Computed Tomography Scanner (CT-Scan) (Pemindai Tomografi Komputer)

PT. SIEMENS INDONESIA

1

Computed Tomography Scanner (CT-Scan) (Pemindai Tomografi Komputer) Oleh: Daniel Kartawiguna, ST., MM., M.Acc. PT. SIEMENS INDONESIA – Medical Solutions [email protected]

Always Thinking Ahead.

DISAJIKAN DALAM RANGKA PELATIHAN CT SCAN DOSEN ATEM JAKARTA, 19 JUNI 2006

2

MATERI PELATIHAN

I. Dasar Tomografi Komputer II. Sistem Tomografi Komputer Spiral III. Sistem Tomografi Komputer Multi Irisan (MSCT) IV.Pemeliharaan dan Perbaikkan V. Aplikasi Klinis MSCT


(c) Daniel Kartawiguna, 2006 [email protected]

1


3

I. Dasar-dasar Tomografi Komputer


4

Dasardasar Tomografi Komputer

POKOK BAHASAN

1. Tomografi Komputer dan Sejarah Perkembangannya 2. Prinsip Tomografi Komputer 3. Komponen2 Tomografi Komputer 4. Sistem Tomografi Komputer yang dikembangkan oleh SIEMENS 5. Kalibrasi 6. Pemeriksaan Kualitas Gambar



2


5

Dasardasar Tomografi Komputer

POKOK BAHASAN

1. Tomografi Komputer dan Sejarah Perkembangannya 2. Prinsip Tomografi Komputer 3. Komponen2 Tomografi Komputer 4. Sistem Tomografi Komputer yang dikembangkan oleh SIEMENS 5. Kalibrasi 6. Pemeriksaan Kualitas Gambar


6

Penemuan Sinar-X Sinar-X ditemukan pada tahun 1895 oleh seorang ahli fisika Jerman yang bernama:

Wilhelm Conrad Röntgen



3


7

Sinar-X memungkinkan orang pertama kali untuk melihat struktur dari tubuh manusia tanpa melakukan operasi/pembedahan. Keterbatasan: • Gambar yang dihasilkan merupakan superimposisi dari obyek yang diamati. • Tidak dapat menggambarkan jaringan lunak.


8

CT memberikan solusi terhadap keterbatasan ini... Pada tahun 1972, Hounsfield dan Ambrose yang bekerja di Central Research Labs. of EMI, Ltd di Inggris menghasilkan gambar klinis pertama dengan CT Scan.



4


9

Gambar CT yang pertama memungkinkan pertama kali kita dapat melihat: - Tomographic atau “Slice” anatomy - Perbedaan densitas

Keterbatasan: • Memerlukan waktu yang lama. Always Thinking Ahead.

• Resolusi perlu ditingkatkan.

10

CT-Scanner Siemens yang pertama

SIRETOM (1974) Waktu scan 7 min, matrix gambar 80x80 Pixel, scan field 25 cm, resolusi spatial 1,3 mm (4LP/cm) Always Thinking Ahead.


5


11

Perkembangan kualitas gambar CT Dari tahun 1972 - 2000

SIRETOM (1974)

SOMATOM Plus 4 UFC (1996)


12

Ringkasan Sejarah Perkembangan Tomografi Komputer 1917

1963 1972 1974 1975 1979 1989 1998 2000

J.H. Radon: Transformasi Radon, gambar dari obyek yang tidak diketahui dapat digambarkan dari proyeksinya. A.M. Cormack: mengembangkan teknik untuk menentukan distribusi penyerapan tubuh manusia. G.N. Hounsfield dan J. Ambrose: menghasilkan gambar CT pertama kali untuk keperluan klinis. 60 unit CT terpasang untuk pemeriksaan kepala First whole body scanner in clinical use Hounsfield dan Cormack dianugerahi hadiah Nobel Spiral CT Multislice CT > 30000 clinical CT installations



6


13

POKOK BAHASAN

1. Tomografi Komputer dan Sejarah Perkembangannya 2. Prinsip Tomografi Komputer 3. Komponen2 Tomografi Komputer 4. Sistem Tomografi Komputer yang dikembangkan oleh SIEMENS 5. Kalibrasi 6. Pemeriksaan Kualitas Gambar Always Thinking Ahead.

14

Prinsip Dasar Pemindai Tomografi Komputer Pelemahan intensitas sinar-X oleh obyek yang ditembus oleh sinarX tersebut. Berkas sinar-X yang menembus suatu obyek akan mengalami pelemahan (kehilangan energi) yang diakibatkan oleh: Penyerapan oleh obyek Sinar-X datang Penyebaran

SINARSINAR-X DIHAMBURKAN

JARINGAN TUBUH

Sinar-X Diteruskan

Penyerapan oleh jaringan adalah sebanding dengan densitasnya Jaringan dengan kepadatan yang tinggi Always Thinking Ahead.

