CN103930030B - The area of computer aided bone scanning evaluation of the quantization assessment with automation lesion detection and bone disease load variations - Google Patents

The area of computer aided bone scanning evaluation of the quantization assessment with automation lesion detection and bone disease load variations Download PDF

Info

Publication number
CN103930030B
CN103930030B CN201280051415.0A CN201280051415A CN103930030B CN 103930030 B CN103930030 B CN 103930030B CN 201280051415 A CN201280051415 A CN 201280051415A CN 103930030 B CN103930030 B CN 103930030B
Authority
CN
China
Prior art keywords
image
lesion
patient
bone
pixel
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN201280051415.0A
Other languages
Chinese (zh)
Other versions
CN103930030A (en
Inventor
马修·谢尔曼·布朗
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
MEDQIA
Original Assignee
MEDQIA
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by MEDQIA filed Critical MEDQIA
Publication of CN103930030A publication Critical patent/CN103930030A/en
Application granted granted Critical
Publication of CN103930030B publication Critical patent/CN103930030B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/0002Inspection of images, e.g. flaw detection
    • G06T7/0012Biomedical image inspection
    • G06T7/0014Biomedical image inspection using an image reference approach
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/10Image acquisition modality
    • G06T2207/10116X-ray image
    • G06T2207/10128Scintigraphy
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30004Biomedical image processing
    • G06T2207/30008Bone
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30004Biomedical image processing
    • G06T2207/30096Tumor; Lesion

Abstract

A kind of computer assisted bone scanning assessment system and method provide the quantization assessment of automation lesion detection and bone disease load variations.

