The Study Of Image Registration And Position Variability In KV Cone-beam Computed Tomography Image-guided Radiotherapy For Hepatic Tumor

Liver cancer in men is the fifth most frequently diagnosed cancerworldwide and the second most frequent cause of cancer death. In women, it isthe seventh most commonly diagnosed cancer and the sixth leading cause ofcancer death. Besides, liver is a frequent site for progression of malignanttumors and are often an indicator of advanced disease. Radiotherapy is animportant modality of non-surgical treatment. With the recent development ofradiotherapy facilities and computer technology, further understanding ofradiobiology, radiotherapy has been evolved remarkably in multi-modalitytreatment for hepatic tumor.The tolerance of the whole liver to radiation is very poor. Radiation-induced liver disease is a common subacute radiation-induced symptom ofliver toxicity. The occurrence of radiation-induced liver disease has beenclosely related to the radiation dose to the whole liver, with the probability of5%after dose of25Gy, and50%after dose of40Gy. Overseas researchfindings show that whole-liver radiotherapy can be delivered to livermetastases patients without hepatic dysfunction, with the probability of5%after dose up to30Gy. However, the radiosensitivity of hepatocarcinoma cellis similar to poorly differentiated squamous cell carcinoma, the fractionationschedule should be60Gy in30daily fractions, delivered6weeks, withcurative intent. For this reason, the radiation tolerance dose of hepatic cell isfar less than that required to eradicate tumor. The use of conventionalradiotherapy was limited because it cannot balance tumor control rate andnormal tissue complication probability. Three dimensional conformalradiotherapy (3D-CRT) is a precise treatment which involves shaping of theprofile of each radiation beam to fit the profile of the target, it allows dose-escalation of target while sparing normal liver.3D-CRT has becomea new promissing modality of radiotherapy. The type of uncertainties in3D-CRT for hepatic tumor include the motion of target, setup error and so on.Image-guided radiation therapy (IGRT) which integrates radiotherapy unitsand imaging equipments can conform the position of target and critical organs,monitor the motions of tumor and normal tissues, adjust all sources of errors.IGRT has improved the accuracy of radiotherapy.Kilovoltage cone-beam computed tomography (kV CBCT) integratedwith linear accelerator is one of the most common volumetric image guidance.Compared with two dimensional images, CBCT provides more informationabout3D soft tissues. Either the entire CBCT image or region of interest canbe registered with planning CT, in these ways, the setup would be guided morebetter. In the whole process of kV CBCT IGRT, image registration isespecially important, the results depend on different image registrationmethods. Therefore, at first, different image registration methods werecompared in this dissertation, then interfraction variability of position inhepatic tumor radiotherapy was investigated and the importance of IGRT wasdiscussed.PartⅠ Comparing different image registration methods in kV CBCTimage-guided radiotherapy for hepatic tumorObjective: To compare different methods of registering kV CBCT scanwith planning CT scan.Methods: A total of15patients with hepatic tumor were investigated.For each patient, a CBCT scan was obtained in the treatment positionimmediately before radiotherapy, once to three times per week,121CBCTscans in all. In the offline study, each CBCT was registered with relativeplanning CT using one manual and four automatic image registration methods.In the manual registration method, the live contour was used as a surrogate forimage registration, the liver at the right side of vertebral body and in the front of vertebral body were registered respectively, took the mean of the tworegistration results as the result of the whole liver. Repeated it four times,took the mean of the results as the result of manual registration then. Fourautomatic match methods, including routine soft-tissue registration, routinebone registration, automatic liver registration and automatic vertebral bodyregistration, were performed using image registration sofeware. The sofewareconsists of four factors: registration template, registration range, registrationparameter and registration axis. The template of routine soft-tissue andautomatic liver registration was automatic soft-tissue registration template, thetemplate of routine bone and automatic vertebral body registration wasautomatic bone registration template. The range of routine soft-tissue and boneregistration was all of the stuctures in the CBCT scan (7.5cm superior/inferiorto the isocenter), the range of automatic liver registration was the whole liver,and the range of automatic vertebral body registration was the vertebral bodyat the level of liver. The parameter and axis of the four automatic imageregistration methods were uniform. The parameter was Thorax3D. The axiswere X (left-right), Y (superior-inferior) and Z (anterior-posterior). Not to quitthe registration procedure, continuously registered CBCT scan and planningCT three times using the same automatic registration method, then the resultswere compared to test the the stability of the sofeware. Repeated to registerimage three times using every automatic registration method, the intervalbetween registrations was about a month, four times per method in all. Theresults were compared then, the reproducibility of automatic imageregistration method was defined as the maximum differences were no morethan3mm. In the last, the mean of automatic registration results was comparedwith the mean of manual registration results to determine which automaticimage registration method was the most similar to manual registration.Results:1Registering kV CBCT scan with planning CT scan using manul methoddemanded10to15minutes, the efficiency of image registration had yet to beimproved. 