PERLEMAHAN LEBIH BESAR


Jaringan Dengan Kepadatan rendah

PERLEMAHAN LEBIH KECIL

7


15

Prinsip Dasar Pengukuran

Ι0

µ Ι

I=I0 ·exp(- ∫ µ ds) “Line Integral”

∫ µ ds = - ln(I/ I0) Koefisien atenuasi, nilai CT Always Thinking Ahead.

16

Proyeksi (Projection)



8


17

Projection To be able to compute in acceptable quality a CT image, a sufficiently high number of attenuation integrals or projection values have to be recorded. One readout of all detector elements is a so called projection. A CT scan for a single image (skice) requires the measurement of a high number of projections ( approx. 800 – 1400) .


18

Generasi CT-Scan



9


19

Prinsip dasar akuisisi data Tabung sinar-X

Parallel beam geometry (1st generation scanner)

Berkas sinar-X

Detektor sinar-X

Obyek pemeriksaan


20

Raw data (Data Mentah) Sequence of attenuation profiles

1. projection

time (projections) attenuation profile (channels) Always Thinking Ahead.


10


21


2. projection

time (projections) attenuation profile (channels) Always Thinking Ahead.

22


Full rotation → CT raw data time (projections) attenuation profile (channels) Always Thinking Ahead.


11


23

REKONSTRUKSI • How do we reconstruct the image?

• One could solve more than 260.000 unknown µ by measuring the attenuation in about 1.400 readings with about 700 detector channels. Disadventage: calculation can be started only after a full revolution, spiral scan wouldn`t be time effective

or • We could simply add the projections by smearing them back upon each other across the image. • This is simple back-projection. Always Thinking Ahead.

24

Rekonstruksi Gambar

Dasar Matematika: Radon (1917) Dalam praktek: a) Fourier method atau Raw data

b) filtered backprojection

CT image



12


25

Unfiltered Backprojection Scanned object

nice reconstruction, but not very accurate... Reconstructed object using 128 64 32 16 8 4 3 2 1 projection(s)


26

Solution: Filter

Filter: convolution kernels

Attenuation profile of a cylinder

Filtered attenuation profile



13


27

Filtered Backprojection Scanned object

This reconstruction looks much better ... Reconstructed object using 128 64 32 16 8 4 3 2 1 projection(s)


28

Convolution Kernel

sharp kernel

smooth kernel



14


29

Convolution kernel


30

Convolution kernel The original profile ( = 1 projection) is multiplicated with the convolution kernel; the result is the concolved projection. Different convolution kernel provide different image sharpness. The use of a smoothing kernel reduces the noise in the image and improves low contrast resolution. The use of an edge enhancing kernel improves high contrast resolution, but increases the noise in the image.



15


31

Convolution Kernel (lanjutan)

Kernel akan menentukan • Ketajaman

Not independent:

• Noise

sharp image → more noise less noise → smoother image

• Batas


32

Simple back-projection w/o and with convolution



16


33

Simple back-projection w/o and with convolution It is necessary to carry out measurements in all directions, i.e. at least over an angular range of 180 0, and to determine many narrowly spaced data points for each projection. For simple backprojection each projection value is added to all the picture elements in the computer memory along the direction in which it has been measured. The farreaching signal contributions due to the backprojection process lead to an unsharp image, which is insufficient for the diagnosis of cpmplex structures. To avoid this unsharpening each projection has to be convoluted before backprojection with a mathematical function, the concolution kernel. This constitutes a pointwise multiplication of the convolution kernel and the attenuation profile and addition of the resulting values. In essence, this represents a high pass filtering procedure which generates over- and undershoots at object boundaries. For a positive signal, negative undershoots are generated. Concolution additionally offers the possibility to influence image characteristics by the choice and design of the concolution kernel – from soft or smoothing to sharp or edge enhancing.


34

Perangkat Keras Sistem CT-Scan Bagian luar... • • • • •

Gantry Meja Generator Konsol Komputer

Gantry

Generator Meja

Konsol Komputer Always Thinking Ahead.


17


35

Perangkat Keras Sistem CT-Scan Bagian dalam ... •

Tabung

•

Detektor

•

DAS*

Tabung

DAS Detektor * Data Acquisitions System (Sistem Akuisisi Data) Always Thinking Ahead.

