabolic assessment is inconclusive.

[0080] iv. If the FDG signal is stable over time and the pH is also stable over time, this indicates a stable tumor.

[0081] v. If the FDG signal is stable over time and the pH decreases over time, this indicates tumor progression.

[0082] vi. If the FDG signal is stable over time and the pH increases over time, this indicates that the metabolic assessment is inconclusive.

[0083] vii. If the FDG signal decreases over time and the pH is also stable over time, this indicates that the metabolic assessment is inconclusive.

[0084] viii. If the FDG signal decreases over time and the pH increases over time, this indicates that the tumor is regressing.

[0085] ix. If the FDG signal decreases over time and the pH decreases over time, this indicates that the metabolic assessment is inconclusive.

[0086] The above table can be implemented as a lookup table in the evaluator EVAL.

[0087] It will be appreciated that the proposed system is configured to distinguish between progression and pseudo-progression of lesions / tumors according to the first and third rows (and items i) and ii)) in the table. In particular, this particular discrimination capability is particularly advantageous for immunotherapy monitoring. More particularly, the relative local changes in metabolic activity are compared with the relative local changes in pH. In this way, an attempt is made to distinguish between the metabolic activity of tumor cells and the metabolic activity of immune cells, and in this way an attempt is made to distinguish between pseudo-progression and real progression of so-called lesions. Constant or increased metabolic activity at an increased pH (less acidic) is interpreted as a signature for a response to treatment, while increased activity at a constant or decreased pH indicates real progression.

[0088] The above indications in the table and their expressions i)-ix) apply equally to imaging modalities other than 18-FDG-PET and CEST-MR.

[0089] As an alternative to the voxel-by-voxel evaluation of thresholds and metrics described above that require empirical learning, machine learning can instead be used to examine the distribution of changes in pH levels and metabolic activity across input images to establish patterns for response to treatment.

[0090] To this end, historical outcome data collected over time from hospital information systems HIS for patients undergoing the same or similar cancer treatments can be used to train a suitable machine learning algorithm. The outcome training image set is naturally labeled for each patient by the outcome of the treatment, since it is already known which image in the training set corresponds to a responsive or non-responsive pattern.

[0091] In one embodiment, a neural network ("NN") having two output nodes representing responsiveness and non-responsiveness to immunotherapy, respectively, or a third node for representing inconclusiveness can be used. Additional output nodes can be added according to any set of indications / conclusions from the third column in the table above. The input layer nodes correspond to the size of the difference image for changes in pH level and metabolic activity. There can be more than one (e.g., 3 or 4) hidden layers in the deep learning architecture. The NN can then be trained using the resulting images as a labeled training set to learn the relationship between i) the distribution of pH level changes and metabolic activity changes and ii) the desired indication / conclusion (e.g., response or non-response to a given type of cancer treatment in a combination of different types of cancer treatments). The network can be trained by a suitable backward / forward algorithm as described elsewhere.

[0092] Once a given input image is suitably trained for changes in pH levels and changes in metabolic activity, it can be input into the NN (e.g., in the form of a difference image derived from a control image and a baseline image) and forward propagated through the NN to directly establish indications by examining the counts at the output nodes. However, NN is not the only machine learning method that can be used, and statistical schemes such as support vector machines (SVM), regression, or other statistical schemes may also be used instead or in combination.

[0093] Hypoxic areas are not conducive to immune effector cells. Therefore, it can be beneficial to add imaging based on hypoxia markers to supplement the information from pH level and metabolic imaging. In hypoxic areas, the metabolic rate is mainly attributed to tumor cells only. The response can cause the size and / or quantity of the hypoxic areas in the lesion to decrease. A suitable imaging modality for hypoxic imaging is 19F MR spectroscopy imaging. This is because tumor cells increase proton (H+) production, and this is associated with incomplete vascularization, and poor extracellular space can create a hypoxic environment. When the method based on combined pH level changes and metabolic activity changes as proposed in this article is inconclusive, imaging for hypoxia can be particularly useful.

[0094] Reference now Figure 2 , Figure 2A flow chart of a method for evaluating an immunotherapy response is shown. It should be understood that the following method steps may also be understood as teachings that proceed on their own. The computer-implemented method as set forth herein comprises the following steps.

