VOL. 88-B, No. 11, NOVEMBER 2006 1513

Very low-dose computed tomography for planning and outcome measurement in knee replacement

THE IMPERIAL KNEE PROTOCOL

J. Henckel, R. Richards, K. Lozhkin, S. Harris, F. M. Rodriguez y Baena, A. R. W. Barrett, J. P. Cobb

From Imperial College, London, England

�

J. Henckel, MRCS, Honorary Research Fellow

�

J. P. Cobb, MCh, FRCS, ProfessorDepartment of Musculoskeletal Surgery, Imperial College London, Charing Cross Hospital, Fulham Palace Road, London W6 8RF, UK.

�

R. Richards, PhD, Senior Research FellowDepartment of Medical Physics & Bio-Engineering, Malet PlaceUniversity College London, Gower Street, London WC1E 6BT, UK.

�

K. Lozhkin, PhD, Principle PhysicistDepartment of Medical PhysicsUniversity College London Hospitals NHS Trust, 250 Euston Road, 1st Floor North Wing, London NW1 2PQ, UK.

�

S. Harris, PhD, Project Manager

�

A. R. W. Barrett, PhD, Project ManagerAcrobot Company Limited, The Leathermarket, Weston Street, London SE1 3ER, UK.

�

F. M. Rodriguez y Baena, PhD, LecturerMechatronics in Medicine Laboratory, Department of Mechanical EngineeringImperial College London, Mechanical Engineering, South Kensington Campus, London SW7 2AZ, UK.

Correspondence should be sent to Mr J. Henckel; e-mail: johann@doctors.org.uk

©2006 British Editorial Society of Bone and Joint Surgerydoi:10.1302/0301-620X.88B11. 17986 $2.00

J Bone Joint Surg [Br]

2006;88-B:1513-18.

Received 4 April 2006; Accepted after revision 7 August 2006

Surgeons need to be able to measure angles and distances in three dimensions in the

planning and assessment of knee replacement. Computed tomography (CT) offers the

accuracy needed but involves greater radiation exposure to patients than traditional long-

leg standing radiographs, which give very little information outside the plane of the image.

There is considerable variation in CT radiation doses between research centres, scanning

protocols and individual scanners, and ethics committees are rightly demanding more

consistency in this area.

By refining the CT scanning protocol we have reduced the effective radiation dose

received by the patient down to the equivalent of one long-leg standing radiograph.

Because of this, it will be more acceptable to obtain the three-dimensional data set

produced by CT scanning. Surgeons will be able to document the impact of implant

position on outcome with greater precision.

The ability to plan a knee replacement pre-cisely and measure the outcome accurately inboth conventional and computer-assistedorthopaedic surgery is crucial for detailedanalysis of the efficacy of these systems.

In this study we sought the optimal com-puted tomography (CT) radiation dose forboth planning and outcome measurement incomputer-assisted and conventionally per-formed knee replacement. In this field, CT isfast becoming the imaging modality ofchoice.

1-4

The perspective distortion, magnifi-cation errors and orientation uncertaintiesassociated with standard radiographs makethem an insufficiently sensitive tool for bothaccurate pre-operative planning and post-operative measurement of accuracy.

5,6

Radia-tion dose is generally represented as theweighted dose (mSv) received by the radio-sensitive organs. A typical CT scan of the pel-vis gives around 10 mSv,

7

which is equivalentto about 4.5 years of background radiation(average United Kingdom background radia-tion is 2.2 mSv per year).

8

The Perth protocolfor lower limb CT scans gives a dose of 2.7 mSv,

9

whereas in our previous trials we used a proto-col that gave 2.2 mSv.

10

In comparison, the tra-ditional long-leg standing radiograph, used todefine the mechanical axis, gives a dose ofabout 0.7 mSv.

11

The National RadiologicalProtection Board estimates that CT is used in5% of annual radiological investigations but

has a disproportionately high annual collectivedose burden of 40%.

12

The current evidencesuggests a linear relationship between effectiveradiation dose and malignancy, with exposureto 10 mSv giving a 1:2000 risk of radiation-induced malignancy.