36

CARA KERJA PESAWAT CT SCAN Sumber Sumber Radiasi Radiasi Sinar-X Sinar-X Detektor Detektor Rekonstruksi Rekonstruksi ++ post post processing processing



18


37

Overview over a CT system

Image Image Processing Processing


38

Overview over a CT system A modern CT can be divided into three parts:

System Control Runs the Operating System software (VMS, UNIX or Win NT.) Runs the Application software to provide a user interface (SOMARIS). Supports Scan Control by distribution of scan parameters and instructions. Scan Control Usually microprocessor based. Directly controls the scanner hardware via firmware routines stored in PROM. It is responsible for the safety of the patient, operator and system hardware. It distributes parameters from the host to the other functional groups in the system. It serves as the interface for all communication between “system control” and the gantry. Image Processing This is a specialized array of computers that are responsible for the reception of data from “scan control” and the subsequent reconstruction of an image.



19


39

Blockdiagram image processor


40

Blockdiagram image processor An image processor consists on 4 functional blocks: 1: Pre- processor 2. Convolver 3. Backprojector 4. Imager

The data measurement system supplies the scan data in serial order to the receiver module, which is a part of the preprocessor.



20


41

Blockdiagram preprocessor


42

Blockdiagram preprocessor A preprocessor has to compensate the measured data for : 1)

Electrical drifts

2)

Dose variations

3)

X-ray attenuation law

4)

Beam hardening

5)

Mechanical deviations of the scanning system

The input (measured data from the aquisition system) is called a reading, the output a projection.



21


43


44

PGA decoding The signals from the detector have a very wide range. To cover the whole dynamic range of input signals, a „Programmable Gain Amplifier“ (PGA) is used (also called FPA, floating point amplifier). PGA: The PPA is an amplifier which selects its gain automatically. The selected gain can be 1, 8 or 64. The gain used is indicated by the two bits called `PGA Bits´. PGA Decoding:In order to calculate the actual attenuation, the PGA bits are decoded in the SMI. This is done in the preprocessing step „PGA decoding“ (also called „FPA decoding“). Amplification 64:If the signal from the detector was very small (i.e., high absorption in the scanfield), the amplifier will have used a factor of 64. The resulting data in the SMI will be the 14 bit from the ADC, preceeded by many zeroes, in other words, a rather small numerical value. Amplification 8:If the amplification was 8, the signal was larger. In the SMI, the 14 bit are shifted 3 bit to the left, equaling a multiplication by 8, or a larger numerical value.



Amplifiaction 1: If the signal from the detector was large (e.g., only air in the scanfield), the PGA will have used an amplification of 1. This will result in a large number in the SMI, because the 14 bit are shifted 6 bit to the left, equaling a multiplication by 64 or a rather large numerical value.

22


45

Offset correction


46

Offset correction Offset voltage : In the DMS, ADCs are used that can´t measure negative voltages. This would falsify the measurement, if very small detector signals ( = high absorptions ) have to be measured. To avoid this, in the DMS an offset voltage is added to the signal, the signal is measured, and in the SMI the offset signal subtracted again, leaving the true value only.

Offsets are channel specific: Because the analogue offset may be slightly different for each ADC, or, to be precise, even for every integrator board channel, the actual offset has to be measured prior to the scan for every channel. Offset measurement:



With each scan start, a measurement is started without X-ray and the data are stored in the image processor as offset data.

23


47

Logarithmation


48

Logarithmation Logarithmation is done because the attenuation of an X-ray beam follows an exponential law. I = Io e - µ d

The calculation of the object attenuation „A“ requires the calculation of the logarithmic value of the measured radiation intensities I and I 0: A= ln I 0 – ln I The logarithmation is done using a table of log values.



24


49

Normalization


50

Normalization During the measurement, the intensity Io of the X-ray beam varies (exaggerated in the picture for clear visualization). Monitor value: A monitor element measures the unattenuated radiation as a reference value.This value is called “Monitor value”. Normalization : During the preprocessing step “Normalization” , this monitor value IM is substracted from each channel value I: In 1/I – In 1/IM



25


51

Calibration


52

Calibration Each detector has a different sensitivity. This can vary with the time and must be compensated. The channel specific sensitivity differences are compensated by calibration. Technically, that means taking an air scan and then subtracting the channel values obtained in air from the normalized channel values. Then preprocessing step calibration requires the base calibration tables, that are measured during the last calibration. During the last tune up, the differences of each combination of kV, mA, slice width etc. were measured, so that only the base calibration is required on a daily basis. The other calibration tables for different settings of kV, mA etc are calculated from the difference tables and the base calibration table. Always Thinking Ahead.