Description

The computer of the quantization assessment with automation lesion detection and bone disease load variations Auxiliary bone scanning evaluation
Prioity claim
This application claims the U.S. Provisional Patent Application No.61/548,498 submitted on October 18th, 2011 and entitled COMPUTER-AIDED BONE SCAN ASSESSMENT WITH AUTOMATED LESION DETECTION AND QUANTITATIVE ASSESSMENT OF BONE DISEASE BURDEN CHANGES(With automation lesion detection and bone The area of computer aided bone scanning evaluation of the quantization assessment of disease burden change)Rights and interests.This application claims in October 16 in 2012 U.S. Provisional Patent Application No.61/714,318 and entitled COMPUTER-AIDED BONE SCAN that day submits ASSESSMENT(Area of computer aided bone scanning is evaluated)Rights and interests.
Merged by quoting
Pass through to quote the U.S. Provisional Patent Application No.61/ that will be submitted on October 18th, 2011 for all purposes The 548,498 and U.S. Provisional Patent Application No.61/714,318 that is submitted on October 16th, 2012 integrally integrates with the application.
Technical field
The present invention relates to medical imaging field.The present invention relates more specifically to the evaluation of bone scanning, osseous lesion and bone disease.
Background technology
Bone tumour may originate from bone, or they may originate from other sites and spread(Transfer)Into skeleton. For example, the secondary tumors in bone results from the prostate cancer of transfer often.Image from bone scanning shows and primary The related lesion of osteocarcinoma or metastatic carcinoma, and explanation to the image from bone scanning is widely used in the diagnosis of disease and controls In treatment.
It has been reported that a few is used for the area of computer aided lesion detecting system of bone scanning.These technologies include half certainly Dynamic Image Segmentation program, the semi-automatic Image Segmentation program takes too long often when being used in clinical setting(For example Those semi-automatic Image Segmentation programs of Erdi et al. and Yin et al.).The semi-automatic method person of needing to use of Erdi et al. descriptions Seed point is inserted into each transport zone on image, it is contemplated that the patient with Bone tumour often has multiple disease positions Point, therefore this is the program of non-trivial.Referring to Erdi YE, Humm JL, Imbriaco M, Yeung H, Larson SM are in J The article Quantitative bone metastases delivered on 38 phases page 1401-1406 in 1997 of Nucl Med analysis based on image segmentation(Quantization Bone tumour analysis based on Image Segmentation).Referring further to Yin The article A computer- that TK, Chiu NT are delivered on 23 phases 639-654 in 2004 of IEEE Trans Med Imaging aided diagnosis for locating abnormalities in bone scintigraphy by a fuzzy system with a three-step minimization approach(It is logical using three step minimization methods in bone scintigraphy Fuzzy system is crossed for positioning abnormal computer-aided diagnosis).
Recently, the full-automatic method developed by Sadik et al. makes the osseous lesion carried out by Image Segmentation detect and pass through The scanning evaluation that artificial neural network is carried out is combined, and they are classified with the possibility that Bone tumour occurs according to patient, So as to the binary scanning grading for being possible to " Bone tumour " or may be " without Bone tumour ".Referring to Sadik M, Jakobsson D, 2006 year 27 phases of Olofsson F, Ohlsson M, Suurkula M, the Edenbrandt L. in Nucl Med Commun The article A new computer-based decision-support system for the delivered on page 417-423 interpretation of bone scans(A kind of new computer based decision support system for explaining bone scanning System).
Although the system shows the good correlation with the assessment of the Bone tumour possibility determined by doctor, the system Without offer to continuously scanning the quantisation metric for being contrasted(metric), the instrument of evaluation treatment results is not provided yet.
Importantly, also not carrying out being related to patient to benefit to the result reported(For example the reduction of skeleton dependent event or The extension of life)True measurement, constitute supervision department license basic measurement in terms of research.
Conversely, the system for Imaging enhanced has been developed, so that the image standardization from continuous scanning, in order to Doctor explains, but the system does not attempt also to recognize lesion.Referring to Jeong CB, Kim KG, Kim TS, Kim SK exist The article Comparison of image delivered on 24 phases page 424-436 in 2011 of J Digit Imaging enhancement methods for the effective diagnosis in successive whole-body bone scans(For the contrast of the Imaging enhanced method of the efficient diagnosis in Whole body bone scan).
Before, being flashed by the bone to the transfer bone disease of load in prostate cancer has carried out quantization assessment, wherein Exploitation comprising measurement, such as bone scanning index(BSI)With the percentage of the positive area in bone scanning(%PABS).Referring to Imbriaco M, Larson SM, Yeung HW, Mawlawi OR, Erdi Y, Venkatraman ES et al. is in Clin The article A new parameter for measuring delivered on 4 phases page 1765-1772 in 1998 of Cancer Res metastatic bone involvement by prostate cancer:the Bone Scan Index(For measure by The new parameter of the transfer bone that prostate cancer is involved:Bone scanning index).Referring further to Noguchi M, Kikuchi H, Ishibashi The article Percentage of that M, Noda S. are delivered on 88 phases page 195-201 in 2003 of Br J Cancer positive area of bone metastasis is an independent predictor of disease death in advanced prostate cancer(The percentage of the positive area of Bone tumour is to advanced prostate cancer disease death Independentpredictor).
BSI and %PABS has gone through initial evaluation as the Prognostic Factors of the patient for suffering from prostate cancer, but is used for The method for calculating these measurements spends the too long of time, since it is desired that carrying out substantial amounts of manual annotation to bone scanning.As evaluation The feasible measurement of therapeutic response, in being continuously developed to the evaluation of %PABS and BSI.Referring to Yahara J, Noguchi M, The article Quantitative evaluation of that Noda S. are delivered on 92 phases page 379-384 in 2003 of BJU Int bone metastases in patients with advanced prostate cancer during systemic treatment(To the quantitatively evaluating of Bone tumour of the patient with advanced prostate cancer during systematic treating).Referring further to Morris MJ, Jia X, Larson SM, Kelly A, Mezheritzky I, Stephenson RD et al. was in 2008 The Post-treatment serial bone scan index delivered on Genitourinary Cancers Symposium (BSI)as an outcome measure predicting survival(Serial bone scanning index after treatment(BSI)As Predict the measurement result of survival).
