CN109350099A

CN109350099A - A Random Event Removal Processing Method Applied in Clinical PET System

Info

Publication number: CN109350099A
Application number: CN201811066685.4A
Authority: CN
Inventors: 牛晓锋; 张勇; 叶宏伟
Original assignee: Zhongshan Mingfeng Medical Instrument Co ltd
Current assignee: Zhongshan Mingfeng Medical Instrument Co ltd
Priority date: 2018-09-13
Filing date: 2018-09-13
Publication date: 2019-02-19

Abstract

The invention discloses a random event removing and processing method applied to a clinical PET system, which comprises two quick and simple specific implementation methods, namely an optimized three-dimensional removing method and a quick two-dimensional removing method. The invention has the following advantages and effects: (1) invalid random coincidence events can be removed up to 75%. (2) The calculation time of invalid coincidence events is removed in an optimized mode, and the data processing speed is improved. (3) And reducing the background noise of the image.

Description

A kind of chance event removal processing method applied to clinical PET system

Technical field

The present invention relates to medical image processing technical fields, more specifically, are related to a kind of applied to clinical PET system Chance event removes processing method.

Background technique

Positron Emission Computed Tomography (Positron Emission Tomography, abbreviation PET) is to pass through To human injection's radioactive tracer drug, with specific cells or tissue certain bioprocess occurs for drug in human body, passes through To the gamma-ray detection of a pair that radionuclide decay generates, to obtain radiopharmaceutical in the intracorporal distribution map of people.

The positive electron that positron radionuclide in PET is generated by decay can fall into oblivion with existing negative electrons a large amount of in human body It goes out, to generate the γ photon of a pair of reversed approximate 180 degree.The line received between the detector of the two γ photons is referred to as Meet line (line of response, abbreviation LOR).If two γ photon sources being detected are in the same electron annihilation Event and wherein at least one photon is scattered with medium, this event are referred to as scattering and meet.If detect two A photon is derived from two different annihilation events, this event is referred to as random coincidence.

Square directly proportional, and true coincidence counting rate and the system counting rate of random coincidence counting rate and system counting rate First power it is directly proportional.So when activity is higher, or the PET system long in the covering of the axial visual field, random coincidence event become One seriously affects the factor of PET image quality, so that PET quantitative analysis be made to lose meaning.Accurately to random coincidence event Correction be PET system image procossing a key technology.

With being widely used for fast blink crystal and answering for flying time technology (time of flight, TOF) With can more accurately estimate generation position of each true coincidence event on LOR.Also useless farthest to remove simultaneously Chance event provide new method.Useless chance event bring benefit is removed before image reconstruction two aspects, (1) counting for reducing useless event, improves the speed of data processing image reconstruction, (2) farthest reduce ambient noise, mention High image quality.

The method for removing and correcting relative to existing random signal, this method are schemed according to the CT or MR of each scanning human body Picture can farthest utilize the temporal resolution performance of PET system, to farthest remove useless Random event Part is preferably minimized influence of the random signal to picture quality.It also proposed a fast and convenient implementation method simultaneously, it can be with Receive range according to the time window that each scanning human body calculates each LOR in real time.

Summary of the invention

It is applied to face for the acquisition of PET system list type (List mode) data the object of the present invention is to provide a kind of The chance event of bed PET system removes processing method, and this method can be fast and convenient or effectively removes useless chance event, The speed of data processing image reconstruction is improved, and utmostly reduces ambient noise, improves picture quality.

Above-mentioned technical purpose of the invention has the technical scheme that a kind of applied to clinic PET system The chance event of system removes processing method, includes the following steps, Step 1: being examined using same period scanning computed tomography image or MRI image Survey the body region of scanning patient；Step 2: extracting location information (a pair of of detector d of each LOR event₁And d₂), it determines Whether this LOR intersects with the body region prestored, and the disjoint LOR of body region for rejecting and prestoring；Step 3: will with it is pre- The LOR for the body region intersection deposited is according to the two point p intersected with body region₁And p₂, two points are calculated to d₁And d₂Time Difference is denoted as T respectively₁And T₂, read the time difference information T of this LOR record_diff, judge this LOR whether in effective time difference range It is interior effective, reject the LOR not in effective range；Step 4: being retained in the LOR in effective range, and carry out Data correction, figure As reconstruction and post-processing operation.

It is further arranged to: in step 1, image preprocessing being carried out (by PET and CT or MRI to scanning result after scanning Image registration and dimension correction), bed board and bracket removal, threshold value optimize selections, image binaryzation, boundary profile extraction with And contour images filling, post processing of image (expansion, corrosion).

It is further arranged to: in step 2, before the location information for extracting LOR event, by patient's 3D CT image in axial direction It is cumulative to form two-dimentional body region figure, and calculate two-dimentional body contour figure；After the location information for extracting LOR event, LOR is calculated Projection on two-dimentional cross section.

Be further arranged to: the time difference effective range in step 3 is T₁-nσ-ε≤T_diff≤T₂+nσ+ε。

In conclusion the invention has the following advantages:

(1) invalid random coincidence event can be removed and be up to 75%.

(2) the calculating time that removal meets event in vain is optimized, data processing speed is improved.

(3) background noise is reduced.

Detailed description of the invention

Fig. 1 is that list type data random signal of the invention removes flow chart；

Fig. 2 is the treatment process of picture of the present invention；

Fig. 3 is the schematic diagram of three-dimensional optimized minimizing technology of the present invention；

Fig. 4 is the schematic diagram of embodiment fast two-dimensional minimizing technology；

Fig. 5 is using the reconstruction image (background parts) after processing method of the present invention；

Fig. 6 is using the reconstruction image (body part) after processing method of the present invention.

