Mammography Physical Principles and Instrumentation Lecture Notes III IT Garbett This section of work considers the instrumentation/engineering aspects of mammography. • X Ray tube Design – Anode & focal spot sizes – cathode – tube port/filtration • X Ray Generator & AEC • Image Receptors – Film/Screen – Digital 1 X Ray Tube Design (i)Anode Anode Material: This has already been well covered earlier. Mo most common (ii) Anode Design • Rotating anode. • Anode angles from 10o to −9o . Tube tilt is used to get enough field coverage for very small angles. Tilt is in cathode-anode axis with cathode nearest to chest wall. Use this orientation because anode-heel effect =⇒ anode ‘end’ of beam has smallest intensity. • Metal Insert with earthed anode Use metal insert around portion of tube instead of glass. Enables higher ratings because deposited cathode filament atoms on a glass wall cause a third electrode for stray electrons from the beam. However glass susceptible to damage from the kinetic energy of these electrons. Metal more robust. In addition by setting anode at the same potential (0V) as the metal insert, electrons rebounding from focal area of target (=anode) are as likely to go to the metal insert as back to anode. This reduces so called ‘extra focal X radiation’. • Focal Spot sizes. The effective (projected) focal spot size is a function of the distance along the cathode-anode axis. Largest spot sizes occur nearest cathode. Actual focal spot iszes are 0.3 to 0.4 mm for contact imaging. 0.1 to 0.15 mm for magnification imaging. More image detail occurs at the end of the image where the effective focal spot is smallest. (Where is this then?) (ii)Cathode The cathode has a single element electrically isolated from the cathode cup. The cathode cup is held at negative potential with respect to the filament when the small focal spot size is selected (eg -400 V applied). 2 Figure 1: mammography system design 3 Figure 2: X ray tube tilt. Source: Bushberg et al:Essential Physics of medical Imaging Figure 3: Mammography tube. Source: Thompson et al:Principles of Imaging Science & Protection 4 Figure 4: focal spot size. Source: Bushberg et al: Essential Physics of medical Imaging 5 Figure 5: Cathode cup bias. Source: IT Garbett 6 Figure 6: space charge limited operation. Source: ITGarbett The cathode filament is much closer to the anode in a mammography tube than in a regular X ray tube =⇒ obtain a given mA value at much lower filament temperatures. =⇒ less deposition of tungsten (filament metal) onto the inner surface of the tube insert =⇒ less attenuation of the beam. This is important because mammography spectrum effective energy is << than regular X ray spectrum, and beam attenuation would be too great. Compensation circuits are required to adjust the filament current. This is because the mammography tube operates at low kV and so operates in the space charge limited region. In order to acquire the ‘set’ tube current the filament current must be altered (by the ‘compenstaion circuits’) accordingly. Thus tube current is a function of filament current in mammography. (Tube current is essentially independent of filament current in normal radiography.) Typical values 100mA for large (0.3mm) f spot and 25mA for small (0.1mm) f spot. 7 Figure 7: Tube diaphragm and Be window Source: Forster Equipment for Diagnostic Radiography (iii) Tube port/filtration Already considered filtration material choice with target material. Typical HVT1 value for 30kVp beam is ∼ 0.3 to 0.4 mm Al. This value dpeends also on filtration used and the composition of the compression paddle. Radation dose requires accurate assessement of the HVT value. Worn out tubes will ↑ or ↓ the HVT value ? (Choose one) A beryllium window is used < attenuating than glass. An additional metal diaphragm is used inside the tube that restricts emission of any extra-focal X radiation. X Ray Generator & AEC system (i) The Generator: 1φ, 3φ or high frequency. High frequency becoming standard (enables 1φ power supply, low ripple, fast response time and stability in use. 8 Figure 8: AEC mammography system Source:Source: Bushberg et al: Essential Physics of medical Imaging (ii) AEC system: Also called photo-timer. AEC system comprised of: • Detector Single ionization chamber or array of three semiconductor-diodes. These detect the X radiation transmitted through the breast, grid (if used), and image receptor. • Amplifier This amplifies the current from the detector that is then used to charge up the ... • Capacitor. When the capacitor charges up the voltage across it builds. This voltage signal goes to the... • Comparator. This produces an output signal when the capacitor voltage reaches the value of the other signal provided to the comparatorthe reference voltage. The output from the comparator causes the ... 