Although computer aided detection before(CAD)System has been applied in bone scanning analysis, but they lack this hair Feature in bright implementation method.For example, single sweep operation of these known systems typically only to patient carries out lesion detection, and Continuous scanning is not contrasted.
The content of the invention
There is a kind of offer the bone scanning evaluation of automation lesion detection and the quantization assessment of bone disease load variations to be System and method.
In various embodiments, therapeutic effect is monitored using bone scanning make use of is carried out in single sweep operation to lesion Accurate cutting and quantization, and the lesion measured value between continuous scanning contrast.Implementation method provides a kind of accurate And reproducibly cut and quantify automated system of the osseous lesion to aid in doctor to be contrasted and patient between in patient in itself.
The present inventor is to combining automation lesion cutting(Comprising image standardization)With the quantization assessment of disease burden Bone scanning computer assisted therapy assessment system has carried out analysis verification.Success point to untreated patient and the patient for the treatment of Group, is used to the ability that evaluation system evaluates therapeutic effect.
Checking display, the system can reduce the otherness of the bone scanning analysis of manual annotations, be obtained so as to self-consistentency Objective, the reproducible and measured value that quantifies so that for pre- between individual bulk measurement and other clinical and laboratory result data Phase association gives basis.
Checking display, the system can accurately be automated bone scanning lesion segmentation(The detection of lesion pixel)And energy The quantization measurement of lesion load is enough provided, the quantization measurement of the lesion load and then can be used for evaluation through treatment and without controlling The change of the morbid state in the patient for the treatment of.
In each implementation method, the present invention can use computer program(Software)Realize.The program can be by image Acquisition device, read work station, server and/or other appropriate devices are performed.Treatment on the server can be helped Interacted in centralized image file system, and contributed to the storage of bone scanning analysis report in the centralized database.Should System can be with(For example via internet)It is accessed remotely through computer networks.
The schematic flow sheet that embodiments of the present invention are referred to equation, algorithm and/or method is described.These sides Method can be implemented separately using instruction set or be realized as a component of system.So, the behaviour of each equation, algorithm, flow Work and/or combinations thereof can be in many ways(Such as hardware, firmware and/or software)To realize.Computer program instructions Can be loaded on computers, realize what is specified by equation, algorithm and/or flow so that computer program instructions offer is a kind of The mode of function.
Brief description of the drawings
Describe the present invention with reference to the accompanying drawings.These accompanying drawings of a part of specification are incorporated herein and formed exemplified with this hair Bright implementation method, and be further used for explaining principle of the invention together with explanation, and cause those skilled in the relevant art Can implement and use the present invention.
Fig. 1 shows that area of computer aided of the invention quantifies bone scanning assessment system.
Fig. 2 shows the general introduction of the estimated journey of area of computer aided bone scanning of the invention.
Fig. 3 shows the more detailed general introduction of the evaluation process of Fig. 2.
Fig. 4 shows the method being based on according to the method for Fig. 2 with reference to image calculating normal bone standardized value RIMEDIAN.
Fig. 5 shows the method comprising the intensity threshold that lesion is indicated using expert and identification.
Fig. 6 shows the second party comprising the intensity threshold that lesion is indicated using expert and identification of the process according to Fig. 2 Method.
Fig. 7 shows that the identification of the process according to Fig. 2 indicates the true positives pixel of lesion.
Fig. 8 shows that the evaluation anatomy of the process according to Fig. 2 is specifically measured.
Fig. 9 shows the simple measurement of all anatomical areas of evaluation of the process according to Fig. 2.
Figure 10 shows the operation scenario of the exemplary assessment system of the process according to Fig. 2.
Specific embodiment
The content being provided below describes the example of some embodiments of the present invention.Design, accompanying drawing and description are The non-limiting example of the implementation method disclosed in it.For example, the other embodiment of disclosed device and/or method can be with Comprising or can not include features described herein.Additionally, disclosed advantage and benefit can be only applicable to spy of the invention Determine implementation method, and be not used in limitation disclosed invention.
As used in this article, term " coupling " includes and is directly connected to and is indirectly connected with.Additionally, work as mentioning the first dress Put when being coupled with second device, there can be the middle device comprising active device between the first device and the second device.
Fig. 1 shows that area of computer aided of the invention quantifies bone scanning assessment system 100.Processing unit 108 is received Image data from image capturing equipment 102.It is from these images and other images, by treatment or untreated Data are chosen for the reference cell 104 of data via for storing, and can be obtained by processor.In each implementation method, place The input of the input comprising user and setting 106 of reason device, such as evaluation, desired image by expert to any patient status Those inputs and setting obtained by quality and equipment performance.In some embodiments, treatment and/or user input extremely Partially carried out in image capturing equipment.
Bone scanning imaging device 102 includes any suitable bone imaging device usually used in nuclear medicine.For example, In each implementation method, imaging device utilizes radio isotope and radiosensitive camera, such as in scintillography system and dress Put middle those radio isotopes and radiosensitive camera for using.Specifically, scintigraphy is that one kind is used in combination radioactivity Isotope and γ cameras are indicating to accumulate the technology of radioisotopic tissue.Here, during γ camera pixels intensity shows bone The regional area of radio isotope and the hypermetabolism activity of accumulation(Indicate the event of osseous lesion).
Reference device 104 includes any suitable instrument for data storage and/or data association message.Typical base Quasi- unit includes digital data storage unit, and the digital data storage unit includes semiconductor memory, move media memory (Such as hard disk drive), optical memory and known similar device and equipment in calculating field.