Specific embodiment

Below in conjunction with Fig. 1 to Fig. 5, invention is further described in detail.

Embodiment one, three-dimensional optimized minimizing technology: one kind is acquired for PET system list type (List mode) data, and Chance event applied to clinical PET system removes processing method, including following operating procedure, step 1 are swept using the same period Retouch the body region of CT image or MRI image detection scanning patient；Step 2 carries out image preprocessing to the image scanned (by PET be registrated with CT or MRI image and dimension correction), bed board and bracket removal, threshold value optimize choose, image two-value Change, boundary profile extracts and contour images filling, post processing of image (expansion, corrosion)；Step 3 extracts each LOR thing The location information of part, determines whether this LOR intersects with the body region prestored, if do not intersected with the body region prestored, Reject part LOR；If part LOR intersects with the body region prestored, intersected according to part LOR with body region Two point p₁And p₂, two points are calculated to d₁And d₂Time difference, be denoted as T respectively₁And T₂, read the time difference of this LOR record Information T_diff, judge whether this LOR is effective within the scope of effective time difference, which is T₁-nσ-ε≤T_diff≤ T₂+ n σ+ε, wherein σ is related to system timing resolution, and σ=FWHMtiming/2.355, n usually takes between 2 ~ 3, and ε is due to suffering from Person body is mobile and bring error.The not LOR in time difference effective range is rejected, is retained in time difference effective range LOR.Step 4 carries out Data correction, image reconstruction and post-processing operation, output to the LOR in effective range of reservation As a result.

Embodiment two, fast two-dimensional minimizing technology: the embodiment is roughly the same with embodiment one, and difference is in step 2 In, before the location information for extracting LOR event, first by patient 3D CT image in the axially two-dimentional body region figure of cumulative formation, and Calculate two-dimentional body contour figure.The location information of LOR event is then extracted again.After the location information for extracting LOR event Need to calculate projection of the LOR on two-dimentional cross section；After extracting LOR, LOR is intersected with the two-dimentional body region prestored to sentence It is disconnected whether to need to reject part LOR.

Table 1 by the 70cm radial direction visual field, the 40cm axial direction visual field PET scan frame for, for 30cm diameter, 40cm length Water mould, the ratio of this method removal random coincidence event is up to 75%, and the influence to effective true coincidence event only has 0.23%。

The removal of 1 random coincidence event invalid data of table is compared

It counts	True coincidence event	Random coincidence event
			A before removing	51,991,822	43,785,697
B after removal	51,872,166	11,062,411
			Removal percentage (a-b)/a × 100%	0.23%	74.74%

Fig. 5 and Fig. 6 is the same die body figure after rebuilding, by adjusting display tonal range, Fig. 5 saliency air background portion Point, Fig. 6 shows die body activity region.The method of the present invention can almost remove air background noise, living simultaneously for die body There was only small influence in degree region.

The mean activity value of area-of-interest (ROI) in Fig. 5 and Fig. 6 is shown in table 2.In activity region, difference is only Have 0.23%, and air background part, this method almost all remove ambient noise.

2 area-of-interest activity value of table compares

ROI activity value	Body part	Background parts
			A before removing	123.7369	0.2761
B after removal	123.9419	1.9E10-7
			Relative different (Shu a-b Shu/a) * 100%	0.16%	99.99%

This specific embodiment is only explanation of the invention, is not limitation of the present invention, and those skilled in the art exist It can according to need the modification that not creative contribution is made to the present embodiment after reading this specification, but as long as in the present invention Scope of the claims in all by the protection of Patent Law.

Claims

1. a random event removal processing method applied to clinical PET system, is characterized in that: comprise the steps,

Step 1. Use the simultaneous scanning CT image or MRI image to detect and scan the body area of the patient;

Step 2: Extract the position information of each LOR event (a pair of detectors d ₁ and d ₂ ), determine whether the LOR intersects with the pre-stored body region, and remove LORs that do not intersect with the pre-stored body region;

Step 3. Calculate the time difference from the two points to d ₁ and d ₂ according to the two points p ₁ and p ₂ intersecting with the body region of the LOR that intersects with the pre-stored body region, and denote them as T ₁ and T ₂ respectively, read Take the time difference information T _diff recorded by this LOR, judge whether this LOR is valid within the valid time difference range, and remove the LORs that are not within the valid range;

Step 4: Retain the LOR within the effective range, and perform data correction, image reconstruction and post-processing operations.

2. a kind of random event removal processing method applied to clinical PET system according to claim 1, is characterized in that: in step 1, after scanning, carry out image preprocessing to scanning result (match PET with CT or MRI image; calibration and size correction), bed plate and bracket removal, optimal threshold selection, image binarization, boundary contour extraction and contour image filling, image post-processing (dilation, erosion).

3. A method for removing random events applied to a clinical PET system according to claim 1, wherein in step 2, before extracting the location information of the LOR event, the 3D patient CT images are accumulated in the axial direction to form 2D body area map, and calculate the 2D body contour map; after extracting the location information of the LOR event, calculate the LOR projection on the 2D cross-section.

4. a kind of random event removal processing method applied to clinical PET system according to claim 1 is characterized in that: the effective range of time difference in step 3 is T ₁ -nσ- _ε≤T diff≤T ₂ +nσ+ ε.