26


53

Channel Correction


54

Channel Correction Technically, the preprocessing step „channel correction“ is a multiplication of each channel value with a correction factor. Because a correction is needed for many reasons, several individual tune- up tables contribute to the resulting one factor in the preprocessing. Channel correction includes: -

Beam hardening correction

-

Cosine correction

-

Channel coefficient correction and

-

Water scaling



27


55

Beam – Hardening Correction


56

Beam – Hardening Correction X-ray spectrum: Tubes generate polychromatic radiation, i.e.different wavelength are contained in the spectrum. Just as with visible light, the higher energies or shorter wavelength can penetrate the objects better than the softer part of the spectrum. Beam hardening causes in homogenous objects (e.g. a water phantom) an inhomogeneity. That means, the CT values in the center are different from the outer values. The correction is done by taking data of a reference phantom (mostly a 20cm water phantom) and the correction data are used during preprocessing step „beam hardening“ for the correction of the scan data. Always Thinking Ahead.


28


57

Cosine - Correction


58

Cosine - Correction Measured attenuation: The Cosine – Correction considers that the x-ray beam is fan shaped. The length of the path of the radiation through an object depends on the angle alpha between the central beam and any other beam. Corrected attenuation: The correction of the different channel outputs is done by using the cosine function. For each channel, the table contains the cosinevalue of the corresponding fan angle „alpha“. The measured attenuation is multiplied with the table values. Always Thinking Ahead.


29


59

Channel Coefficient Correction


60

Channel Coefficient Correction Just like „Calibration“, the pre-processing step ´Channel Correction´ compensates for sensitivity differences of the detector. The difference is that the Channel correction compensates for nonlinearities in the area of attenuated radiation, i.e., with an object is in the scan field. Parameter: The parameters which determine the detected radiation energy are: Tube Voltage Slice Thickness Object Attenuation (Head or body) Correction tables: The sensitivity compensation is done with values which are determined during the tune-up. Always Thinking Ahead.


30


61

Water Scaling


62

Water Scaling The water scaling sets CT value of water to 0 HU. This factor depends on the energy received by the detector; the parameters for the scaling are: - tube voltage - tube current - slice thickness



31


63

Generasi Tomografi Komputer “Generasi” digunakan untuk membedakan rancangan konfigurasi tabung-detektor pada CT.


64

Generasi ke-1: Pencil Beam



32


65

Generasi ke-1 • • • • • • • •

• •

Gerakan translasi dan rotasi Berkas sinar-X berbentuk pensil (pencil beam) Geometri berkas sinar paralel. FOV (field of view) 24 cm. Menggunakan 2 buah detektor sehingga sekali scan dapat menghasilkan 2 irisan. 160 berkas paralel/proyeksi. 180 proyeksi dengan interval 1 derajat. Detektor tidak dapat mendeteksi perbedaan intensitas sinar-X yang sangat besar, oleh karena itu kepala yang diperiksa harus dikelilingi oleh kantong berisi air. Kristal NaI yang digunakan sebagai detektor memiliki waktu “afterglow” yang nyata. Keuntungan: pengaruh hamburan radiasi pada detektor ditiadakan karena berkas sinar-X yang berbentuk pensil.


66

Generasi ke-2: Partial Fan Beam



33


67

Generasi ke-2 • Menggunakan 30 linear array detector. • Kerugian: adanya pengaruh radiasi hamburan dan meningkatnya intensitas kearah tepi dari berkas sinar-X yang berbentuk kipas. Hal ini diatasi dengan penambahan filter dasi kupu-kupu pada jendela tabung sinar-X. • Keuntungan: waktu scan lebih singkat, yaitu antara 18 hingga 30 detik/irisan.


68

Generasi ke-3: Fan Beam



34


69

Konstruksi CT Generasi ke-3


70

Generasi ke-3: Rotasi/Rotasi, Fan Beam • • • • •

Konfigurasi rotasi/rotasi. Berkas sinar-x berbentuk kipas (fan beam). Menggunakan detektor array. Waktu scan 1 detik. Kekurangan: kemungkinan terjadinya ring artefact karena adanya kerusakan kanal detektor.



35


71

Generasi ke-4: Fan Beam + Detector Ring


72

Generasi ke-4: Rotasi/Diam • • • • •

Tabung sinar-X berputar dan detektor diam. Detektor tersusun melingkar berbentuk lingkaran. Sekitar 8000 buah detektor diperlukan. Waktu scan 1 detik. Kerugian: harga mahal, dosis radiasi pada pasien lebih tinggi. • Keuntungan: tidak terjadi ring artefact. • Masalah: jarak antara tabung sinar-X dan elemen detektor tidak semuanya sama -> diatasi dengan kalibrasi dan normalisasi saat scan.



36


73

Generasi ke-5: Electron Beam Technique


Electron Beam Computed Tomography Scanner, waktu scan 50ms Digunakan untuk scan jantung.