User input unit 106 is included to be used for any suitable work for being delivered to processor 108 of user Tool.In each implementation method, the input processing instrument of keyboard, mouse, touch-screen and correlation is used(Such as personal computer) In any one.
In some embodiments, there is provided adjustment unit or regulation station 114 1 are used to enter processor output data 109 Row enhancing and quality control.Although can such as use pattern identification technology make the function automate, in each implementation method In, by expert(Such as nuclear medicine radiologist with nuclear medicine professional knowledge)To strengthen and/or correct bone scanning image/number According to.Such as identify the situation by previously having existed(Such as joint disease and fracture)Caused wrong lesion is indicated, and accordingly Ground is solved.
Bone scanning treatment 108 includes processing equipment, method and process.Suitable equipment is included in known in calculating field Any suitable message processing device.Specifically, in each implementation method, comprising microprocessor, personal computer, work Stand, any one of Large-scale parallel computing facility and supercomputer or various digital processing devices provide what is be adapted to Processing function.
Visualization device 110 is included in known any suitable equipment in calculating field, comprising display and printer. Display includes CRT, LED, plasma scope, fluorescence display and el display device.Printer is included will letter Breath is fixed on the device in tangible medium, such as laser printer and the device with similar purpose.
Graph visualization aid 112 is used to visualize physical arrangement, and particularly for sweeping the bone of instruction lesion Retouch data visualization.Visualization aid comprising synthetic image special case, the synthetic image using show base image or The form of the image set that basic is covered by upper layer image.The translucent and/or transparent quality of upper layer image is caused can be same When observation basic at least part of and upper layer image it is at least part of.
As it is shown in the figures, processor output end 109 aids in information conveyance to visualization device 110, graph visualization One or more in instrument 112 and adjustment unit 114.In each implementation method, visualization device and aid are caused Unregulated processor output 111 and adjusted processor output one or both of 115 are visualized.
Following accompanying drawing is described in more detail the side performed in area of computer aided quantifies bone scanning assessment system 100 Method and program, are included in the methods and procedures performed in processing unit 108.
Fig. 2 shows the general introduction 200 of area of computer aided bone scanning assessment procedure of the invention.Initialization step 202 Enable standardized test image step 204 and the evaluation to testing image, to the evaluation of testing image comprising identification lesion and Generating quantification measurement 206.As illustrated, follow-up testing image evaluation need not typically repeat initialization step.
In initialization step 202, one or more reference images are obtained.Typically, from for primary bone cancer or turn Shifting property osteocarcinoma indicates to obtain multiple reference images in multiple patients that there is the positive to indicate.Selected from various patient populations Select one group of reference images and be favorably improved following possibility:Benchmark norm(reference norm)To indicate subsequent right Than the osseous lesion in testing image, without being the failure to indicate the osseous lesion in subsequent contrast test image.
Initialization step 202 includes the normalization factor for determining the normal bone strength of instruction.Here intensity refers to visual Luminous intensity, such as intensity of pixel in the image obtained by γ cameras.Initialization is also included:Determined using reference images Indicate the intensity threshold of osseous lesion.In each implementation method, normalization factor and intensity that storage is determined by initialization step Threshold value, in case using in the future.In some embodiments, these values are stored in reference cell 104.
Complete after initialization step, be then standardized step 204 and evaluation procedure 206.Walked in standardization In rapid, testing image is obtained by image capturing equipment 102 or miscellaneous equipment(Such as represent the data of image), and according to it is following enter The methodological standardization image of one step description.
Standardization 204 has prepared the testing image for evaluating, and the evaluation is recognized and measurement generation 206 comprising lesion. In each implementation method, for the repeatability for improving lesion segmentation and quantify, standardization is reduced due to showing in build, radioactivity The dosage level of track agent and/or the temporal difference between tracer is administered and scanning is obtained and the intensity variation that produces Influence.After intensities normalised, the image pixel intensities of normal bone are consistent between time point such that it is able in serial patient Reproducible lesion segmentation and quantization assessment are obtained in image.
Measurement from evaluation procedure 206 provides the quantization measurement to lesion load.Such as the institute in deciding step 208 Show, extra testing image can be standardized and be evaluated, without repeating initialization step 202.Having processed After some testing images, end step 210 is reached.
In one embodiment, it is that a specific patient makes multiple testing images and these testing images are carried out Treatment.Each testing image is each provided with the quantization measurement to the lesion load of the patient, is so made in different time Whether whether whether testing image is monitored there is provided patient health, wherein having reaction to treatment comprising disease, stablizing or in development.
Fig. 3 shows the more detailed general introduction 300 of the implementation method of area of computer aided bone scanning evaluation of the invention.It is such as preceding It is described, it is standardized step 204 and evaluation procedure 206 after initialization step 202.
Initialization step 202 is included:Reference images are obtained, the standardized value of normal bone is determined using reference images (“RIMEDIAN”)400, and determine the specific intensity threshold of anatomical area using reference images(“ITr”)500、600.Such as It is upper described, more than 302 reference images are obtained typically from multiple osteocarcinoma patients.
Normalization step 204 is included:Testing image 304 is obtained, testing image normal bone intensity level is determined(“TI75”) 314, and 324 are standardized to testing image image pixel intensities.
As in equationi, calculated according to testing image normal bone strength TI75 and reference images standardized value RIMEDIAN Normalization factor NF.
Equation 1, normalization factor:NF=(RIMEDIAN/TI75)
