1 computed tomography scanner (ct-scan) pemindai...

PT. SIEMENS INDONESIA

(c) Daniel Kartawiguna, 2006 [email protected] 1

1

Always Thinking Ahead.

1


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

DISAJIKAN DALAM RANGKA PELATIHAN CT SCAN DOSEN ATEM

JAKARTA, 19 JUNI 2006

2


2


MATERI PELATIHAN

I. Dasar Tomografi KomputerII. Sistem Tomografi Komputer SpiralIII.Sistem Tomografi Komputer Multi

Irisan (MSCT)IV.Pemeliharaan dan PerbaikkanV. Aplikasi Klinis MSCT



3


3


I. Dasar-dasar Tomografi Komputer

4


4


POKOK BAHASAN1.Tomografi Komputer dan Sejarah

Perkembangannya2.Prinsip Tomografi Komputer3.Komponen2 Tomografi Komputer4.Sistem Tomografi Komputer yang

dikembangkan oleh SIEMENS5.Kalibrasi6.Pemeriksaan Kualitas Gambar

Dasar-dasar Tomografi Komputer



5


5





Dasar-dasar Tomografi Komputer

6


6


PenemuanPenemuan SinarSinar--XXSinar-X ditemukan pada

tahun 1895 oleh seorang ahli

fisika Jerman yang bernama:

Wilhelm Conrad Röntgen



7


7


Keterbatasan:• Gambar yang dihasilkan merupakan superimposisi

dari obyek yang diamati. • Tidak dapat menggambarkan jaringan lunak.

Sinar-X memungkinkan orangpertama kali untuk melihatstruktur dari tubuh manusia

tanpa melakukanoperasi/pembedahan.

8


8


Pada tahun 1972, Hounsfield dan Ambrose yang bekerja di Central Research Labs. of EMI, Ltd di Inggris menghasilkangambar klinis pertamadengan CT Scan.

CT memberikan solusi terhadapketerbatasan ini...



9


9


memungkinkanpertama kali kita dapatmelihat:

• Memerlukan waktu yang lama.

• Resolusi perlu ditingkatkan.

GambarGambar CT yang CT yang pertamapertama

- Tomographic atau“Slice” anatomy

- Perbedaan densitas

Keterbatasan:

10


10


CT-Scanner Siemens yang pertama

Waktu scan 7 min, matrix gambar 80x80 Pixel, scan field 25 cm, resolusi spatial 1,3 mm (4LP/cm)

SIRETOM (1974)



11


11


Perkembangan kualitas gambar CTDari tahun 1972 - 2000

SIRETOM (1974)

SOMATOM Plus 4 UFC(1996)

12


12


Ringkasan Sejarah PerkembanganTomografi Komputer1917 J.H. Radon: Transformasi Radon, gambar dari

obyek yang tidak diketahui dapat digambarkan dari proyeksinya.

1963 A.M. Cormack: mengembangkan teknik untukmenentukan distribusi penyerapan tubuh manusia.

1972 G.N. Hounsfield dan J. Ambrose: menghasilkan gambar CT pertama kali untuk keperluan klinis.

1974 60 unit CT terpasang untuk pemeriksaan kepala1975 First whole body scanner in clinical use1979 Hounsfield dan Cormack dianugerahi hadiah Nobel1989 Spiral CT1998 Multislice CT2000 > 30000 clinical CT installations



13


13





14


14


Prinsip Dasar Pemindai Tomografi KomputerPelemahan intensitas sinar-X oleh obyek yang ditembus oleh sinar-X tersebut.

Berkas sinar-X yang menembussuatu obyek akan mengalamipelemahan (kehilangan energi) yang diakibatkan oleh:

Penyerapan oleh obyekPenyebaran

Sinar-X datang Sinar-X Diteruskan

SINARSINAR--X DIHAMBURKANX DIHAMBURKAN

JARINGANTUBUH

Penyerapan oleh jaringan adalah sebanding dengan densitasnya

JaringanDengan

Kepadatanrendah

Jaringandengan

kepadatanyang tinggi

PERLEMAHAN LEBIH BESAR PERLEMAHAN LEBIH KECIL



15


15


Prinsip Dasar Pengukuran

I=I0 ·exp(- ∫ µ ds)

“Line Integral”∫ µ ds = - ln(I/ I0)

Koefisien atenuasi, nilai CT

Ι0

Ι

µ

16


16


Proyeksi (Projection)



17


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


18


Generasi CT-Scan



19


19


Prinsip dasar akuisisi data

Parallel beam geometry(1st generation scanner)

Tabung sinar-X

Berkas sinar-X

Detektor sinar-X

Obyek pemeriksaan

20


20


Raw data (Data Mentah)

Sequence ofattenuation profiles

time(projections)

attenuation profile(channels)

1. projection



21


21


time(projections)

2. projection




22


22


time(projections)

Full rotation → CT raw data






23


23


• How do we reconstruct the image?