74

EBT Electron Beam Technique In a conventional CT scanner, the x-ray tube moves round the patient, the xray beam is attenuated by the patient and the differences of the emergent beam are registered by a bank of detectors. This information is digitised and converted into a cross-sectional image. The exposure time for each slice is restricted by the time it takes physically to move the x-ray tube. The Imatron CT Scanner has no x-ray tube but an electron gun which produces a beam of electrons at 130kV which is accelerated along a tube. The beam of electrons is focused, by an electro-magnetic coil, onto a small focal spot on a tungsten ring. This target area is then moved along the ring.



The x-rays are generated by this deceleration process and collimators define them into a fan-beam which ‘sweeps’ the patient. The x-ray density differences are registered by a bank of solid-state detectors and the output is digitised by the Data Acquisition System. Data is stored in the bulk memory and the transferred to disc storage which is then converted into a crosssectional image. There are no moving parts therefore exposure times can be reduced to 50ms per slice. Up to 17 slices per second can be taken, enabling the CT Scanner Unit to image moving structures such as the heart.

37


75

EBT Electron Beam Technique 2


76

EBT Electron Beam Technique 3 Front view

Side view Always Thinking Ahead.


38


77

Generasi ke-6: Spiral atau Helical CT • Teknologi Slip-Ring sekitar tahun 1990-an. • Akuisisi data dilakukan dengan meja yang bergerak sementara tabung sinar-X berputar sehingga gerakan tabung sinar-X membentuk pola spiral terhadap pasien saat dilakukan akuisisi data. • Diterapkan pada konfigurasi rancangan CT generasi ke-3 dan ke-4.


78

Teknologi Slip-ring Aliran listrik disalurkan melalui sejumlah konduktor berbentuk cincin yang disusun secara paralel, sehingga tidak menggunakan kabel lagi.

Ë Rotasi Gantry yang kontinu Ë Merupakan syarat untuk CT-Scan spiral/helical



Non SlipSlip-ring Scanner

SlipSlip-ring Scanner

39


79

Apakah yang dimaksud dengan “Spiral Scan” ? -jawaban 4“C”


Continuously rotating tube/detector system

Continuously generating X-ray

Continuously table feed

Continuously data acquisition

CT Spiral dikelompokkan ke dalam CT generasi ke-6.

80

Akuisisi Data secara kontinu

A

Volum Data

B

Memungkinkan rekonstruksi gambar pada sembarang posisi (baik berurutan atau tumpang tindih) dalam daerah volum yang di-scan.

Jarak antar irisan disebut dengan “Increment”



40


81

Rekonstruksi Gambar Secara Berurutan

}

Tebal Irisan

Increment = Tebal Irisan Ë Tidak ada tumpang tindih Ë Tidak ada sela

{

Increment Always Thinking Ahead.

82

Rekonstruksi Gambar secara Tumpang Tindih SliceThickness

}}

Overlap }

Increment < Slice Thickness Overlap of slices Closer image interval More images created

Ë Ë Ë



{

Increment

41


83

Rekonstruksi Gambar dengan Antara/Sela

}

Slice Thickness

Increment > Slice Thickness Ë Sela antara 2 potongan Ë Images are further apart Ë Jumlah gambar yang lebih sedikit. sedikit.


{

Increment

84

Deep Inspiration

Shallow Inspiration

Standard CT / Slice Imaging Misregistration due to different

respiratory levels between slices

Unable to resconstruct images at arbitrary position

Partial Volume Effect Always Thinking Ahead.


Slice imaging is slow

42


85

Spiral CT / Volume Imaging Scan the whole region of interest in one breath hold No gaps since radiation always transmits the whole volume

Reconstruction of overlapping images without additional dose Retrospective reconstruction of slices in arbitrary position within the scanned volume Always Thinking Ahead.

86

Kuntungan Spiral CT

Melakukan scan satu bagian volume dengan sekali tahan nafas.

Mengurangi efek partial volume.

Tidak ada celah.

Rekonstruksi gambar secara tumpang tindih dapat dilakukan tanpa tambahan dosis radiasi.

Meningkatkan kualitas data untuk keperluan rekonstruksi 3D (3D-rendering)



43


87

Mengapa pemeriksaan Spiral CT dilakukan? Waktu scan yang cepat terhadap valume yang besar. Akuisisi data tanpa celah dlm satu kali tahan napas. Mengoptimalkan penggunaan media kontras. Dapat dilakukan rekonstruksi dengan menentukan nilai increment yang dikehendaki.

CT Scan Konvensional

CT Scan Spiral

Rekonstruksi secara tumpang tindih akan memberikan resolusi sumbu-z yang lebih baik.


88

Kapan pemeriksaan Spiral CT dilakukan?

Pemeriksaan dengan kontras.

Pemeriksaan seluruh bagian tubuh.