Normalization factor NF is used to be standardized the image pixel intensities in testing image.As shown in equation 2, to test shadow As pixel is standardized, wherein TIPIiIt is the testing image image pixel intensities of specific pixel, TIPINiIt is the standardization of the pixel Testing image image pixel intensities.
Equation 2:TIPINi=TIPIi x NF
After the standardization, the normal bone strength in reference images is consistent with the normal bone strength in testing image.
Standardization 204 has prepared testing image to evaluate 206.Evaluation procedure is included:Lesion 700 is indicated, anatomy is evaluated Region specifically measures 800, and evaluation collects measurement 900 for all anatomical areas, and each step will be entered below Row is further described.
These measurements provide the quantization measurement to lesion load.As shown in deciding step 338, can be to extra Testing image is standardized and evaluates, without repeating initialization step 202.Processed all of testing image it Afterwards, end step 348 is reached.
Fig. 4 shows the method 400 that normal bone standardized value RIMEDIAN is calculated according to reference images.School district is dissected in identification Domain 411, anatomy segmentation 413 is carried out to benchmark bone image set, and normal bone intensity level 415 is recognized in each image, and really Surely the benchmark bone strength value 417 of all reference images is represented.
In step 411, anatomical area is recognized.These regions typically represent backbone region.In embodiments, dissect Credit cuts by the way that image collection of illustrative plates is contrasted with following anatomical landmarks to recognize anatomical area:Backbone, rib, head, Four limbs and pelvis.
In step 413, automatic segmentation benchmark bone image set.Here segmentation corresponds to above-identified dissection school district Domain.
In step 415, normal bone intensity level is recognized in each reference images.In each implementation method, each The specific region selection of reference images has the intensity that statistics is worth to represent normal bone strength.The statistics assignment for being used can To be the experience that is based on, evaluated based on trial-erroneous procedures, or it is true with another way well known by persons skilled in the art Fixed.
In the exemplary cases based on inventor's experience, normal bone strength by particular anatomical region intensity histograms 75% value RI75x(The quantity in 1≤x≤region)Indicate.In embodiments, selected by from epiphysis area intensity histograms 75% value RI75xTo determine the normal bone strength in reference images.
In step 417, it is determined that representing the benchmark bone strength value of all reference images.In each implementation method, the generation Table bone strength value is and 75% value RI75 above-mentionedxThe corresponding intermediate value RIMEDIAN of set.
Fig. 5 and Fig. 6 show the method 500,600 for determining the specific intensity threshold of anatomy from reference images. Fig. 5 shows the method 500 comprising the intensity threshold that lesion is indicated using expert and identification.First step 511 provides expert, For example understanding nucleus medical image and the expert especially in the field of the bone scanning image of the patient with osseous lesion.In the step In rapid, the lesion in Expert Location reference images.In second step 513, expert is indicated to evaluate.The evaluation determines to refer to Show the intensity threshold of lesion.
Fig. 6 shows the second method 600 comprising the intensity threshold that lesion is indicated using expert and identification.Step includes note Release 611, classification 613 and determine intensity threshold ITr
Annotation 611 make use of expert, such as expert mentioned above.Herein, expert is carried out to each reference images Annotation, to indicate lesion.613 pairs of expert's signs of classification are classified, so that pixel is associated with lesion.In embodiments, Using binary classifier system, so as to the expert's sign that will indicate lesion is categorized as true positives pixel, and other bone pixels are divided Class is true negative pixel.
For each anatomical area, intensity threshold determines the single intensity threshold IT of 615 determinationsr, help to reappear The classification of all patients in for benchmark group.
For example, for each anatomical area, all patients in finding intensity threshold for group make true positives Number is maximized(Increased average sensitivity), while minimizing the number of false positive(Increased average specificity).
In each implementation method, lesion is carried out to standardization image via anatomical area specific intensity threshold Segmentation, with by will specific threshold application to standardize image on then carry out be connected element filter detect each dissect Lesion in the domain of school district.
In addition, in each implementation method, using receiver operating characteristic curve(ROC or ROC curve)It is above-mentioned to evaluate The performance of binary classifier system.TPF in by marking the positive at each threshold value setting(TPR=true positives ratio Rate)False positive part in contrast feminine gender(FPR=false positive rates)To create curve/criterion.TPR is also known as susceptibility, And FPR is 1 to subtract specific or Kidney-Yin sex rate.
Change discrimination threshold(Here it is intensity threshold(ITr))To determine ITrValue, the ITrValue helps to optimize average sensitivity Degree and average specificity.For example, by true positives lesion pixel labeled as the function of false positive lesion pixel ROC curve by typical case Ground has the optimal IT of instructionrThe peculiar change of the slope of value.
Fig. 7 shows that identification indicates the true positives pixel 700 of lesion.As described above, identification anatomical area 711.Make every Image pixel intensities and corresponding intensity threshold IT in one testing image anatomical arearMatch, carry out following test 713.
Equation 5, the instruction of lesion:TIPINr,i>ITr
Obviously, the equation is by the standardized test image pixel intensity in particular anatomical region and is derived from reference images The specific intensity threshold IT of anatomyrCompare.When testing image image pixel intensities are more than corresponding intensity threshold, the pixel is Indicate the true positives pixel that there is lesion.
Fig. 8 shows evaluation anatomy specifically measurement 800.Specifically, for each anatomical area, to kidney-Yang The number Z of property pixelrCarry out counting 813.Additionally, in subsequent step 815, the intensity summation to all true positives pixels SUMIr.These steps 817 are repeated for each anatomical area.
Fig. 9 shows that evaluation collects measurement 900.Collect measurement comprising all anatomical areas.
As shown in equation 6, evaluation collects osteopathy variable area 911, wherein PARepresent 1 area of pixel.
Equation 6, collects bone scanning lesion area(BSLA)
As shown in equation 7 below, evaluation collects osteopathy intensity adjustable 913.