[0095] At step S210, baseline images and control images of an object that has undergone or is undergoing cancer treatment (e.g., immunotherapy) are collected. The baseline image is collected before starting immunotherapy, and the control image is collected during or after the treatment is administered. The input image is sensitive to extracellular pH levels and metabolic activity. In one embodiment, this can be achieved by using MRI imaging for pH levels and PET imaging for capturing metabolic activity, but combinations of other imaging modalities are also contemplated. It is necessary to convert the image as supplied by the modality into a suitable quantity, for example, in PET, the radiological intensity as recorded in the natural PET image is converted into SUV or SUVmax for better comparison. In addition to functional images for metabolic activity and pH, morphological imaging modalities (e.g., CT or radiography) can also be used to capture the spatial characteristics of the lesion. This can preferably be accomplished by using a tandem CT-PET imager. Instead of PET, SPECT can also be used in this article.

[0096] At optional step S220, the pH level sensitivity baseline is registered with the control image, and the metabolic activity sensitivity baseline is also registered with the control image. This step may not necessarily be required if the images are already naturally registered.

[0097] At optional step S230, segmentation is performed on the input image to segment the cancer lesions. The order of S220 and S230 can be reversed so that the image is segmented first and then the segmented parts are used as clues when performing registration.

[0098] Intelligent segmentation schemes are also contemplated herein, which provide not only global ROI segmentation, but also fine segmentation into sub-regions within the ROI. It is also possible to merge the segmentation step and the registration step into a single algorithm.

[0099] In each embodiment, the segmentation is preferably configured to account for changes in the size / volume of the ROI across corresponding baseline and control images. This is useful when the size of the tumor ROI may change during treatment. Segmentation schemes contemplated herein include: elastic or inelastic segmentation schemes, model-based or non-model-based segmentation schemes, optical flow-based segmentation schemes, and other segmentation schemes, used alone or in combination.

[0100] At step S240, the image values ​​in the input image are then used to evaluate changes in pH level and metabolic activity over time. This can be done by forming a corresponding difference image, which is formed by voxel-by-voxel subtraction of a baseline image based on pH level sensitivity from a control image on the one hand and voxel-by-voxel subtraction of a metabolic activity sensitivity baseline from a control image on the other hand. Each voxel value in the difference image corresponds to a pH level change or a metabolic activity change, respectively. The difference thus obtained can be compared with a threshold value. Additionally or alternatively, ratios can be formed by the pH level change over the metabolic activity change, and these ratios can then be compared with a threshold value. As a further alternative, a difference image is formed by voxel-by-voxel ratio images of the baseline image and the control image according to |P1 / M1-P0 / M0|, where " / " indicates voxel-by-voxel division.

[0101] Instead of a voxel-by-voxel evaluation, the evaluation may be coarsened into chunks, e.g., chunks of sub-ROIs of the entire ROI. The voxel-by-voxel evaluations may then be integrated or summed or otherwise combined into a corresponding single score or metric for the entire ROI or sub-ROI under consideration. All or the above described herein for voxel-by-voxel processing therefore equally applies to a coarser processing based on chunks (that is, subsets of voxels) (e.g., the ROIs or sub-ROIs mentioned).

[0102] At step S250, after comparing the difference or ratio with a suitable set of thresholds or by calculating other suitable metrics based on the (absolute or relative) difference or ratio, an indication about the applied cancer treatment, e.g., responsiveness or non-responsiveness or inconclusiveness, may be established. The indication may be refined into any one or more of the (sub-) indications according to the third column in the table above.

[0103] In one embodiment, the segmented ROI according to step S230 includes the entire lesion, but this can be refined into areas of similar metabolic activity by sub-segmentation and / or into a core area and a peripheral or border area, or refined based on CT or MRI morphological information. The segmentation step S240 is then performed for each sub-segmentation. At step S250, an overall conclusion / indication can be derived by forming a (weighted) average result over the entire tumor lesion or by forming different metric expression contents. In immunotherapy, it can be expected that there will be an earlier response at the border area of ​​the tumor, so more attention will be paid to the border area than the core area in the evaluation.

[0104] In one embodiment, each voxel is treated individually as described above, wherein each voxel has its own evaluation and indication. An overall conclusion / indication for the entire ROI (or sub-ROI) can then be formed by a majority vote across all voxels considered or by taking a simple average or weighted average of all evaluations (and then deriving an overall indication based on the simple average or weighted average according to the above table). Evaluations for voxels from border areas (or from any other specified area) can attract higher weights in the averaging process. Border area voxels can also attract more votes in a majority vote. Preferably, the evaluation is limited to voxels / pixels in the segmented area only.

[0105] If morphological information can be obtained based on the baseline CT images and the control CT images, the factor of volume change can be considered in the above evaluation. In particular, correction for lesion volume change will be necessary. In one embodiment, volume change can be taken into consideration by standardizing the above evaluation value based on volume size.

