Biopsy-free detection and staging of cancer using a virtual staging score

• US 10,424,411 B2
• Filed: 09/19/2014
• Issued: 09/24/2019
• Est. Priority Date: 09/20/2013
• Status: Active Grant

• Current Assignee: Siemens Healthineers AG (Siemens AG)
• Original Assignee: Siemens Healthcare GMBH (Siemens AG)
• Inventors: El-Zehiry, Noha Youssry
• Primary Examiner(s): Sherali, Ishrat I
• Application Number:
  US15/309,165
• Publication Number:
  US 20170116387A1
• Time in Patent Office: 1,831 Days
• Field of Search: 382128, 382131
• CPC Class Codes: A61B 5/0066 Optical coherence imaging
• CPC Class Codes: A61B 5/032 Spinal fluid pressure
• CPC Class Codes: A61B 5/037 Measuring oesophageal pressure
• CPC Class Codes: A61B 5/055 involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
• CPC Class Codes: A61B 8/08 Detecting organic movements or changes, e.g. tumours, cysts, swellings A61B8/02 - A61B8/06 take precedence
• CPC Class Codes: G06F 18/2411 based on the proximity to a decision surface, e.g. support vector machines
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06T 2207/10081 Computed x-ray tomography [CT]
• CPC Class Codes: G06T 2207/10088 Magnetic resonance imaging [MRI]
• CPC Class Codes: G06T 2207/10101 Optical tomography; Optical coherence tomography [OCT]
• CPC Class Codes: G06T 2207/10104 Positron emission tomography [PET]
• CPC Class Codes: G06T 2207/10108 Single photon emission computed tomography [SPECT]
• CPC Class Codes: G06T 2207/10132 Ultrasound image
• CPC Class Codes: G06T 2207/30096 Tumor; Lesion
• CPC Class Codes: G06T 2207/30101 Blood vessel; Artery; Vein; Vascular
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/62 of area, perimeter, diameter or volume
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/30 for calculating health indices; for individual health risk assessment
• CPC Class Codes: G16H 50/70 : for mining of medical data, e.g. analysing previous cases of other patients
• CPC Class Codes: G16Z 99/00 : Subject matter not provided for in other main groups of this subclass

Abstract

A method for predicting a cancer staging score from medical image data includes receiving patient data for a plurality of patients, where patient data for each of the plurality of patients includes one or more of an image volume of a suspected tumor in an organ, blood test data, demographic data, and ground truth tumor staging scores for the suspected tumor in the organ, extracting features from the patient data, and using the features extracted from the patient data to train a classifier to predict a cancer staging score for a new patient from one or more of an image volume of a suspected tumor in the organ, patient blood test data and patient demographic data of that new patient.

Claims

1. A computer-implemented method of predicting a cancer staging score from medical image data, the method comprising the steps of:

receiving patient data for a plurality of patients, wherein patient data for each of said plurality of patients includes one or more of an image volume of a suspected tumor in an organ, blood test data, demographic data, and ground truth tumor staging scores for the suspected tumor in said organ;

extracting features from said patient data, wherein features from said demographic data include a patient'"'"'s weight, height, age, body mass index, body surface area, organ size, any past history of cancer occurrence, and any health index test results; and

using the features extracted from said patient data to train a classifier to predict a cancer staging score for a new patient from one or more of an image volume of a suspected tumor in said organ, patient blood test data and patient demographic data of that new patient.

2. The method of claim 1, wherein the image volume comprises one or more of an anatomical image volume, a functional image volume, or a fused anatomical/functional image volume, and features extracted from said image volume include one or more of anatomical features and functional features from said organ.

3. The method of claim 2, wherein extracting anatomical and functional features of said organ from said image volume comprises:

segmenting the organ in each of the anatomical and functional images to define one or more regions of interest (ROIs); and

extracting anatomical and functional features from the one or more regions of interest.

4. The method of claim 3, wherein a region of interest (ROI) is one of the whole organ or a segment of the organ.

5. The method of claim 2, wherein anatomical features include a volume of an ROI, a surface area of the ROI, a diameter of the ROI, texture features associated with the ROI, and geometric descriptors of a surface of the ROI.