13

Although there is no true safe dose of radia-tion, one can facilitate the reduction of effec-tive doses by manipulating and reducing thedose parameters. This study looked at ways ofminimising radiation dose while maintainingimage quality for pre-operative orthopaedicplanning and post-operative assessment inknee replacement.

Materials and Methods

The initial part of this study was performedusing a standard radiological (tissue equiva-lent) phantom pelvis and proximal femora(cadaver human bones in a rubber envelope).Owing to the unavailability of commercialleg phantoms, a simulated phantom of bothlegs was constructed using human cadaverfemora, patellae, tibiae, fibulae and tarsalbones in a polythene envelope filled withwater to represent the soft-tissue envelope ofthe legs.

The imaging was performed using the Sie-mens Sensation 4, four-slice multislice CTscanner (Siemens Medical Solutions, Erlangen,Germany). Both phantoms were placed in thesupine position in the scanner to simulate the

1514 J. HENCKEL, R. RICHARDS, K. LOZHKIN, S. HARRIS, F. M. RODRIGUEZ Y BAENA, A. R. W. BARRETT, J. P. COBB

THE JOURNAL OF BONE AND JOINT SURGERY

pelvis and lower limbs of the human. A scout film (Fig. 1a)was taken between the iliac crest and the feet; this was usedto specify the regions to be imaged (Fig. 1b). The whole

phantom was initially scanned to reproduce a conventionalscan. Subsequent scans were performed, reducing the vol-ume areas imaged to include only the femoral head, 10 cmabove and below the joint line of the knee and 2 cm to 3 cmon either side of the tibiotalar joint.

Following minimisation of the volume imaged, the milli-ampere-seconds (mAs) and collimation width were variedfor each of the three regions (hip, knee and ankle). The kilo-voltage (kV) was incrementally reduced from 140 to 100for each of the three regions, and the mAs varied between120 and 75 at the pelvis and 100 and 30 for both the kneeand the ankle. Collimation width was increased from 4 mmx 1 mm sliced sequentially. For the three regions the

x

,

y

and

z

axes were kept fixed for the duration of the scan to main-tain the relative position of the three regions with respect toeach other.

Image quality and resolution at these different parame-ters were evaluated by two of the authors (RR, SH).

The effective dose in mSv was calculated from actualscans using two commercially-available software packages(CT DOSE; National Board of Health, Herlev, Denmark,and CT-EXPO; Medizinische Hochschule Hannover, Han-nover, Germany) for the different parameters used.

Following optimisation of the scanning protocol (TableI), human subjects were initially scanned using the Sensa-tion 4 (Siemens Medical Solutions) CT scanners, and whenthe 16 and 64 (Siemens Medical Solutions) multislice scan-ners became available this protocol was adapted for theiruse and the effective radiation doses recalculated (Table II).A conventional lower limb trauma splint was applied tostabilise the leg while in the scanner and thus reduce therisk of movement artefact, especially during movement ofthe table, in the regions not exposed to radiation. Followingthe scan the digital imaging and communications in medi-cine (DICOM) data were further checked for presence ofsuch artefacts.

Results

Field reduction.

The phantom studies showed that we wereable to restrict the region scanned to include only the whole

Table I. Calculated doses for the Imperial knee protocol (Sensation four-slice scanner; Siemens Medical Solutions,Erlangen, Germany)

Milliampere-seconds

Effective dose (mSv)

KilovoltageScan length (cm)

Collimation (mm)

Calculation using CT DOSE program*

Calculation using CT-EXPO† program

Area scanned Male Female

Hip 120 80 5 4 x 2.5 0.610 0.5 0.7Knee 120 100 20 4 x 1 0.120Ankle 120 45 5 4 x 2.5 0.0046

Scout film/scanogram

80 35 102.4 4 x 1 0.016 0.016

Total effective dose (mSv) 0.751 0.641 0.841

* National Board of Health, Herlev, Denmark† Medizinsche Hochschule Hannover, Hannover, Germany

Fig. 1a

a) Scout film to demonstrate the Imperial knee protocoland b) the regions scanned.