• One could solve more than 260.000 unknown µ by measuring the attenuation in about 1.400 readingswith 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.

REKONSTRUKSI

24


24


Rekonstruksi Gambar

Raw data CT image

Dasar Matematika:Radon (1917)

Dalam praktek:a) Fourier method

atau

b) filteredbackprojection



25


25


Unfiltered Backprojection

Scanned object

Reconstructed object

using projection(s)12348163264128

nice reconstruction,

but not very accurate...

26


26


Solution: Filter

Attenuation profileof a cylinder

Filteredattenuation profile

Filter:convolution kernels



27


27


Filtered Backprojection

Scanned object This

reconstructionlooks much

better ...

Reconstructed object

using projection(s)12348163264128

28


28


Convolution Kernel

sharp kernel

smooth kernel



29


29


Convolution kernel

30


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.



31


31


Convolution Kernel (lanjutan)

• Ketajaman

• Noise

• Batas

Not independent:sharp image → more noiseless noise → smoother image

Kernel akan menentukan

32


32


Simple back-projection w/o and with convolution



33


33


Simple back-projection w/o and withconvolution

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


34


Perangkat Keras Sistem CT-Scan

Bagian luar...

Konsol

Komputer

Meja

Gantry

Generator• Generator• Konsol• Komputer

• Gantry• Meja



35


35


Bagian dalam ...

• Tabung

Tabung

Detektor

• Detektor

* Data Acquisitions System (Sistem Akuisisi Data)

DAS• DAS*

Perangkat Keras Sistem CT-Scan

36


36


CARA KERJA PESAWAT CT SCAN

Rekonstruksi + post processingRekonstruksi + post processing

DetektorDetektor

Sumber Radiasi Sinar-XSumber Radiasi Sinar-X



37


37


Overview over a CT system

Image ProcessingImage Processing

38


38


Overview over a CT systemA 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.



39


39


Blockdiagram image processor

40


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 pre-processor.



41


41


Blockdiagram pre- processor

42


42


Blockdiagram pre- processor

A pre- processor 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.



43


43


44


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 theADC, 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 bitto 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), thePGA 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.



45


45


Offset correction

46


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 slightlydifferent 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.



47


47


Logarithmation

48


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 calculationof 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.



49


49


Normalization

50


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



51


51


Calibration

52


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.



53


53


Channel Correction

54


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 pre-processing.

Channel correction includes:

- Beam hardening correction

- Cosine correction

- Channel coefficient correction and

- Water scaling



55


55


Beam – Hardening Correction

56


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.



57


57


Cosine - Correction

58


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 cosine-value of the corresponding fan angle „alpha“.

The measured attenuation is multiplied with the table values.



59


59


Channel Coefficient Correction

60


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 VoltageSlice ThicknessObject Attenuation (Head or body)

Correction tables:The sensitivity compensation is done with values which are determined

during the tune-up.



61


61


Water Scaling

62


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



63


63


Generasi Tomografi Komputer

“Generasi” digunakan untuk membedakan rancangankonfigurasi tabung-detektor pada CT.

64


64


Generasi ke-1: Pencil Beam



65


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 harusdikelilingi oleh kantong berisi air.

• Kristal NaI yang digunakan sebagai detektor memiliki waktu“afterglow” yang nyata.

• Keuntungan: pengaruh hamburan radiasi pada detektorditiadakan karena berkas sinar-X yang berbentuk pensil.

66


66


Generasi ke-2: Partial Fan Beam



67


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 penambahanfilter dasi kupu-kupu pada jendela tabung sinar-X.

• Keuntungan: waktu scan lebih singkat, yaitu antara 18 hingga 30 detik/irisan.

68


68


Generasi ke-3: Fan Beam



69


69


Konstruksi CT Generasi ke-3

70


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.



71


71


Generasi ke-4: Fan Beam + Detector Ring

72


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 dannormalisasi saat scan.



73


73


Generasi ke-5: Electron Beam Technique

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

74


74


EBT Electron Beam TechniqueIn a conventional CT scanner, the x-ray tube moves round the patient, the x-ray 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 memoryand the transferred to disc storage which is then converted into a cross-sectional 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.



75


75


EBT Electron Beam Technique 2

76


76


EBT Electron Beam Technique 3

Front

view

Side

view



77


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 gerakantabung sinar-X membentuk pola spiral terhadap pasiensaat dilakukan akuisisi data.

• Diterapkan pada konfigurasi rancangan CT generasi ke-3 dan ke-4.