6. The method of claim 5, wherein texture features are calculated from a Gray Level Co-occurrence Matrix of the ROI.

7. The method of claim 2, wherein functional features include a standard uptake value, an apparent diffusion coefficient, a permeability surface area product per unit volume of tissue, a fraction of plasma per unit volume of tissue, and a rate constant of an efflux of contrast media from extracellular space back to plasma.

8. The method of claim 2, wherein imaging modalities that capture anatomical information include magnetic resonance imaging (MRI), computed tomography (CT), ultrasound (US), optical coherence tomography (OCT), imaging modalities that capture functional information include positron emission tomography (PET), single photon emission computed tomography (SPECT), dynamic contrast enhanced MRI (DCE-MRI), and diffusion weighted MR1 (DW-MR1), and a fused anatomical/functional modality that captures anatomical and functional information includes any combination of a modality that captures anatomical information and a modality that captures functional information.

9. The method of claim 1, wherein the classifier is trained using an image-based boosting ridge regression method.

10. The method of claim 3, wherein said classifier predicts a local staging score for each segment or ROI of the organ and a global staging score by combining local staging scores for each segment or ROI.

11. The method of claim 1, wherein the organ is a prostate gland, image features are extracted from a fused MR/PET image, features from said blood test data include a prostate specific antigen (PSA) level, a PSA velocity, a PSA density, and a prostate health index, and the predicted cancer staging score is a Gleason score.

12. The method of claim 1, wherein the organ is a lung, image features are extracted from a fused MR/PET image, and the predicted cancer staging score is a TNM score.

13. The method of claim 1, wherein the organ is a breast, image features are extracted from a fused MR/PET image, and the predicted cancer staging score is one of a Nottingham Score or a Scarf-Bloom-Richardson Score.

14. A computer-implemented method of predicting a cancer staging score from medical image data, the method implemented by the computer comprising the steps of:

receiving patient data for a patient, wherein said patient data for the patient includes one or more of an image volume of a suspected tumor in an organ, blood test data, and demographic data;

extracting features from said patient data, wherein features from said demographic data include a patient'"'"'s weight, height, age, body mass index, body surface area, organ size, any past history of cancer occurrence, and any health index test results; and

providing said features to a classifier, wherein said classifier predicts a cancer staging score that is indicative of whether the patient'"'"'s organ is cancerous.

15. The method of claim 14, wherein the image volume comprises one of an anatomical image volume, a functional image volume, or a fused anatomical/functional image volume, and features extracted from said image volume include one or more of anatomical and functional features from said organ.

16. The method of claim 15, wherein imaging modalities that capture anatomical information include magnetic resonance imaging (MRI), computed tomography (CT), ultrasound (US), optical coherence tomography (OCT), imaging modalities that capture functional information include positron emission tomography (PET), single photon emission computed tomography (SPECT), dynamic contrast enhanced MRI (DCE-MRI), and diffusion weighted MRI (DW-MRI), and a fused anatomical/functional modality that captures anatomical and functional information includes any combination of a modality that captures anatomical information and a modality that captures functional information.

17. The method of claim 15, wherein extracting anatomical and functional features from said image volume comprises:

segmenting the organ in said anatomical and functional images to define one or more regions of interest (ROIs); and

extracting anatomical and functional features from the one or more regions of interest,

wherein a region of interest is one of the whole organ or a subregion of the organ.

18. The method of claim 17, wherein said classifier predicts a local staging score for each ROI of the organ and a global staging score by combining local staging scores for each ROI.

19. The method of claim 14, wherein the classifier is trained using one or more of features extracted from a plurality of image volumes of said organ, features extracted from a plurality of patient blood test results, features extracted from patient demographic data, and a plurality of ground truth staging scores from a plurality of organ biopsies.

20. The method of claim 19, wherein the classifier is trained using an image-based boosting ridge regression method.

21. The method of claim 19, wherein each of the plurality of image volume comprises one or more of an anatomical image volume, a functional image volume, or a fused anatomical/functional image volume, and features extracted from said image volume include one or more of anatomical features and functional features from said organ.