Fig. 1b

VERY LOW-DOSE COMPUTED TOMOGRAPHY FOR PLANNING AND OUTCOME MEASUREMENT IN KNEE REPLACEMENT 1515

VOL. 88-B, No. 11, NOVEMBER 2006

femoral head, 20 cm at the knee (10 cm on either side of thejoint line) and 5 cm at the ankle (distal tibia and the talus),and this protocol was subsequently used for our patientstudies (Fig. 2).

Milliamperage reduction.

On viewing images of the hip, withsequential reduction of the milliamperage we found that at anintensity of < 80 mAs neither of the observers felt able to iden-tify the centre of the hip with confidence. At the knee, an mAsof 80 was sufficient pre-operatively, with 100 mAs neededpost-operatively (higher mA needed to reduce image artefactfrom the metal components). At the ankle an mAs of 45was necessary to define the centre of the talus reliably.

The collimation width needed depended on the scannerused. In a four-slice scanner the width could be increased to2.5 mm at both the hips and ankles while still providingadequate image quality, whereas at the knee 1.0 mm colli-mation was retained for image definition. In 16- and 64-slice scanners the collimation widths are narrower, but withthe increased number of detectors in these new scanners theeffective doses are further reduced (Table II).

With the reduction in mAs and scanned volume and opti-misation of collimation width the effective dose was0.841 mSv in females and 0.641 mSv in males, and 0.74 mSvin females and 0.54 mSv in males for the new 64-slice scan-ners, while maintaining a sufficient image resolution for ourpurposes. This contributed to an average effective dose tothe hip of 0.7 mSv in females and 0.5 mSv in males, to theknee 0.1 mSv and the ankle 0.005 mSv in both genders(Table I). This procedure is available via the Imperial Col-lege orthopaedic website.

14

Discussion

In this study we looked at the effects of minimising the vol-ume of a patient’s body that is scanned and the dose of radi-ation delivered. For the purposes of measuring limbalignment we chose to use the mechanical axis. Thisrequires that only the centre of the femoral head be clearlyidentified proximally. Finding this point by limited low-dose scanning is simple and reproducible. By reducing thevolume scanned in this particularly radiosensitive region,this protocol almost spares the ovaries in the female andnormally avoids the testes in the male (while recognising

Table II.

Calculated doses for the Imperial knee protocol for the different scanners

Imperial protocol for each scanner Collimation (mm)

Area Kilovoltage Milliampere-seconds

Scan length (cm) Four-slice 16-slice 64-slice

Hip 120 80 5 4 x 2.5 16 x 1.5 32 x 1.2Knee 120 100 20 4 x 1 16 x 0.75 32 x 0.6Ankle 120 45 5 4 x 2.5 16 x 1.5 32 x 1.2

Scout film/scanogram 80 35 102.4 4 x 1 2 x 1 6 x 0.6Total radiation dose (mSv)

Female 0.84 0.83 0.74Male 0.64 0.53 0.54

Fig. 2

Images of the knee showing the scanned regions.

1516 J. HENCKEL, R. RICHARDS, K. LOZHKIN, S. HARRIS, F. M. RODRIGUEZ Y BAENA, A. R. W. BARRETT, J. P. COBB

THE JOURNAL OF BONE AND JOINT SURGERY

A

B D C

B D C

A

Fig. 3

Computer-assisted images showing the planned component positioning.

Fig. 4a

Figure 4a - 3D images of the plan (left) and post-operative scan (right). Figure 4b – The implant position relative to mechanical axes.

Fig. 4b

VERY LOW-DOSE COMPUTED TOMOGRAPHY FOR PLANNING AND OUTCOME MEASUREMENT IN KNEE REPLACEMENT 1517

VOL. 88-B, No. 11, NOVEMBER 2006

the wide normal anatomical variation in the relative posi-tion of these organs).

The reduction in the volume imaged in the region of thefemoral head taken together with the minimisation of mAsin this extremely dose-sensitive region had a substantialimpact in reducing the effective dose (Table I).