Equation 7, collects bone scanning lesion intensity(SBLI):
Therefore, the quantization that osteopathy variable area represents the size and number of active region in bone scanning is collected, and bone scanning Lesion intensity represents the level of bon e formation activity.
In each implementation method, evaluation osseous lesion counts 915.In embodiments, will using lesion area identifier Lesion is identified as comprising at least 5 regions of the separation of contiguous pixels, and each pixel is above the intensity threshold for determining.At each In implementation method, the size of lesion identifier and contiguous pixels group not only considers the feature for being large enough to merit attention, but also Consider whether to there is a possibility that as follows:The group of selected size will simultaneously be subject to most common failure(Such as defective scanning Instrument camera pixel)Influence.
Equation 8, collects the counting of poroma lesion(BSLC):
Higher than at least 5 numbers of the separated region of contiguous pixels of the intensity threshold for determining.
Reaction for quantization assessment patient to treating, can calculate from given patient's during evaluation is reacted The change of the lesion load measurement between serial bone scanning.The percentage that can be measured using lesion load changes to evaluate treatment Progress and/or reaction, and there is separation in the percentage change for describing each reaction classification.For example, bone scanning image is sick Variable area may be considered development when increasing by 30% or more, and may be considered reaction when reducing 30% or more.
In operation, thus it is possible to vary step in above-mentioned evaluation is being suitable for image, the data from image and benchmark shadow As the availability of process step.For example, after testing image is standardized, can determine to dissect school district from reference images The specific intensity threshold in domain.In another embodiment, multiple reference images collection can be processed, and by corresponding RIMEDIAN and TIr values are for one or more testing images.Similarly, those of ordinary skill in the art should recognize from the present disclosure that with above The order different of the image processing step of description is in some cases appropriate, for example, wherein there is search best base The situation of quasi- image set.Therefore, it can change said system operation be suitable for it is specific the need for and limit.
Figure 10 is exemplified with exemplary assessment system operation scenario 1000.Generally, at using the data from reference images Reason input image, to produce the quantization assessment to bone disease load.To a certain extent, there is the patient's shadow in treatment interval Picture, the disease to the reaction, the disease of development or stabilization treated is indicated in the change of the quantization mark of disease.
As illustrated, patient's baseline and the image 1020 of the 6th week can be obtained.Evaluation process is provided in the way of being substantially the same Baseline and the 6th week quantization disease marker of image.
Initially, anatomy segmentation is carried out to split input image 1002.During Image Segmentation, image is divided into solution Cut open school district domain.Anatomical area selected in the example is backbone, rib, head, four limbs and pelvis.The dissection of input image Credit is cut there is provided the segmentation image similar to the segmentation image 1022 for illustrating.
Input image to splitting carries out image intensity standardization 1004.The standardization is produced similar to illustration standardization base Line and the 6th week image of image 1024.
Lesion segmentation or identification 1006 are carried out after image standardization 1004.Herein, by each of input image Intensities normalised pixel in region is contrasted with the intensity threshold of the corresponding region from reference images as explained above. Area-specific strength threshold value from reference images indicates the lesion in input image.
Optional user checks and edits(With obtain user accreditation lesion segmentation 1008 and figure 1 illustrates Regulation 114)So that manual adjustment can be carried out to evaluation.For example, can solve to be caused due to joint disease and knochenbruch at this moment False positive.Checked with and without optional user and the lesion segmentation of editor is produced similar to the baseline and the 6th for illustrating All lesions indicate the image of image 1026.
Checked and editor 1008 in lesion segmentation 1006 and user(If any)Afterwards, the meter of lesion load is carried out Calculate 1010.In this step, the lesion of the measured value, lesion intensity and specific region and/or all regions of lesion area is determined Count.In embodiments, the lesion area to all regions is sued for peace, and lesion intensity to all regions is asked With.In each implementation method, lesion is counted(The lesion in such as all regions count and)With lesion area and lesion intensity Aggregate value together provides the instrument for quantifying lesion load.
In each implementation method, patient reaction's evaluation 1012 includes patient reaction's classification report, patient reaction classification Report shows for example to react or develop or stablize.In embodiments, chart 1028 provides baseline and measurement in the 6th week is swept in bone Retouch the quantization contrast in lesion counting, bone scanning lesion area and bone scanning lesion intensity.
In some embodiments, image contrast make use of base image with translucent coverage diagram.In an implementation method In, there is the area image of coloring or non-staining automatic segmentation, as in bone scanning image(It is original or standardized) On translucent coverage diagram.So it is particularly useful as adjusting as needed and/or editing operator's nondominant hand of image 114 Section.
In one embodiment, base image(Such as Baseline Images 1026 with lesion segmentation)With the 6th week image(Or Similar image from another treatment interval)Translucent coverage diagram be used together.Herein, it is possible to use color or coloring Strengthening " before and after " visualization difference between situation.It should be evident that these image contrasts are provided is considered as fixed The information of " directly perceived " information of property.To a certain extent, the details being stored in raw video is noted, they also embody bone disease The quantization measurement of sick load variations.
Although each implementation method of the invention has been described above, it should be understood that they are only with for example and not limitation Mode be suggested.As those skilled in the art it will be evident that can without departing from the spirit and scope of the present invention in the case of Under, can in form and details carry out various change.So, width of the invention and scope should not be restricted by above-mentioned example reality The limitation of mode is applied, but should be limited according to claims below and its equivalent.