[0106] With respect to expression of the indication, lesion volume change may be expressed as a separate metric, or the volume change may be combined into a metabolic activity-pH level metric as discussed above.

[0107] The indication may be established as a binary indication as to whether a response is present or absent, or the indication may be expressed in terms of probability or statistics (eg, a percentage value).

[0108] For example, for a segmented ROI, a percentage value of those ratios or differences above a relative threshold relative to those below the threshold may be calculated to yield a value between 0 and 1 that may be interpreted as a probability for treatment response.

[0109] Instead of the above threshold-based method, a machine learning algorithm or a statistical algorithm may be used. It should be understood that, in particular, in the machine learning algorithm, the above steps of evaluating S240 changes and establishing S250 indications are substantially combined into one step.

[0110] Then, at step S260, the indication may be output on the monitor device MT, for example on the screen of a laptop computer, a desktop computer, a smartphone or in other ways.

[0111] The indications may be suitably color coded or otherwise graphically plotted, for example, "red" indicating no response and "blue" indicating a response, or shades in between, etc.

[0112] Instead of or in addition to displaying the indication on the display device, the data may be stored on a memory MEM (eg a database) or the data may be processed in another way.

[0113] It will be appreciated that at different times If multiple control images are acquired at the same time, the segmentation step S240 can be repeated multiple times (N>1). The change in ratio can then be plotted over time for each voxel. However, in principle, the proposed method can operate on a single pair of images or a baseline image and a control image.

[0114] All of the above embodiments can be supplemented, modified or extended by the following:

[0115] Instead of FDG, other direct lactate imaging can also be used. As an alternative to MRI pH imaging, pH-specific PET contrast agents can be used.

[0116] Systemic LDH (lactate dehydrogenase) measurements can be used to compare with the integrated FDG activity of all lesions before and after the start of treatment.

[0117] LDH and cell counts in the blood can be monitored before and during treatment to gain additional information about the patient's response in conjunction with radiological information.

[0118] FLT (fluorothymidine F-18)-PET can be added to visualize the proliferative activity of tumor lesions. This is primarily interpreted as a signature of progression, however, this must be corrected for potential contributions from the proliferation of immune cells. FLT information can help further increase confidence in conclusions drawn from metabolic and pH imaging.

[0119] More acidic tumor regions may not allow successful effector cell activity and thus would be expected to have lower metabolic activity, even when recruited to the tumor.

[0120] Immune cell recruitment (more immune cells arriving at the tumor site) can also be visualized by adding immuno-PET imaging (e.g. using radiolabeled anti-CD8 agents).

[0121] One or more features or methods disclosed herein may be configured or implemented as / with circuits encoded in a computer-readable medium and / or a combination thereof. The circuits may include discrete and / or integrated circuits, application specific integrated circuits (ASICs), systems on chips (SOCs) and combinations thereof, machines, computer systems, processors and memories, computer programs.

[0122] In a further exemplary embodiment of the present invention, a computer program or a computer program element is provided, which is characterized by being adapted to execute the method steps of the method according to one of the preceding exemplary embodiments on an appropriate system.

[0123] Therefore, the computer program element can be stored in a computer unit, which can also be part of an embodiment of the present invention. The computing unit can be suitable for executing or causing the execution of the steps of the above-mentioned method. In addition, the computing unit can be suitable for operating the components of the above-mentioned device. The computing unit can be suitable for automatic operation and / or running the user's command. The computer program can be loaded into the working memory of a data processor. Therefore, a data processor can be equipped to perform the method of the present invention.

[0124] This exemplary embodiment of the invention covers both a computer program that right from the beginning uses the invention and a computer program that is converted by means of an update into a program that uses the invention.

[0125] Furthermore, the computer program element may be able to provide all necessary steps to fulfill the procedure of an exemplary embodiment of the method as described above.

[0126] According to a further exemplary embodiment of the present invention, a computer-readable medium, for example a CD-ROM, is proposed, wherein the computer-readable medium has a computer program element stored on the computer-readable medium, the computer program element being described by the preceding sections.

[0127] The computer program may be stored and / or distributed on a suitable medium (particularly, but not necessarily, a non-transitory medium), such as an optical storage medium or a solid-state medium supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunications systems.

[0128] However, the computer program may also be present on a network, such as the World Wide Web, and can be downloaded from such a network into a working memory of a data processor. According to a further exemplary embodiment of the present invention, a medium for making a computer program element available for download is provided, the computer program element being arranged to perform a method according to one of the previously described embodiments of the present invention.