Pemeriksaan anak dan pasien trauma, yang memerlukan pemeriksaan secara cepat. Pemeriksaan anatomi yang panjang. Spiral selalu digunakan untuk melakukan 3D postprocessing (contoh CTA) !



44


89

Generasi ke-7: Multi Detector Array CT • Tabung sinar-X memiliki kapasitas panas yang terbatas. Hanya 1% dari energi yang dikonversi menjadi sinar-X. • Dengan detektor multi array maka apabila kolimator dibuka lebih lebar akan diperoleh data proyeksi lebih banyak. Dengan demikian maka penggunaan energi sinar-X lebih efisien. • Pada detektor array tunggal, apabila kolimator dibuka lebih lebar maka akan diperoleh irisan yang lebih tebal yang akan mengurai resolusi spatial. • Masalah: cone beam artefact. • Keuntungan: meningkatkan waktu scan hingga 0,33 detik, resolusi dalam arah sumbu-Z hingga < 0,4 mm, dan dosis radiasi lebih rendah.


90

Pembentukan Gambar - Irisan Sinar-X dilewatkan pada kolimator sehingga hanya menembus bagian potongan aksial dari obyek, yang disebut dengan irisan atau "slice"



45


91

Image Generation - „The Slice“ The geometry of a slice Imagine a line emanating from the top center of one’s head and extending through the body toward the feet. This is known as the “long axis” of the body (like a chicken on a rotisserie). When speaking of scanner geometry, the long axis of the patient is aligned to the “Z” axis of the gantry. It is shown as going into the gantry, the same direction as table feed.The picture above also shows that the X-ray source and detector system rotate around this same Z axis. The slice shown then represents a “transaxial” section, that is perpendicular to the long axis. Always Thinking Ahead.

92

Patient Orientation

Z = caudo-cranial X = left-right Y = posterior-anterior



46


93

Patient Orientation Scanner Geometry All the directions in a CT scanner follow a medical coordinate system rather than a mathematical one.


94

Pembentukan Gambar - “Voxel” Irisan dari suatu obyek terbagi dalam elemen volum yang kecil yang disebut dengan “voxels”. Masing2 voxel memiliki suatu nilai tertentu yang menyatakan atenuasi rata-rata sinar-X oleh obyek pada posisi tersebut.



Elemen gambar dalam bidang 2 dimensi disebut “pixels”

Nilai tersebut dikenal dengan nama Houndsfield Unit.

47


95

Image Generation - The Voxel About the VOXEL or Volume Element... A voxel is an “intangible object”. The operator of the CT has some control over its size, yet never really thinks about. The width of a single detector element plays a primary role in setting the “face” dimensions (length and width) of a voxel. But a voxel is three dimensional, so it also has depth which runs in the direction of the “Z” plane. This dimension is determined by the thickness of the chosen slice - typically a whole number between 1 to 10 mm. This is selected by the operator. In theory, we can calculate the attenuation of the X-ray beam by a single voxel. This is carried out by making many measurements at different angular positions around the object. The “Image Processor” will take this information and process it further until it has a single numerical value that represents the density of a tiny piece of the object scanned. It then assigns a luminance value and position of where this data is to be deposited in an “image matrix”. Always Thinking Ahead.

96

CT image

CT image



48


97

CT image

CT image

CT = Computed tomography - perception in slices. CT provides transverse slice images of the human body in digital form. A coordinate system then results which basically conforms to the anatomical main axes and planes. One volume part of the reconstructed image ( = voxel) is represented by the pixel size in (x, y)-plane and the slice thickness (s) in z-axis.


98

Bilangan CT

Ι0

µ Ι

I=I0 ·exp(- ∫ µ ds) “Line Integral”

∫ µ ds = - ln(I/ I0) Koefisien atenuasi, nilai CT Always Thinking Ahead.


49


99

Dari µ ke Bilangan CT (CT number) The distribution of µ was originally measured. To get rid from the energy dependency of the linear attenuation coefficient, one takes the adventage that most of the coefficients for different tissues show the same relative dependence from energy. S‘o a quotient of the 2 coefficients of 2 tissues should be energy independent in first order. Mr. Hounsfield had the great idea to scale the µ values relative to water by

CT number =

µtissue µwater

µwater

X 1000


100

Houndsfield Unit & Houndsfield Unit didefinisikan :

µ obj − µ wat HU = × 1000 µ wat HU = Houndsfield Unit (-1000 HU sampai +3000 HU) µobj = koefisien pelemahan objek µwat = koefisien pelemahan air



50


101

CT numbers

Ct numbers

•Water has CT number 0 •tissues more dense than water have a positive CT number •tissues less dense than water have a negative CT number. Water

