Digital Radiography : Flat Panel

Transcription

1 Digital Radiography : Flat Panel

2 Flat panels performances & operation How does it work? - what is a sensor? - ideal sensor Flat panels limits and solutions - offset calibration - gain calibration - non linearity - Remanence (LAG Ghosting effect) - Blooming (over-brightness) Main characteristics - pixel size and number of pixels - analog to digital convertor (amount of grey levels) - basic spatial resolution (SR b ) according to EN signal/noise ratio (SNR) according to EN dead pixels and dead lines - range of energy

3 How does it look?

4 How does it work? A flat panel is a huge amount of small sensors (> ) called pixels. Once irradiated, each one will deliver an electrical signal which is proportional to the dose. X-rays Plastic layer X-rays light Scintillator Sensors (photodiodes) Plastic layer Glass substrate

5 How does it work? Each sensor converts the radiation into an analogic voltage and afterwards into a grey level. V Vmax Analogic voltage Grey value X-rays pixel ADC

6 Analogic value Ideal answer of the pixels Same answer for all pixels X-ray dose No radiation 0 V

7 Flat panels performances & operation How does it work? - what is a sensor? - ideal sensor Flat panels limits and solutions - offset calibration - gain calibration - non linearity - Remanence (LAG Ghosting effect) - Blooming (over-brightness) Main characteristics - pixel size and number of pixels - analog to digital convertor (amount of grey levels) - basic spatial resolution (SR b ) according to EN MTF (Modulation Transfer Function) - signal/noise ratio (SNR) according to EN dead pixels and dead lines - range of energy

8 First issue: offset Analogic value Pixel 4 X-ray dose Pixel 2 No radiation 0 V Pixel 1 Pixel 3

9 First issue: offset Summary A pixel delivers a voltage even without radiation. This voltage is called the OFFSET Each pixel has a different offset The offsets are temperature sensitive

10 First issue: offset Dark image with offset Grey levels up to 1300 Image is not dark (due to offset) Image is not uniform (different offset for each pixel)

11 First issue: offset Dark image with offset Grey levels up to 1300 Image is not dark (due to offset) Image is not uniform (different offset for each pixel)

12 How to solve the offset issue? Switch ON the detector and wait until temperature is stabilized (30 minutes) Acquire an image without radiation (This is called a dark image) Look at the offset values and memorize them in a table Substracting that table to future images will correct the offset errors This is called the OFFSET calibration

13 First issue: offset Dark image after offset calibration Grey levels up to 30 Image is dark Image is uniform

14 First issue: offset Dark image before & after offset calibration

15 Second issue: gain Analogic value Same answer for all pixels X-ray dose Pixel 2 Pixel 4 Pixel 1 Pixel 3

16 Second issue: gain Summary Each pixel delivers a different voltage for the same dose. The GAIN is different for each pixel The gains are temperature sensitive

17 Second issue: gain Effect of gain dispersion Grey levels dispersion 1500 Image is not uniform when flat panel is uniformly irradiated

18 Second issue: gain Effect of gain dispersion Grey levels dispersion 1500 Image is not uniform when flat panel is uniformly irradiated

19 How to solve the gain issue? Switch ON the detector and wait until temperature is stabilized (30 minutes) Adjust kv so that averaged grey level is around 8000 (max value/2). Look at the grey values, calculate a correction factor for each pixel and memorize them in a table Multiplying that table with future images will correct the gain errors This is called the GAIN calibration

20 Second issue: gain Image after gain calibration Image is uniform

21 Second issue: gain Image before & after gain calibration

22 Gain calibration Why calibrate around max grey value? 2 Grey value Pixel 4 Pixel 3 Max= SATURATION Pixel 2 Pixel 1 X-ray dose For a high dose, pixels having a high gain will saturate. In above example, pixels 3 and 4 cannot be corrected. Calibration cannot be done.

23 Gain calibration Why calibrate around max grey value? 2 Grey value Pixel 4 Pixel 3 Max=16384 Pixel Pixel 1 X-ray dose With a reduced dose, calibration can be done Pixel 1: correction factor = 1 Pixel 2: correction factor = 6862/7380= Pixel 3: correction factor = 6862/7690= Pixel 4: correction factor = 6862/8052=0.8522

24 Third issue: non linearity Calibration is performed for ONE grey level only Grey value X-ray dose Calibrate the gain with a grey level close to the one used for parts inspection

25 Calibration summary A flat panel needs: An offset calibration (dark acquisition, no X-ray) A gain calibration (with X-ray) performed through a material having similar thickness Those calibrations must be done when flat panel temperature is stable.

26 LAG Remanence - Ghosting Detector LAG is defined as a residual signal from previous acquisitions. It can produce severe ghosting effects. Grey level ideal real STOP radiation Time

27 Over brightness - Blooming When the irradiation of a sensor is too high, an overflow of electrons in adjacent the pixels will occur. This is called «blooming effect». This will generate difficulties to inspect: parts edges parts with important thickness variations High irradiation may also produce severe damages to the detector sensors and to the electronics

28 X-RAYS part detector Over brightness - Blooming Solutions Part edges: compensate the thickness using a collimator blooming

29 X-RAYS part detector Over brightness - Blooming Solutions Thickness variations: compensate whith contrast flattening (filtration of the beam) blooming Filter (Cu, ) Contrast ratio = 10 5

30 Over brightness

31 Over brightness

32 Over brightness

33 Flat panels performances & operation How does it work? - what is a sensor? - ideal sensor Flat panels limits and solutions - offset calibration - gain calibration - non linearity - Remanence (LAG Ghosting effect) - Blooming (over-brightness) Main characteristics - pixel size and number of pixels - analog to digital convertor (amount of grey levels) - basic spatial resolution (SR b ) according to EN signal/noise ratio (SNR) according to EN dead pixels and dead lines - range of energy

34 width X Pixel size and amount of pixels pixel X length Detector type Length [mm] Width [mm] X [µm] Number of pixels FP Digit x 1536 FP Digit x 1024 FP Digit x 2048

35 Analog to digital convertor (ADC) The ADC converts the analogic signal delivered by the pixel to a digital value (grey level) X-rays pixel Analogic voltage ADC Grey value The most frequent resolutions are: 14 bits and 16 bits 14 bits means 2 14 = different grey levels 16 bits means 2 16 = different grey levels Why is that convertor resolution so important?