9 • Termination circuit to switch off the power to the X ray tube. The reference voltage used by the comparator is set by density adjustment at the control panel. This provides 5 to 10 steps to increase or decrease exposure. Generally ∼ 10 to 15 % of exposure per step. A backup timer ends exposure if insufficient exposure occurs beyond a certain time. If this occurs adjustment of the kVp and not the AEC density control must be effected. Why? If the AEC density control is DEcreased then this MAY cause termination of the exposure without the backup timer operating...however the density of the radiograph will be underexposed. If the AEC density control is INcreased then the required exposure for termination is set even higher, and yet again the back-up timer will operate. By increasing the kVp the beam penetration increases and allows enough ionization at the detector site and so enough optical density on a film/screen radiograph. Image Detectors (i) Film/Screen Cassettes low attenuation carbon fibre. Gadolinium oxysulphide (Gd2 O2 S) is mostly used for the fluorescent screen. Green light produced =⇒ green sensitive film.(green dye) Conventional screen/film system for mammography uses a single emulsion film with a single screen. In this case the film is situated proximal to the X ray tube than the screen with the emulsion layer adjacent the fluorescent screen. Most interactions (absorptions) of X photons occur at the top of the screen-near the emulsion. This means that MOST of the interactions are near the active film emulsion and so the generated visible light photons have less travel path to the film. A smaller path of travel =⇒ ↓ lateral spread of light and ↑ spatial resolution. Double screen and double emulsion film systems exist also (see figure). Back screen thickness is same as the single screen system. Front screen thickness is 12 thickness of the back screen. Using this design, both screens contribute the same amount of light to their adjacent emulsions. 10 Figure 9: Film/screen system Source:Source: Bushberg et al: Essential Physics of medical Imaging Also ∼ equal light blurring (from lateral spreading) occurs in both films. There is an anti-crossover barrier (film base material) used to stop visible light crossing to the opposite side film region. (Why do this ?) The QDE for this 2-screen 2-film system is about double that of the single screen single film system. The trade off is worse resolution. (But becomes acceptable in magnification imaging). There is a clear trade off between spatial resolution and sensitivity (aka ‘speed’), when using fluoresent screens. In addition the non-linear response of film to (visible) photon flux also Figure 10: Film/screen system resolution. Source: Smith, A.P Fundamentals Of Digital mammography in Radiology Management Sep/Oct 2003 11 Figure 11: Film/screen dynamic range. Source: Smith, A.P Fundamentals Of Digital mammography in Radiology Management Sep/Oct 2003 means that there is a very small tolerance to the exposure values that can be used to show contrast. (‘small exposure latitude’) (ii) Digital Full field digital mammography (FFDM) employs an area beam detector. There are basically two designs of the ‘flat panel’ detectors used in FFDM. • Indirect Conversion • Direct Conversion (a) Indirect Conversion (IC) Earliest design. A two-step process. step 1: X photons absorbed within fluorescent material CsI(Tl). =⇒ visible light photons generated. step 2: v light photons interact with photo-diodes. These diodes arranged in an array. Each diode produces a brief flow of current in proportion to v light energy 12 Figure 12: Indirect Conversion CsI/TFT & Resolution. Source: Smith, A.P Fundamentals Of Digital mammography in Radiology Management Sep/Oct 2003 incident upon the diode⇒ transistor ⇒ current amplification ⇒ charge built up and then stored (eg capacitor on the output side of transistor circuit). Effectively there are two overlying layers of arrays: The photo-diode array & the thin film transistor (TFT) array. These are both deposited onto a glass substrate. A further layer of transistors involved with ‘read-out’ is also used in addition to the ‘charge collector’ array of transistors. Unfortunately, IC designs ⇒ light spread problem (like film/screen systems). Thicker fluorescent screens =⇒ ⇓ spatial resolution (To some degree the use of CsI (long thin needle crystals) reduces this effect) However, a problem occurs compared to screen/film: The photo-diode & transistor arrays are NOT transparent to X photons, unlike film, =⇒ cannot put the ‘array layer’ proximal to the X ray tube. Instead it must be situated at exit surface of the fluorescent layer.