22. The method of claim 14, wherein the organ is a prostate gland, image features are extracted from a fused MR/PET image, features from said blood test data include a prostate specific antigen (PSA) level, a PSA velocity, a PSA density, and a prostate health index, and the predicted cancer staging score is a Gleason score.

23. The method of claim 14, wherein the organ is a lung, image features are extracted from a fused MR/PET image, and the predicted cancer staging score is a TNM score.

24. The method of claim 14, wherein the organ is a breast, image features are extracted from a fused MR/PET image, and the predicted cancer staging score is one of a Nottingham Score or a Scarf-Bloom-Richardson Score.

25. A non-transitory program storage device readable by a computer, tangibly embodying a program of instructions executed by the computer to perform operations for predicting a cancer staging score from medical image data, the operations comprising:

receiving patient data for a plurality of patients, wherein patient data for each of said plurality of patients includes one or more of an image volume of a suspected tumor in an organ, blood test data, demographic data, and ground truth tumor staging scores for the suspected tumor in said organ;

extracting features from said patient data, wherein features from said demographic data include a patient'"'"'s weight, height, age, body mass index, body surface area, organ size, any past history of cancer occurrence, and any health index test results; and

using the features extracted from said patient data to train a classifier to predict a cancer staging score for a new patient from one or more of an image volume of a suspected tumor in said organ, patient blood test data and patient demographic data of that new patient.

26. The computer readable program storage device of claim 25, wherein the image volume comprises one or more of an anatomical image volume, a functional image volume, or a fused anatomical/functional image volume, and features extracted from said image volume include one or more of anatomical features and functional features from said organ.

27. The computer readable program storage device of claim 26, wherein extracting anatomical and functional features of said organ from said image volume comprises:

segmenting the organ in each of the anatomical and functional images to define one or more regions of interest (ROIs); and

extracting anatomical and functional features from the one or more regions of interest.

28. The computer readable program storage device of claim 27, wherein a region of interest (ROI) is one of the whole organ or a segment of the organ.

29. The computer readable program storage device of claim 26, wherein anatomical features include a volume of an ROI, a surface area of the ROI, a diameter of the ROI, texture features associated with the ROI, and geometric descriptors of a surface of the ROI.

30. The computer readable program storage device of claim 29, wherein texture features are calculated from a Gray Level Co-occurrence Matrix of the ROI.

31. The computer readable program storage device of claim 26, wherein functional features include a standard uptake value, an apparent diffusion coefficient, a permeability surface area product per unit volume of tissue, a fraction of plasma per unit volume of tissue, and a rate constant of an efflux of contrast media from extracellular space back to plasma.

32. The computer readable program storage device of claim 26, wherein imaging modalities that capture anatomical information include magnetic resonance imaging (MRI), computed tomography (CT), ultrasound (US), optical coherence tomography (OCT), imaging modalities that capture functional information include positron emission tomography (PET), single photon emission computed tomography (SPECT), dynamic contrast enhanced MRI (DCE-MRI), and diffusion weighted MRI (DW-MRI), and a fused anatomical/functional modality that captures anatomical and functional information includes any combination of a modality that captures anatomical information and a modality that captures functional information.

33. The computer readable program storage device of claim 25, wherein the classifier is trained using an image-based boosting ridge regression method.

34. The computer readable program storage device of claim 27, wherein said classifier predicts a local staging score for each segment or ROI of the organ and a global staging score by combining local staging scores for each segment or ROI.

35. The computer readable program storage device of claim 25, wherein the organ is a prostate gland, image features arc extracted from a fused MR/PET image, features from said blood test data include a prostate specific antigen (PSA) level, a PSA velocity, a PSA density, and a prostate health index, and the predicted cancer staging score is a Gleason score.

36. The computer readable program storage device of claim 25, wherein the organ is a lung, image features are extracted from a fused MR/PET image, and the predicted cancer staging score is a TNM score.

37. The computer readable program storage device of claim 25, wherein the organ is a breast, image features arc extracted from a fused MR/PET image, and the predicted cancer staging score is one of a Nottingham Score or a Scarf-Bloom-Richardson Score.