78


78


Teknologi Slip-ringAliran listrik disalurkan melalui sejumlah konduktorberbentuk cincin yang disusun secara paralel, sehingga tidak menggunakan kabel lagi.

Rotasi Gantry yang kontinuMerupakan syarat untuk CT-Scan spiral/helical

Non SlipNon Slip--ring ring ScannerScanner

SlipSlip--ring Scannerring Scanner



79


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


80


Memungkinkan rekonstruksigambar pada sembarang posisi(baik berurutan atau tumpangtindih) dalam daerah volum yang di-scan.

Jarak antar irisan disebutdengan “Increment”

A BVolum Data

Akuisisi Data secara kontinu



81


81


{Increment

}Tebal Irisan

Increment = Increment = TebalTebal IrisanIrisanTidakTidak adaada tumpangtumpang tindihtindihTidakTidak adaada selasela

Rekonstruksi Gambar Secara Berurutan

82


82


{Increment

}

}}Overlap

}SliceThickness

Rekonstruksi Gambar secara TumpangTindih

Increment < Slice ThicknessIncrement < Slice ThicknessOverlap of slicesOverlap of slicesCloser image intervalCloser image intervalMore images createdMore images created



83


83


Increment > Slice ThicknessIncrement > Slice ThicknessSelaSela antaraantara 2 2 potonganpotonganImages are further apartImages are further apartJumlahJumlah gambargambar yang yang lebihlebih sedikitsedikit..

Rekonstruksi Gambar dengan Antara/Sela

{Increment

}Slice Thickness

84


84


Standard CT / Slice ImagingStandard CT / Slice ImagingDeep InspirationDeep Inspiration Shallow InspirationShallow Inspiration

MisregistrationMisregistration due to differentdue to differentrespiratory levels between slicesrespiratory levels between slices

Partial Volume EffectPartial Volume Effect

Unable to Unable to resconstructresconstruct images atimages atarbitrary positionarbitrary position

Slice imaging is slowSlice imaging is slow



85


85


Scan the whole region of Scan the whole region of interest in one breath hold interest in one breath hold

Reconstruction of overlapping Reconstruction of overlapping images without additional doseimages without additional dose

Retrospective reconstruction Retrospective reconstruction of slices in arbitrary position of slices in arbitrary position within the scanned volumewithin the scanned volume

No gaps since radiation alwaysNo gaps since radiation alwaystransmits the whole volumetransmits the whole volume

Spiral CT / Volume ImagingSpiral CT / Volume Imaging

86


86


Kuntungan Spiral CT

Melakukan scan satu bagian volume dengansekali tahan nafas.

Mengurangi efek partial volume.

Tidak ada celah.

Rekonstruksi gambar secara tumpang tindihdapat dilakukan tanpa tambahan dosis radiasi.

Meningkatkan kualitas data untuk keperluanrekonstruksi 3D (3D-rendering)



87


87


Mengapa pemeriksaan Spiral CT dilakukan?

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

CT Scan SpiralCT Scan Konvensional

Akuisisi data tanpa celah dlm satu kali tahan napas.Waktu scan yang cepat terhadap valume yang besar.

Mengoptimalkan penggunaan media kontras.Dapat dilakukan rekonstruksi dengan menentukannilai increment yang dikehendaki.

88


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 melakukan3D postprocessing (contoh CTA) !



89


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 lebihefisien.

• Pada detektor array tunggal, apabila kolimator dibuka lebihlebar 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 dosisradiasi lebih rendah.

90


90


Pembentukan Gambar - IrisanSinar-X dilewatkan pada kolimator sehinggahanya menembus bagian potongan aksial dariobyek, yang disebut dengan irisan atau "slice"



91


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.

92


92


Patient Orientation

Z = caudo-cranial

X = left-right

Y = posterior-anterior



93


93


Patient Orientation

Scanner Geometry

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

94


94


Irisan dari suatu obyekterbagi dalam elemenvolum yang kecil yang disebut dengan“voxels”.

Masing2 voxel memilikisuatu nilai tertentuyang menyatakanatenuasi rata-rata sinar-X oleh obyekpada posisi tersebut.

Elemen gambar dalambidang 2 dimensi disebut“pixels”

Pembentukan Gambar - “Voxel”

Nilai tersebut dikenaldengan namaHoundsfield Unit.



95


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”.

96


96


CT imageCT image



97


97


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.