2The reproducibility of routine soft-tissue and bone registration werebest, the range of difference among the four times image registration resultswould not exceed1mm. The reproducibility of automatic vertebral bodyregistration was better than automatic liver registration.3The registration range affected automatic registration result.4The automatic liver registration result was the most similar to manualregistration result, the percentages of registration results had an absolutedifference no more than3mm in left-right, superior-inferior and anterior-posterior directions were84.3%,77.7%and92.6%. There still be largerdifferences in left-right, superior-inferior and anterior-posterior directions(13.4mm,12.6mm and14.9mm).Conlusions: In order to improve the efficiency of image registration, thisstudy recommends to use automatic liver registration. As registration rangeaffects registration result, the individual volumetric region of interest (ROI)should be determined according to liver position in each image registration. Asthere still be larger difference using this method, after automatic registration,the result should be adjusted carefully according to live contour, especiallydiaphragmatic dome contour in planning CT.PartⅡ Application of kV CBCT to research interfraction variability ofposition in hepatic tumor radiotherapyObjective: To evaluate the interfraction variability of position for hepatictumor patients using kV CBCT image-guided radiotherapy.Methods: A total of15hepatic tumor patients were investigated. Foreach patient, a CBCT scan was obtained in the treatment position immediatelybefore radiotherapy, once to three times per week,121CBCT scans in all. Inthe offline study, each CBCT was registered with relative planning CT usingliver contour and vertebral body contour as surrogate. The liver and vertebralbody registration results were defined as interfraction variability of liverposition and vertebral body position respectively, the difference between the results was defined as the variability of liver position relative to vertebral body.The spatial variabilities were evaluated then.Results:1The interfraction variability of liver position was1.0(-16.0~8.0) mm,-2.1±8.4mm,-1.2±2.3mm in left-right (X), superior-inferior (Y) and anterior-posterior (Z) directions, respectively. Of which, the largest amplitude of livermotion was in Y direction (median6.3mm,90th percentile14.9mm,0～19.3mm), the second was in X direction (median2.8mm,90th percentile6.6mm,0～16.0mm), and the smallest was in Z direction (median1.5mm,90th percentile4.4mm,0～7.6mm), there were statistically significantdifferences between groups.2The interfraction variability of vertebral body position was-1.0(-16.0~8.0) mm,-3.4±8.1mm,-2.0(-10.0~3.0) mm in X, Y and Z directions,respectively. Of which, the largest amplitude of vertebral body motion was inY direction (median6.0mm,90th percentile14.0mm,0～18.0mm), the secondwas in X direction (median3.0mm,90th percentile7.4mm,0～16.0mm), andthe smallest was in Z direction (median2.0mm,90th percentile7.0mm,0～10.0mm), there were statistically significant differences between groups.3The interfraction variability of liver position relative to vertebral bodywas1.5±2.5mm,1.2±6.7mm,1.1(-2.9~10.3) mm in X, Y and Z directions,respectively. Of which, the largest amplitude of liver relative to vertebral bodymotion was in Y direction (median4.8mm,90th percentile10.8mm,0.1～19.3mm), the second was in X direction (median2.0mm,90th percentile4.4mm,0～8.6mm), and the smallest was in Z direction,(median1.4mm,90thpercentile3.6mm,0～10.3mm), there were statistically significant differencesbetween groups.4If patients position was aligned using vertebral body registration, theamplitude of liver motion may increase, the possibilities in left-right,superior-inferior and anterior-posterior directions were39.7%,42.1%and43.0%.5Without image guidance in hepatic tumor radiotherapy, the margins for compensation of CTV-PTV would have been8.6mm,17.4mm,4.7mm inleft-right, superior-inferior and anterior-posterior directions to ensure95%radiation prescription dosage can deliver to95%CTV.Conlusions: The interfraction variability of liver position relative tovertebral body is an important source of geometric uncertainty. Using vertebralbody as surrogate is not suitable for hepatic tumor IGRT. Without IGRT, themargins for compensation of CTV-PTV would have been8.6mm,17.4mm,4.7mm in left-right, superior-inferior and anterior-posterior directions.