[0129] It must be noted that embodiments of the present invention are described with reference to different subject matters. In particular, some embodiments are described with reference to method claims, while other embodiments are described with reference to apparatus claims. However, unless otherwise stated, a person skilled in the art will infer from the above and following descriptions that, in addition to any combination of features belonging to one type of subject matter, any combination of features relating to different subject matters is also considered to be disclosed in the present application. However, all features can be combined to provide synergistic effects that are more than the simple addition of the features.

[0130] Although the present invention has been illustrated and described in detail in the drawings and the foregoing description, such illustration and description should be considered illustrative or exemplary rather than restrictive. The present invention is not limited to the disclosed embodiments. Those skilled in the art will be able to understand and implement other variations to the disclosed embodiments in practicing the claimed invention by studying the drawings, the disclosure, and the claims.

[0131] In the claims, the word "comprising" does not exclude other elements or steps, and the word "a" or "an" does not exclude a plurality. A single processor or other unit may fulfill the functions of several items recited in the claims. Although certain measures are recited in mutually different dependent claims, this does not indicate that a combination of these measures cannot be used to advantage. Any reference signs in the claims should not be construed as limiting the scope.

Claims

1. A computerized system for monitoring cancer treatment, comprising: an input port (IN) for receiving two or more input images acquired of a subject, the two or more input images being capable of recording metabolic activity and extracellular pH level; an evaluator (EVAL) configured to evaluate, based on the input image, i) changes in metabolic activity over time and ii) changes in extracellular pH levels over time, and to establish, based on the evaluation, an indication as to whether a response to a cancer treatment associated with a lesion in the subject is present or whether the evaluation is inconclusive, wherein said evaluator (EVAL) comprises an image segmenter (SEG) for segmenting said two or more input images with respect to a region of interest "ROI", wherein said evaluation of said metabolic activity and said changes in said extracellular pH level is based on image values ​​in said ROI, wherein said assessing said changes in said metabolic activity and said extracellular pH level is based on calculating one or more ratios of changes in said metabolic activity and changes in said extracellular pH level; and An output interface (OUT) is used to output the indication.

2. The system of claim 1, wherein the evaluator (EVAL) is configured to establish an indication that there is a response to the cancer treatment when: i) there is an increase in metabolic activity and there is an increase in extracellular pH levels, or ii) there is a decrease in metabolic activity and an increase in extracellular pH levels.

3. The system according to claim 2, wherein: The evaluator (EVAL) is configured to establish an indication of whether the response is a pseudo-progression or a true progress.

4. The system according to any one of claims 1 to 3, wherein: The evaluation of the metabolic activity and the change in the extracellular pH level is based on image values ​​in a boundary portion of the ROI.

5. The system according to any one of claims 1 to 3, wherein: The at least one image recording the metabolic activity is based on a PET image and / or wherein the at least one image recording the pH level is based on an MR image.

6. A computer-implemented method for monitoring cancer treatment, comprising the steps of: receiving two or more input images acquired of a subject, the two or more input images being capable of recording metabolic activity and extracellular pH levels; evaluating i) changes in metabolic activity over time and ii) changes in extracellular pH levels over time based on the input image; segmenting the two or more input images with respect to a region of interest "ROI", wherein said evaluation of said metabolic activity and said change in said extracellular pH level is based on image values ​​in said ROI, wherein said evaluation of said changes in said metabolic activity and said extracellular pH level is based on calculating one or more ratios of changes in said metabolic activity and changes in said extracellular pH level; establishing an indication based on the evaluation as to whether there is a response to a cancer treatment associated with the subject's lesion or whether the evaluation is inconclusive; and The indication is output.

7. The method according to claim 6, wherein: The establishing includes establishing that there is an indication of a response to the cancer treatment when: i) there is an increase in metabolic activity and there is an increase in extracellular pH levels, or ii) there is a decrease in metabolic activity and an increase in extracellular pH levels.

8. The method according to claim 6 or 7, wherein: The established indications are able to distinguish pseudo-progression from real progression.

9. The method according to any one of claims 6 to 7, wherein: The at least one image recording the metabolic activity is based on a PET image and / or wherein the at least one image recording the extracellular pH level is based on a MR image.

10. The method according to any one of the preceding claims 6-7, wherein: The establishing step includes applying machine learning to the image.

11. A computer program element, which, when run by at least one processing unit (PU), is adapted to cause the processing unit (PU) to perform the method according to any of claims 6-10.

12. A computer-readable medium having stored thereon a program element according to claim 11.

Images