CT image

98


98


Bilangan CT

I=I0 ·exp(- ∫ µ ds)

“Line Integral”∫ µ ds = - ln(I/ I0)

Koefisien atenuasi, nilai CT

Ι0

Ι

µ



99


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


100


Houndsfield Unit

Houndsfield Unit didefinisikan :

1000×−

=wat

watobjHUµ

µµ

HU = Houndsfield Unit (-1000 HU sampai +3000 HU)µobj = koefisien pelemahan objekµwat = koefisien pelemahan air



101


101


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 HU

Bone = +3000 HU

Air = - 1000 HU

Ct numbers

102


102


Houndsfield Unit

Water

Mamma

Air

Bone

Spleen

Fat

Pancreas

Lung

Kidneys

AdrenalGland

Blood

Heart

Liver

Intestine

Tumor

Bladder

3000

60

40

0

-100

-200

-900-1000

HU *

Rule of thumb: Nilai CT air adalah 0 dan udara1000. The relative values of the other tissues are calculated relative to that of water

Rule of thumb: Nilai CT air adalah 0 dan udara1000. The relative values of the other tissues are calculated relative to that of water

HU = Hounsfield Unit, also known as CT number



103


103


CT number flexibility

• We can change the appearance of the image by varying thewindow 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.

CT number flexibility

104


104


Pembentukan Gambar - “Matrix”

35 36 3934 3331 34 33 35 3231 80 85 9078

Nilai koefisien pelemahan radiasi diukur kemudiandikodekan dan ditransfer ke komputer.Oleh komputer akan ditampilkan dalam gambar 2 dimensi yang disebut dengan matriks.

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



105


105


Image Generation - The Matrix

Image Processing -

This illustration implies the metamorphosis of a voxelinto 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.

106


106


Pembentukan Gambar - A>D>A*

Nilai masing2 elemn matriks dikonversikan kedalamgambar hitam putih sesuai dengan grey scale-nya.

35 36 3934 3331 3433 35 3231 80 85 9078

*Analog - Digital - Analog



107


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


108


Image matrix size affects resolution

8 by 8looks likesquare!

25 by 25this is closerto a circle!



109


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


110


Reconstructed matrix

measured data

preprocessing

convolution

backprojection

reconstructed

matrix: 512 2 pixel

imaging

image matrix:

1024 2 pixel

(interpolated)



111


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


112


Displayed matrix

imaging

image matrix:

1024 2 pixel

(interpolated)



113


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


114


Penampilan Gambar - Windowing

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

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

Hounsfieldunit

+3000

-1000

0

Window width W

Window center C

Gray scaledisplay

White

Black

CT Windowing



115


115


Image Display - WindowingClose 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


116


Pengaruh Windowing pada gambarNilai densitas CT didefinisikan berkisar antara -1000 hingga +3000, tetapi mata manusia paling baik hanya dapatmembedakan 30 - 40 tingkat terang gelap saja.

Dengan demikian pengatruan pilihan windwosharus disesuaikan dengan struktur yang akandilihat.

Lung Window Mediastinum Window



117


117


Image Display - Windowing


Narrow Window Width Broad Window Width

Lebar window Sempit : Gambar memiliki tingkat kontrasyang baik, tetapi struktur diluardaerah window tidak terlihatdengan jelas.

Lebar window Luas : Perbedaan densitas yang kecil akan nampak homogendan mungkin akantersamar/tidak terlihat.

118


118


Teknik Double Window


Narrow Window Width Broad Window WidthHounsfieldunit

+3000

-1000

0

Window 1

Gray scaledisplay

White

Black

CT WindowingWindow 2

Digunakan untuk menampilkan dua jenis jaringanyang memiliki nilai densitas sangat berbeda, sepertiparu2 & mediastinum.



119


119


Hasil Penggunaan Double Windows


Narrow Window Width

Untuk melihatparu2 danmediastinumsecara bersama-sama dalam satugambar.

*Penggunaan Double window tidak direkomendasikan untukdiagnosa.

120


120


Pengaruh Windowing pada Gambar

Narrow Window Width

Lung Window Mediastinum WindowDouble Window

Gambar 2: Hanyaparu2 yang terlihat.

Gambar 1: Paru2, dindingtoraks & mediastinum terlihat.

Gambar 3: Hanya dindingthorax & mediastinum Yang dapatterlihat.



121


121


Extended CT Scale

Secara normal bilangan CT yang dapat diukur berkisar -1024 sampai +3071, tetapi dengan SOMATOM Plus 4 dapatdiperluas (x10) dari -10240 sampai +30710 untuk dapatmenampilkan benda logam. Sehingga dimungkinkan untukmenampilkan bilangan CT yang nyata tidak tergantung dimanadan bagaimana window diposisikan.

Post operative femuralhead replacement – nilaibilangan CT = 6000 HU

122


122


Extended CT ScaleWhat’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.



123


123





124


124







125


125





126


126







127


127


128


128


1 computed tomography scanner (ct-scan) pemindai...

Documents