=⇒ ⇓ spatial resolution compared to film/screen system. 13 Figure 13: Direct Conversion. Source: Smith, A.P Fundamentals Of Digital mammography in Radiology Management Sep/Oct 2003 (b) Direct Conversion (DC) Direct Conversion (DC) employs a single-step process in production of the image signals. Uses a photo-conductive layer. This ‘photoconductor’ produces electron-hole pairs when X-photons are absorbed by the photoconductor material. An Electric field is applied across the photoconductor. Electrons and holes then move in direction of the field. → − Electrons or holes (depends on direction of the applied E field) move toward an array electrode (each electrode ≡ a pixel). This charge then stored on an electrode for that pixel area. → − NOTE: Due to movement along the axis of the E field, there is no lateral spreading of the electron or hole in its trajectory =⇒ very narrow point spread function (PSF ' 1 micrometre). Because of the action of the applied electric field, the thickness of the photoconductor can be freely chosen. 14 Figure 14: Direct Conversion and resolution. Source: Smith, A.P Fundamentals Of Digital mammography in Radiology Management Sep/Oct 2003 15 µ plot Source: Smith, A.P Fundamentals Of Digital ρ mammography in Radiology Management Sep/Oct 2003 Figure 15: Selenium In practise, thickness is chosen to stop 95% of the incident X ray beam energy. The photoconductor used is amorphous SELENIUM. The virtue of this material (apart from a well established manufacturing history in zeroradiography) is its attenuation characteristics. The K-absorption edge is well placed in relation to the photon energies used in mammographic spectra. The TFT array for DC detector systems does not require the addition of the photo-ddiode array used in the IC detectors. This enables the same manufacturing process well established in large area TFT/LCD arrays, commonly used in computer displays. ( It should be noted that an alternative to TFT arrays for DC detectors is the use of charge coupled devices (CCDs), however these will not be discussed here and also suffer from light spread problems). 16 Pixel Size and Resolution Pixel size and IC detectors: The smaller the pixel size (≡ array detector element) then the greater the ability of the image to display items that are separate as actually separate (ie the greater the spatial resolution). However, IF the information presented TO the detector array itself does not contain two separate signals, then there is no way that the image will display separation, even if it actually exists within the irradiated structure. This will be the case no matter how SMALL the pixel size. It is because of the light scattering in the fluorescent screen that the IC detectors are ‘scintillator’ limited in their resolution. =⇒ there is no reason to reduce the pixel size below a certain value for IC detectors. This limiting value ' 100 micrometres. This is just at the limit of ‘use’ in mammography. (Calcifications of 100 micrometres need to be resolved). Pixel size and DC detectors: Because of the lack of lateral spread of the generated electron-holes in their trajectory to the detector panel, the resolution for DC detectors is limited by the size of the pixel. If pixel size is reduced to very small sizes then eventually a size is reached that may be inefficient in regard to the incident photon flux (ie a large percentage of the pixel elements may have no incident photons ! ). (Quantum limited) At least one manufacturer(Hologic) uses a pixel size of 70 micrometres. There is, inevitably, a practical limit (as well as a quantum limit) on the use of very small pixel sizes. If one halves the pixel size (side of a square) then the number of individual stored numbers increases by fourfold. Both data storage requirements and network bandwidth needed for image distribution becomes more costly. Intrinsic Efficiency Intrinsic efficiency is the ratio of energy absorbed in the detector energy incident upon the detector Screens used in film/screen mammography =⇒ 50% to 70%. IC detectors using CsI(Tl) =⇒ 50% to 80%. As discussed earlier the use of Se photoconductor material =⇒ 95% intrinsic detector efficiency for DC detectors. Unlike conventional screen/film mammography, the optimum effective photon energy has not been fully established for digital mammography. 17 Figure 16: absorption percentage. Source: Smith, A.P Fundamentals Of Digital mammography in Radiology Management Sep/Oct 2003 18 Detectors with a high intrinsic efficiency have potential to allow the use of higher effective photon energy (why?) This potential arises in both IC and DC detector designs.