Claims (15)

1. a kind of for being quantified to osteopathy varying duty to determine the device of patient reaction, the device includes:
Radioactive tracer formulation scanner, processor and store the digital data memory of anatomy collection of illustrative plates;
Patient bone scan-image, it is obtained by the radioactive tracer formulation scanner and is formed by pixel, the patient bone Scan-image has image intensity;
The processor is using the anatomy collection of illustrative plates and provides the dissection for the patient bone scan-image be based on atlas Credit is cut, to recognize the anatomical area collection on the patient bone scan-image;
Intensity to the patient bone scan-image is standardized, to cause the normal bone in the patient bone scan-image Intensity it is consistent with the intensity of the normal bone in one or more benchmark bone scanning images;
The intensity of the pixel in the region based on the patient bone scan-image is swept with from one or more of benchmark bones Difference between the area-specific strength threshold value of shading picture detects the osteopathy in the region of the patient bone scan-image Become;And,
Quantify osteopathy varying duty using the feature of set of pixels corresponding with the osseous lesion for detecting.
2. device according to claim 1, further includes:
For the patient bone scan-image, counted to determine at least one amount according to a group lesion area, lesion intensity and lesion Change osteopathy varying duty mark;
For patient bone scan-image treated before, corresponding quantization osteopathy varying duty mark is determined;And
Disease person's development is determined based on the contrast of the quantization osteopathy varying duty mark.
3. device according to claim 2, wherein, the quantization osteopathy varying duty mark is directed to all regions to be added up 's.
4. device according to claim 1, further comprising the steps:
For the patient bone scan-image, counted to determine at least two amounts according to a group lesion area, lesion intensity and lesion Change osteopathy varying duty mark;
For patient bone scan-image treated before, determine that corresponding at least two quantify osteopathy varying duty mark;And
Disease person's development is determined based on the contrast of the quantization osteopathy varying duty mark.
5. device according to claim 1, further comprising the steps:
For the patient bone scan-image, counted to determine multiple quantization osteopathy according to lesion area, lesion intensity and lesion Varying duty is marked;
For patient bone scan-image treated before, determine that corresponding multiple osteopathy varying duties that quantify are marked;And,
Disease person's development is determined based on the contrast of the quantization osteopathy varying duty mark.
6. device according to claim 1, further comprising the steps:
Selection is standardized and the processed image based on patient's Baseline Images of lesions showed;
Selection be standardized and the processed patient using lesions showed time after a while image as translucent covering image;And
Will be superimposed basis and covering image be presented to unprofessional person, as explain given patient lesion load have occurred that why The means of the change of sample.
7. device according to claim 1, further comprising the steps:
Image based on selection patient's raw video or patient's standardization image;
The translucent coverage diagram of the lesion that selection is detected is used as covering image;And,
The image of superposition is presented, as making the degree of the lesion load of given patient and the visual means of distribution.
8. device according to claim 7, wherein, the covering image is and base image image of the same period.
9. device according to claim 2, further comprising the steps:
Normal bone standardized value is determined from multiple reference scans, the normal bone standardized value is not that region is specific;
Normal bone intensity level is determined from the patient bone scan-image;And
Using the normal bone standardized value of the reference scan and the normal bone strength of the patient bone scan-image to implement State normalization step.
10. device according to claim 9, further comprising the steps:
Multiple reference scans are annotated to indicate lesion;
It is true positives pixel or true negative pixel by annotation category;And
Determine that one helps to reappear the intensity threshold of the classification for each anatomical area.
A kind of 11. devices for processing patient bone scan-image and quantify osteopathy varying duty, described device includes:
Using the anatomical structure of scanner scanning bone, to produce patient bone scan-image, the patient bone scan-image is by picture Element is formed, and the pixel has the intensity for indicating Bone m etabolism rate;
Using anatomy collection of illustrative plates, the patient bone scan-image anatomy is divided into multiple regions;
Using from the normal bone label in benchmark bone scanning image set and from the normal bone in the patient bone scan-image Mark, is standardized come the intensity to patient bone scan-image pixel;
Detect described by the intensity of patient bone scan-image pixel and from the area-specific strength threshold value of reference images Lesion pixel in patient bone scan-image;And
If it there is lesion pixel in the patient bone scan-image, feature according to the lesion pixel quantifies osteopathy Varying duty.
12. devices according to claim 11, it is further comprising the steps:
For the patient bone scan-image, it is multiplied by elemental area according to the number of the lesion pixel in all regions to determine Bone scanning lesion area;
For patient bone scan-image treated before, corresponding quantization osteopathy varying duty mark is determined;And
Disease person's development is determined based on the contrast of the quantization osteopathy varying duty mark.
13. devices according to claim 12, it is further comprising the steps:
For the patient bone scan-image, determine what is amounted to according to the summation of the intensity of the lesion pixel in all regions Intensity;
For the patient bone scan-image, the lesion picture for amounting to is determined according to the summation of the lesion pixel in all regions Element;
The bone scanning lesion intensity is determined divided by the lesion pixel of the total according to the intensity of the total;
For patient bone scan-image treated before, corresponding quantization osteopathy varying duty mark is determined;And,
Disease person's development is determined based on the contrast of the quantization osteopathy varying duty mark.
14. devices according to claim 11, it is further comprising the steps:
The number k of the contiguous pixels that selection can not possibly be influenceed by most common failure simultaneously, the most common failure is swept for defective Retouch instrument camera pixel;
For the patient bone scan-image, and for all regions, it is determined that with more than the k number of the group of lesion pixel j;And
Bone scanning lesion is counted and is equal to j.
A kind of 15. devices for processing patient bone scan-image and quantify osteopathy varying duty, described device includes:
Radioactive tracer formulation scanner, processor and digital data memory;
Storage anatomy collection of illustrative plates in memory;
Normal bone label is derived from reference scan and is stored in memory, and indicates the area-specific strength of lesion Threshold set is derived and is stored in memory from the reference scan;
The scanner is operable as obtaining the patient bone scan-image formed by pixel;
The processor is operable as carrying out anatomy segmentation to the patient bone scan-image using the anatomy collection of illustrative plates;
The processor is operable as using the normal bone label come to the strong of the pixel in the patient bone scan-image Degree is standardized;
The processor is operable as detecting lesion pixel using the intensity threshold;And,
The processor is operable as calculating quantization lesion load measurement according to lesion pixel feature.
CN201280051415.0A 2011-10-18 2012-10-16 The area of computer aided bone scanning evaluation of the quantization assessment with automation lesion detection and bone disease load variations Expired - Fee Related CN103930030B (en)