= 0

Bone

= +3000 HU

HU

Air

= - 1000 HU


102

Houndsfield Unit 3000

Blood

60

40

0

Pancreas

Bone

Heart Adrenal Gland

Intestine

Bladder

Water

-100 Mamma -200 Fat -900 -1000

Liver Tumor

Kidneys

Spleen

Air

HU *

Lung

Rule Ruleof ofthumb: thumb: Nilai CT Nilai CTair airadalah adalah00dan danudara udara 1000. The relative values 1000. The relative valuesof ofthe the other tissues are calculated other tissues are calculated relative relativeto tothat thatof ofwater water

HU = Hounsfield Unit, also known as CT number



51


103

CT number flexibility

CT number flexibility

• We can change the appearance of the image by varying the window width and level. • This can spread a small range of CT numbers over a larger range of grey scale values. • This makes it easy to detect very small changes in CT number.


104

Pembentukan Gambar - “Matrix” Nilai koefisien pelemahan radiasi diukur kemudian dikodekan dan ditransfer ke komputer. Oleh komputer akan ditampilkan dalam gambar 2 dimensi yang disebut dengan matriks. 35 36 34 39 33 31 34 33 35 32 31 78 80 85 90



Ukuran matriks gambar : 80x80, 256x256, 512x512

52


105

Image Generation - The Matrix Image Processing This illustration implies the metamorphosis of a voxel into a pixel. It starts out as a three dimensional volume of matter. We measure it’s ability to attenuate X-ray. This information then is sent to a special computer where it is processed along with other data and becomes a “number” that will specify a “light intensity” or luminance value on a CRT display. The area that is filled by this “light” is known as a pixel and it has only two dimensions - length and width. A group of pixels placed in a specific order, side by side, result in the image that we see it displayed on the monitor. Always Thinking Ahead.

106

Pembentukan Gambar - A>D>A* Nilai masing2 elemn matriks dikonversikan kedalam gambar hitam putih sesuai dengan grey scale-nya. 35 36 34 39 33 31 3433 35 32 31 78 80 85 90

*Analog - Digital - Analog Always Thinking Ahead.


53


107

Image Generation - A>D>A* Data Acquisition, Processing and Image Display Here the process continues, hiding the magic of “image reconstruction”, but revealing the desired end result - an image. This slide also illustrates the conversion process that takes place, beginning with the acquisition of data in the gantry, where it is an analog signal. The “data” measured can be very small in amplitude. In order to maintain it’s integrity, it is converted to a digital value before it is sent over to the image processor (SMI5). The SMI5 uses this digital data to calculate an image. But before an image can be displayed on a monitor, it is necessary to convert the data back into analog form.


108

Image matrix size affects resolution

8 by 8 looks like square!

25 by 25 this is closer to a circle!



54


109

Image matrix size affects resolution The image matrix mostly is a 1024x1024 matrix, which represents the object, measured in the scan field. The FOV ( field of view), selected by the operator, shows either the full scan field or just a part of it. The FOV is displayed with the image matrix. When the FOVis equal to the scan fiel, e.g. 500mm, than each pixel of the matrix represents approx. 1mm2 of the FOV (> pixel size ~1mm2). With a smaller FOV, e.g. 50mm, the same image matrix is used; therefore the pixel size becomes smaller (> pixel size ~0.1mm2). The result of the variation of the pixel size effects the resolution of the system: the smaller the pixel size, the better.


110

Reconstructed matrix imaging

image matrix: 1024 2 pixel (interpolated)

measured data

preprocessing

backprojection

reconstructed matrix: 512 2 pixel

convolution Always Thinking Ahead.


55


111

Reconstructed matrix The image processor writes all backprojected data into a 5122 matrix, which is the so called reconstructed matrix. It is on most CT- system a 5122 matrix. The pixel size, mentioned on previous page, is a function of the reconstructed matrix. The imager, responsible for windowing and calculation of HU values, is supplied from the backprojector with the 5122 reconstructed matrix. A linear interpolation from 5122 to 10242 pixels is done, before the digital data of the matrix are converted to analog and connected as RGB to the monitor.


112

Displayed matrix

imaging

image matrix: 1024 2 pixel (interpolated)



56


113

Displayed matrix The displayed matrix is on most CT systems a 10242 interpolated matrix. There are some systems working with smaller matrix sizes like 5122 or 3602, especially economic systems, to keep the price low.


114

Penampilan Gambar - Windowing

Hounsfield unit

+3000

Gray scale display White

Window width W

Window center C

0 -1000

Black

CT Windowing

Window width (W): the density range represented within the gray scale. Always Thinking Ahead.


Window center (C): the center of the density range.