Further down the lower limb we found that scanning10 cm either side of the joint line at the knee and 5 cm at theankle, including the talus, was sufficient to identify thenecessary landmarks (knee and ankle centres). The volumesbetween the hip, knee and ankle were not exposed to directradiation. Although the reduction in effective dose achievedby this modification may be modest, any reduction in doseis worthwhile. By reducing the size of the data set, manipu-lation of the models was made faster, producing anunintended benefit from this protocol.

The low mAs used here leads to no significant degrad-ation or reduction in quality of the bone image. The greatvariation in patient shape and size posed no challenge inthis respect, as although this may reduce the quality of thesoft-tissue image, the resolution of the interface betweenbone and soft tissue remains very clearly defined. There-fore, for the purpose of pre-operative planning and post-operative assessment in knee replacement only bony win-dows need be reconstructed.

For pre-operative planning the DICOM files areexported to the planner and segmentation of the bones isperformed. This is followed by detailed planning of theimplant position, with the precise location and orientationof each implant optimised in three dimensions (Fig. 3). Thesurgeon can then choose where to put each component and

track back and forth between images to ensure that thechoice of size and position is correct.

In computer-assisted surgery the plans made from the CTdata are co-registered with points taken in the knee in theoperating theatre. When a satisfactory match is achieved,the operation can be performed, using active or passive sur-gical navigation systems.

Following surgery, detailed and accurate comparisonsbetween the planned and achieved component positionscan be made using both 2D and 3D methods, if a pre-operative plan has been made (Fig. 4a). Co-registration ofthe precisely planned models with surface models from thepost-operative scan gives real measurements of implantposition, enabling the measurement of the accuracy of jointreplacement in all six degrees of freedom, giving both trans-lation and rotation errors in all three planes. Even withouta pre-operative CT-based plan this protocol allows meas-urement of implant position relative to the mechanical axis(Fig. 4b).

The aim of this exercise was to find a compromisebetween radiation exposure and the acquisition of imageswhich are good enough to enable the surgeon to generate areliable 3D data set for planning and assessing kneereplacement procedures.

The dose burden per CT scan is now down, reduced tobetween 0.53 mSv and 0.84 mSv and compares favourablywith traditional long-leg measurement films at 0.7 mSv(Fig. 5).

15-18

By providing an entire 3D data set of the largejoints of the lower limb, it offers significant advantages overlong-leg films for the accurate determination of the positionof implants in three dimensions and provides a permanentmodel of the patient’s anatomy for future reference.

The time needed to position and scan the patient isanother source of concern with both imaging modalities.There are anecdotal reports of problems with patient-positioning when long-leg films are taken and the need forfilm to be repeated. With CT there is a greatly reduced riskof re-imaging, as the scanning time is now measured in sec-onds not minutes. The time spent in processing the pre-operative data in this study is still measured in minutes, notseconds. We expect that as the software is improved thetime and skill needed will be substantially reduced.

1

Post-operative analysis is still time consuming and exper-imental; it being used as a research tool. As the softwaredevelops, it is envisaged that appropriately-trained staffwill perform these analyses routinely. Such analysis willprovide valuable information in poorly functioning orpainful joints and allow detailed assessment to be under-taken before embarking on any further intervention.

Compared with the Perth Protocol

6

we have shown afourfold reduction in effective radiation dose with no com-promise in data quality. Like the Perth protocol theseimages are non weight-bearing, but because the bones maybe measured separately the combined mechanical axes canbe calculated with complete correction for any rotatorymalalignment. Although this study does not compare the

0

1

2

3

4

5

6

7

8

9E

ffec

tive

rad

iati

on

do

se (

mS

v)

Bariu

m en

ema

Perth

pro

toco

l

Old p

roto

col

New p

roto

col

Annual bac

kround

radiat

ion

Antero

posterio

r rad

iogra

ph

(pelv

is to

ankle

)

Fig. 5

Comparison of effective radiation doses for different types of imaging(Barium enema,15 Perth protocol,16 old protocol,17 annual backgroundradiation,18 anteroposterior radiograph (pelvis to ankle)15

1518 J. HENCKEL, R. RICHARDS, K. LOZHKIN, S. HARRIS, F. M. RODRIGUEZ Y BAENA, A. R. W. BARRETT, J. P. COBB

THE JOURNAL OF BONE AND JOINT SURGERY

accuracy of traditional long-leg radiographs with CT scans,this area will be addressed in future work.