Applications Claiming Priority (5)

Application Number Priority Date Filing Date Title
US201161548498P 2011-10-18 2011-10-18
US61/548,498 2011-10-18
US201261714318P 2012-10-16 2012-10-16
US61/714,318 2012-10-16
PCT/US2012/060394 WO2013059177A1 (en) 2011-10-18 2012-10-16 Computer-aided bone scan assessment with automated lesion detection and quantitative assessment of bone disease burden changes

Publications (2)

Publication Number Publication Date
CN103930030A CN103930030A (en) 2014-07-16
CN103930030B true CN103930030B (en) 2017-06-16

Family

ID=48141274

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201280051415.0A Expired - Fee Related CN103930030B (en) 2011-10-18 2012-10-16 The area of computer aided bone scanning evaluation of the quantization assessment with automation lesion detection and bone disease load variations

Country Status (2)

Country Link
CN (1) CN103930030B (en)
WO (1) WO2013059177A1 (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107274389B (en) * 2017-05-25 2021-06-01 中国科学院苏州生物医学工程技术研究所 Femur and acetabulum anatomical parameter obtaining method based on CT three-dimensional sequence image
US20190341150A1 (en) * 2018-05-01 2019-11-07 Google Llc Automated Radiographic Diagnosis Using a Mobile Device
TWI709147B (en) * 2019-10-16 2020-11-01 中國醫藥大學附設醫院 System of deep learning neural network in prostate cancer bone metastasis identification based on whole body bone scan images

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7283857B1 (en) * 1998-11-30 2007-10-16 Hologic, Inc. DICOM compliant file communication including quantitative and image data
US6246745B1 (en) * 1999-10-29 2001-06-12 Compumed, Inc. Method and apparatus for determining bone mineral density
DE10044844A1 (en) * 2000-09-11 2002-04-04 Siemens Ag Method, device and software for separating the individual objects of an anatomical structure segmented from 3D data sets of medical examination methods
JP2006512938A (en) * 2002-09-16 2006-04-20 イメージング セラピューティクス,インコーポレーテッド Imaging markers for musculoskeletal diseases
ATE484231T1 (en) * 2003-01-07 2010-10-15 Imaging Therapeutics Inc DEVICE FOR PREDICTING MUSCLE/SKELETAL DISEASES
JP5426170B2 (en) * 2005-11-11 2014-02-26 ホロジック, インコーポレイテッド Estimating the risk of future fractures using a three-dimensional bone density model

Also Published As

Publication number Publication date
CN103930030A (en) 2014-07-16
WO2013059177A1 (en) 2013-04-25

Similar Documents

Publication Publication Date Title
US9002081B2 (en) Computer-aided bone scan assessment with automated lesion detection and quantitative assessment of bone disease burden changes
Belal et al. Deep learning for segmentation of 49 selected bones in CT scans: first step in automated PET/CT-based 3D quantification of skeletal metastases
Petrick et al. Evaluation of computer‐aided detection and diagnosis systems a
Li et al. Unsupervised machine learning of radiomic features for predicting treatment response and overall survival of early stage non-small cell lung cancer patients treated with stereotactic body radiation therapy
US20220319008A1 (en) Automated tumor identification and segmentation with medical images
CN101231678B (en) Medical image-processing apparatus and medical image processing method
Nakajima et al. Bone scan index: a new biomarker of bone metastasis in patients with prostate cancer
Campadelli et al. A fully automated method for lung nodule detection from postero-anterior chest radiographs
Horikoshi et al. Computer-aided diagnosis system for bone scintigrams from Japanese patients: importance of training database
Giger et al. Computerized characterization of mammographic masses: analysis of spiculation
Van Rikxoort et al. A method for the automatic quantification of the completeness of pulmonary fissures: evaluation in a database of subjects with severe emphysema
CN113710159A (en) System and method for automated and interactive analysis of bone scan images to detect metastasis
Brown et al. Computer-aided quantitative bone scan assessment of prostate cancer treatment response
Tokuda et al. Investigation of computer-aided diagnosis system for bone scans: a retrospective analysis in 406 patients
Pu et al. Computerized assessment of pulmonary fissure integrity using high resolution CT
Takahashi et al. Assessment of bone scans in advanced prostate carcinoma using fully automated and semi-automated bone scan index methods
Aslantas et al. CADBOSS: A computer-aided diagnosis system for whole-body bone scintigraphy scans
Liu et al. Deep learning for the automatic diagnosis and analysis of bone metastasis on bone scintigrams
CN103930030B (en) The area of computer aided bone scanning evaluation of the quantization assessment with automation lesion detection and bone disease load variations
Gc et al. Variability measurement for breast cancer classification of mammographic masses
Xu et al. Severity assessment of COVID-19 using a CT-based radiomics model
Nichols et al. Texture analysis for automated evaluation of Jaszczak phantom SPECT system tests
Lindgren Belal et al. Automated quantification of PET/CT skeletal tumor burden in prostate cancer using artificial intelligence: The PET index
Marias et al. A mammographic image analysis method to detect and measure changes in breast density
JP2004213643A (en) Computer aided reconciliation method

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20170616

Termination date: 20211016

CF01 Termination of patent right due to non-payment of annual fee