(see percentage absorption graph) DQE in digital mammography The DQE is a measure of the transfer of the signal to noise ratio within the system. Recall (from study of fluoroscopy) that: DQE = 2 SN Rout 2 SN Rin In fact the DQE varies with the spatial frequency. (ie DQE is a function of the spatial frequency). This arises because in fact both the output noise and the output signal are dependent upon the spatial frequency (Recall spatial frequency is the rate of change of a structure with distance, thus 2 line pairs per mm is a frequency of 2 mm−1 and yields: low exposure value → high exposure value →low exposure value → high exposure value across one millimetre). If this ’variation’ in radiation distribution is not reproduced well at the output, there is a consequent low signal and (supposing for simplicity the noise is constant) this means a low SN R out value. This would indeed be the case where the signal cannot be discerned (eg from two edges of two structures) because of light blurring (as happens in both film/screen and also IC digital detectors). Consequently we expect the DQE variation with spatial frequency to be poorer in the case of both screen film and also IC detectors. Assessment of Image Quality A standard quality control procedure for accreditation of mamography centres employs the use of a standard imaging phantom. The resultatnt images need to match minimum values for certain characteristics. One such phantom is the ACR (American College of Radiologists) phantom. This perspex phantom is made equivalent to a 4.2 cm compressed breast consisting of a 50% glandular 50% adipose mix of tissue. It contains fine and coarse structures (fibres, specks and masses) that range from visible through to invisible on a standard mamographic film image. Spatial resolution checks are routinely made using a line pair test object. For an ideal digital system the pixel size would determine the spatial resolution. 19 Figure 17: DQE vs spatial frequency. Source: Smith, A.P Fundamentals Of Digital mammography in Radiology Management Sep/Oct 2003 Figure 18: ACR Mammography Phantom. Source: Smith, A.P Fundamentals Of Digital mammography in Radiology Management Sep/Oct 2003 20 Figure 19: ‘Line Pair’ Test Object. Source: Smith, A.P Fundamentals Of Digital mammography in Radiology Management Sep/Oct 2003 For example a pixel size (side of a square) of 100 micrometres means that effectively one lead strip of 100 micrometres can occupy one pixel width, and the gap of 100 micrometres (before the next strip) can occupy fully the adjacent pixel. This means 1 line pair occupies 200 micrometres. Thus the limit of resolution for this system would be 5 line pairs per millimtre. However, as previously mentioned, IC digital detectors have a spatial resolution imposed ultimately by the light spread in the fluorescent material. Pixel size is only a good predictor of spatial resolution before this limit is reached (100 micrometres). Detector Size and Field of View. For film/screen there are two cassette sizes used: 18cm x 24 cm and 24cm x 30 cm. Larger size can be used with larger breast. Use of larger cassette with smaller breast wasteful of film and processing materials. Use of smaller cassette size with larger breast =⇒ need two overlapping irradiations to ensure coverage of the breast contents. =⇒ ⇑ absorbed dose, multiple compressions, and is time consuming. For digital receptors, these flat panel detectors are very costly and also heavy and unwieldy to change =⇒ use just one detector. =⇒ use a large detector. The large digital detector does not have same problems as film/cassette because the image can be cropped and there are no processing materials. There is also a bonus in using the larger detector area for the smaller 21 breast if magnification mode is used: This means that a pixel on the magnified image represents a smaller actual breast dimension. Suppose a breast that ‘fits’ a 18cm x24 cm using contact imaging, is projected onto a larger 24cm x 30 cm detector using magnification technique. 5 30 × 24 = Then area magnification, am = 18 × 24 3 3 This means that a detector element represents now = 0.6 of the object 5 area it represented with no magnification. Suppose a detector element area (pixel area) is actually 70 micrometres x 70 micrometres. Then the magnified pixel represents 70 × 70 × 0.6 = 2940 (square micrometres) of the breast. This √ is equivalent to a pixel that is of size (along one side of its square area): 2940 = 54.221767µm ' 54.2µm Therefore imaging a ‘regular size’ breast using magnification with a large area detector is in fact equivalent to using a detector in contact mode that has pixel dimensions of 54 µm. This represents a benefit in image resolution for ' 80% of women. 22