57


115

Image Display - Windowing Close the window - I can’t see anything... The human eye can only resolve about 40 shades of gray at best. The human body is made up of tissue containing mostly water, but also calcium, phosphorous, as well as other minerals. However, all soft tissue is not the same, even though it consists of the same basic building blocks. There is a difference in density between muscle and the liver, for example. The same goes for bone - you have a variety of bone density in your body from the cartilage in your nose to the calcium rich bone of your pelvis. All of these organs attenuate X-rays differently, so they appear to us as different gray levels in the reconstructed image. Remember, each shade or intensity value correlates to how effectively an object attenuates X-rays. Bright white means a very effective attenuater, such as dense bone. Black means the opposite or virtually no attenuation, such as air. Our image display system is designed to display no more than 256 gray shades at one time, but the HU scale extends from -1000 to 3096 or 4096 values in total! You could say that each value corresponds to a unique intensity or “gray level”. When a doctor is looking for details in the liver, he may be looking for small density changes. In order to see these small changes, we use a device called “windowing”. Since the liver only represents a very small portion of the HU scale, we choose a “center value” that corresponds to about the mean value of that organ and “open” the window just enough to see the desired detail. The maximum number of gray shades will not exceed 256 regardless of the window setting!


116

Pengaruh Windowing pada gambar Nilai densitas CT didefinisikan berkisar antara -1000 hingga +3000, tetapi mata manusia paling baik hanya dapat membedakan 30 - 40 tingkat terang gelap saja. Lung Window



Mediastinum Window

Dengan demikian pengatruan pilihan windwos harus disesuaikan dengan struktur yang akan dilihat.

58


117

Image Display - Windowing Narrow Window Width

Lung Window

Lebar window Sempit : Gambar memiliki tingkat kontras yang baik, tetapi struktur diluar daerah window tidak terlihat dengan jelas. Always Thinking Ahead.

Broad Window Width

Mediastinum Window

Lebar window Luas : Perbedaan densitas yang kecil akan nampak homogen dan mungkin akan tersamar/tidak terlihat.

118

Teknik Double Window Hounsfield Narrowunit Window Width +3000

Gray Width scale Broad Window display

White Window 1 Lung Window

Mediastinum Window

0 Black

Window 2 -1000

CT Windowing

Digunakan untuk menampilkan dua jenis jaringan yang memiliki nilai densitas sangat berbeda, seperti paru2 & mediastinum. Always Thinking Ahead.


59


119

Hasil Penggunaan Double Windows Narrow Window Width

Untuk melihat paru2 dan Mediastinum Window mediastinum secara bersamasama dalam satu gambar.

Lung Window


*Penggunaan Double window tidak direkomendasikan untuk diagnosa.

120

Pengaruh Windowing pada Gambar Narrow Window Width

Double Window

Gambar 1: Paru2, dinding toraks & mediastinum terlihat. Always Thinking Ahead.


Lung Window

Gambar 2: Hanya paru2 yang terlihat.

Mediastinum Window

Gambar 3: Hanya dinding thorax & mediastinum Yang dapat terlihat.

60


121

Extended CT Scale Secara normal bilangan CT yang dapat diukur berkisar -1024 sampai +3071, tetapi dengan SOMATOM Plus 4 dapat diperluas (x10) dari -10240 sampai +30710 untuk dapat menampilkan benda logam. Sehingga dimungkinkan untuk menampilkan bilangan CT yang nyata tidak tergantung dimana dan bagaimana window diposisikan.

Post operative femural head replacement – nilai bilangan CT = 6000 HU Always Thinking Ahead.

122

Extended CT Scale What’s the real value? Besides just looking at an image to gain information, doctors will often request a quantitative measurement of an area of interest on the image. This can be performed by drawing an “ROI” (region of interest) over the object to be measured. The computer software then evaluates all of the data confined to that area and reports several statistics about it. The key one is called the mean value and it is expressed as HU’s. This is a representation of the X-Ray absorption characteristic of that object expressed in a numerical value. Remember, we normally only have about 4071 values to represent all objects in the human body. Also realize that anything very dense is displayed as white and if the density of that object is beyond 3071 HU - it is still going to be displayed at the maximum brightness level. In addition, if a quantitative measurement is taken - the result will be false, because the scale ends at 3071! This is where the value of being able to multiply the normal CT scale by a factor of ten comes into play. We now have over 30,000 values to choose from making very high density objects visible and measurable. This feature of expanding the Hounsfield Unit scale is also available on the Somatom AR product line. Note: Extended windowing is only useful if the data is there to begin with. Depending on scanning parameters chosen and limitation of the scanning equipment, it may not be possible to accurately represent extremely dense objects.



61


123

POKOK BAHASAN


124

POKOK BAHASAN



62


125

POKOK BAHASAN


126

POKOK BAHASAN



63


127


128



64

Computed Tomography Scanner (CT-Scan) (Pemindai Tomografi Komputer)

Recommend Documents