36 X-rays Why is the ADC so important? part Detector A thin defect will generate very small dose differences impacting the pixels. Pixels will generate small analogic differences. But is the ADC able to provide different grey levels for small analogic differences? With a 10 bits ADC: smallest difference will be 1/1024 = 0.098% With a 14 bits ADC: smallest difference will be 1/16384 = % With a 16 bits ADC: smallest difference will be 1/65536 = % CONCLUSION: good ADC will detect small contrasted structures

37 Flat panels main characteristics Detector type Length [mm] Width [mm] X [µm] Number of pixels ADC resolution FP Digit x bits FP Digit x bits FP Digit x bits

38 Basic Spatial Resolution (SR b ) According to EN ISO :2010 The SR b is a measurement of the spatial resolution of the flat panel. It is not a measurement of the image quality! Flat panel Duplex IQI (EN462-5 or ASTM E ) The duplex IQI must be put on the flat panel with an angle between 2 and 5.

39 Basic Spatial Resolution (SR b ) According to EN ISO :2010 Line pair number 13D 12D 11D 10D 9D 8D 7D 6D 5D 4D 3D 2D 1D Duplex IQI specification (according to EN 462-5) Wire diameter and distance between line centers [mm] 0,050 0,063 0,080 0,100 0,130 0,160 0,200 0,250 0,320 0,400 0,500 0,630 0,800 U i Intreseque unsharpness [mm] Basic spatial resolution SR b [pl/mm] Basic spatial resolution SR b [µm]

40 Basic Spatial Resolution (SR b ) According to EN ISO :2010 First pair having dip < 20% of background

41 Basic Spatial Resolution (SR b ) Measurement according to EN ISO : Use of Duplex wire (EN 462-5) - focal spot to detector distance must be 1 meter or more - focal spot size must be 1 mm or less - Duplex IQI must have an angle between 2 and 5

42 SR b understanding The SR b indicates the smallest defects the flat panel can distinguish. It can not be better than the pixel size. d d d d Best situation: wires aligned on pixels Resolution = pixel size Pixels can distinguish the wires GOOD SR b d Worst situation: wires not aligned on pixels Resolution > pixel size Pixels can NOT distinguish the wires BAD SR b d BUT. SR b does not indicate the smaller defect that can be seen

43 SR b smaller defect detectable The SR b uses pairs of wires. Its measurement indicates the ability to distinguish both wires. BUT Let s put a 50 µm wire on a 127 µm detector you ll see it! Why? d As the wire is covering a significant part of the pixel, the pixel will produce a little less electrical signal. If the ADC is good enough, a different grey level will be available. ADC resolution is quite important to detect small differences

44 SR b smaller defect detectable A 50 µm wire (W19) on a 127 µm pixel size W19 = 50 µm

45 Final spatial resolution is better thanks to the geometrical magnification X 2 Small details are magnified. <50 µm line can be distinguished.

46 Magnification limit

47 Magnification limit

48 BALTOMATIC AIS203RD Inspection Real Time System for metallic & non-metallic components

49 Signal to noise ratio (SNR) Definition: Noise is an unwanted signal superposed with the interesting information. Impact: Noise gives an image a grainy or snowy appearance. consequence: Noise can reduce the visibility of small defects within an image, especially for lowcontrast details.

50 Signal to noise ratio (SNR) Effect of increasing noise added to the information

51 Signal to noise ratio (SNR) SNR = 54 SNR = 180

52 Signal to noise ratio (SNR) Origins of noise 1. Quantum noise due to the physics of the radiation. X-ray photons are not evenly distributed over the detector surface. 2. Electronic/electrical noise due to external disturbances and to the electronic of the detector

53 Signal to noise ratio (SNR) Origins of quantum noise Random impacts on each pixel

54 Signal to noise ratio (SNR) Origins of quantum noise Waiting for an average of 100 photons Waiting for an average of 1000 photons Information = 100 Information = 1000 Noise= 9 (9%) Noise= 33 (3.3%) Quantum noise is a random phenomenon. The random effect is reduced when quantity of photons increases

55 How can we improve the SNR? Increasing the number of photons impacting the flat panel will improve the SNR. -Increase the current in the X-ray source. Always work at the maximum ma -Increase the irradiation time, so that pixels will capture more photons. Increase the integration time Increase the number of frames averaged

56 SNR measurement According to EN ISO : Measurement must be done outside the weld, on an area where grey levels are homogeneous - Measurement must be done inside a window of 50 X 50 pixels - Minimum SNR is 70 (class A)

57 Dead pixels Due to the manufacturing process. A few pixels are defective for several reasons: they are not delivering a voltage when irradiated their gain is not high enough Their offset is too high Newly manufactured detectors may have several dead pixels (on a total around to ). Some manufacturers are proposing a «very high grade quality», selecting the detectors. Number of dead pixels is close to zero. They are used only for tomography applications as price is very high.

58 Dead pixels Example of dead pixels

59

60

61 Questions?