No benefits in any form have been received or will be received from a commer-cial party related directly or indirectly to the subject of this article.

References

1. Nabeyama R, Matsuda S, Miura H, et al.

The accuracy of image-guided kneereplacement based on computed tomography.

J Bone Joint Surg [Br]

2004;86-B:366-71.

2. Stockl B, Nogler M, Rosiek R, et al.

Navigation improves accuracy of rotationalalignment in total knee arthroplasty.

Clin Orthop

2004;426:180-6.

3. Siebert W, Mai S, Kober R, et al.

Technique and first clinical results of robot-assisted total knee replacement.

Knee

2002;9:173-80.

4. Amiot LP, Poulin F.

Computed tomography-based navigation for hip, knee, and spinesurgery.

Clin Orthop

2004;421:77-86.

5. Jazrawi LM, Birdzell L, Kummer FJ, Di Cesare P.

The accuracy of computed tom-ography for determining femoral and tibial total knee arthroplastry component rota-tion.

J Arthroplasty

2000;6:761-6.

6. Chauhan SK, Scott RG, Breidahl W, Beaver RJ.

Computer-assisted knee arthro-plasty versus a conventional jig-based technique: a randomised, prospective trial.

JBone Joint Surg [Br]

2004;86-B:372-7.

7. Dawson P.

Patient dose in multislice CT: why is it increasing and does it matter?

BrJ Radiol

2004;77:10-13.

8. Watson SJ, Jones AL, Oatway WB, Hughes JS.

Ionizing radiation exposure ofthe UK population: 2005 Review.

Health Protection Agency

. HPA-RPD-001. http://www.hpa.org.uk/radiation/publications/hpa_rpd_reports/2005/hpa_rpd_001.htm

9. Chauhan SK, Clark GW, Lloyd S, et al.

Computer-assisted total knee replacement:a controlled cadaver study using a multi-parameter quantitative CT assessment ofalignment (the Perth CT Protocol).

J Bone Joint Surg [Br]

2004;86-B:818-23.

10. Jakopec M, Harris SJ, Rodriguez y Baena F, et al.

The first clinical application ofa “hands-on” robotic knee surgery system.

Comput Aided Surg

2001;6:329-39.

11. Hart D, Hillier MC, Wall BF, Shrimpton PC, Bungay D.

Doses to patients frommedical x-ray examinations in the UK: 1995 review, NRPB-R289. London: HMSO,1994.

12. Shrimpton PC, Jones DG, Hillier MC, et al.

Survey of CT practice in the UK. Part2: dosimetric aspects, R249. London, UK: National Radiological Protection Board,1991.

13. National Radiological Protection Board.

Protection of the patient in x-ray com-puted tomography. Doc NRPB 1992;3.

14. Cobb J.

www1.imperial.ac.uk/medicine/people/j.cobb (last accessed 19th Septem-ber 2006).

15. Hart D, Hillier MC, Wall BF, et al.

Doses to patients from medical x-ray examina-tions in the UK – 1995 review, NRPB-R289. London: HMSO 1994.

16. Chauhan SK, Clark GW, Lloyd S, et al.

Computer-assisted total knee replacement:a controlled cadaver study using a multi-parameter quantitative CT assessment ofalignment (the Perth CT Protocol).

J Bone Joint Surg [Br]

2004;86-B:818-23.

17. Jakopec M, Harris SJ, Rodriguez y Baena F, et al.

The first clinical application ofa “hands-on” robotic knee surgery system.

Comput Aided Surg

2001;6:329-39.

18. Watson SJ, Jones AL, Oatway WB, Hughes JS.

Ionizing radiation exposure ofthe UK population: 2005 Review. Health Protection Agency. HPA-RPD-001.