Faculty Bibliography

Purpose: Radiation induced interstitial pneumonitis and late renal dysfunction are major concerns for patients undergoing total body irradiation (TBI). The purpose of this work is to evaluate the dosimetry of VMAT-based TBI plans generated using Varian Eclipse scripting.
Method(s): Three full-body CT datasets (two patients, one anthropomorphic CIRS phantom) were used. An in-house Eclipse script was developed to generate optimized field arrangements using the body contour, user origin, and couch longitudinal travel. Plans consisted of a lower-body AP/PA portion and an upper-body VMAT portion (8 full arcs with 4-isocenters). Treatment plans to 1320 cGy (165 cGy x 8fx) were generated with dose directives: [PTV V100% >=90- 95%; Total lung Dmean <900 cGy; Kidneys Dmean <1100 cGy]. All plans used 6MV photons and were calculated using the AAA algorithm. Upperbody VMAT plan dosimetry was evaluated 'in-phantom' placing 12 OSLDs in different key locations (lung, kidneys, bone, and soft tissue). Additionally, dosimetric verification was performed for the three plans using Varian portal dosimetry, PerFraction(SNC) and ArcCheck(SNC) with a global gamma criterion of 2%/2 mm.
Result(s): Planning objectives were met for the three treatment plans with the following averages: PTV V100% = 94.02%, total lung Dmean = 872.9 cGy, and kidneys Dmean = 1075.8 cGy. The dose deviation between Eclipse and the OSLDs (relative to the prescribed dose) averaged 0.98%, with each individual dose deviation within +/-4%. Dose ranged between 52.5 cGy (lung) and 187.5 cGy (bone) for OSLD measurements. The average passing rate for all 24 fields (8 per plan) was 98.0%, 99.76% and 98.6% for portal dosimetry, PerFraction and ArcCheck respectively. The lowest passing rate of any individual field was 95.4%, 99.0% and 91.8% for portal dosimetry, PerFraction and ArcCheck respectively.
Conclusion(s): Eclipse scripting can assist in creating robust multi-isocentric VMATbased TBI treatment plans to block lungs and kidneys without compromising target coverage. Dosimetric accuracy and deliverability was confirmed using in-phantom OSLD dosimetry, Varian portal dosimetry, PerFraction and Arc-Check verification

BACKGROUND:High b-value diffusion-weighted imaging has application in the detection of cancerous tissue across multiple body sites. Diffusional kurtosis and bi-exponential modeling are two popular model-based techniques, whose performance in relation to each other has yet to be fully explored. PURPOSE/OBJECTIVE:To determine the relationship between excess kurtosis and signal fractions derived from bi-exponential modeling in the detection of suspicious prostate lesions. MATERIAL AND METHODS/METHODS:This retrospective study analyzed patients with normal prostate tissue (n = 12) or suspicious lesions (n = 13, one lesion per patient), as determined by a radiologist whose clinical care included a high b-value diffusion series. The observed signal intensity was modeled using a bi-exponential decay, from which the signal fraction of the slow-moving component was derived ( SFs). In addition, the excess kurtosis was calculated using the signal fractions and ADCs of the two exponentials ( KCOMP). As a comparison, the kurtosis was also calculated using the cumulant expansion for the diffusion signal ( KCE). RESULTS:Both K and KCE were found to increase with SFs within the range of SFs commonly found within the prostate. Voxel-wise receiver operating characteristic performance of SFs, KCE, and KCOMP in discriminating between suspicious lesions and normal prostate tissue was 0.86 (95% confidence interval [CI] = 0.85 - 0.87), 0.69 (95% CI = 0.68-0.70), and 0.86 (95% CI = 0.86-0.87), respectively. CONCLUSION/CONCLUSIONS:In a two-component diffusion environment, KCOMP is a scaled value of SFs and is thus able to discriminate suspicious lesions with equal precision . KCE provides a computationally inexpensive approximation of kurtosis but does not provide the same discriminatory abilities as SFs and KCOMP.

Purpose: TBI treatment at our institution has moved from traditional hand calculation to CT-based planning to incorporate dose heterogeneities and organs at risk dose limits. The main objective of this work is to report our institutional experience with CT-based TBI and to show a comparison with the traditional approach. Methods: Ten patients were CT simulated supine with arms immobilized for lung shielding. Legs are separated to achieve a width similar to umbilicus separation; rice bags were placed between the legs for compensation. Four plans (P1, P2, P3 and P4) were created for each patient, all prescribed at midplane-umbilicus. The first three plans use lateral 15X beams, with head compensation. P1 was planned using a hand calculation. P2 includes heterogeneity corrections and inferior subfield to improve coverage. P3 includes heterogeneity corrections, inferior subfield, and adjustment of field weights to maintain coverage while keeping mean lung doses below 10.5 Gy (prescription dose 12 Gy). P4 uses AP-PA 6X beams. Dose to target (mean, max, D98%, D95%, min), mean lung and liver doses are calculated for all plans; reported doses when unitless and normalized to prescription dose. Results: Coverage of the target (Body-2 cm), indicated by D98% was 84.1 +/- 2.8, 84.7 +/- 3.9, 81.0 +/- 1.8, and 92.2 +/- 1.9 whereas the maximum doses were 123 +/- 5, 135 +/- 4, 129 +/- 4, and 124 +/- 5 for P1, P2, P3, and P4 respectively. The mean relative lung and liver doses were lowest for P3 with values of 87.8 +/- 0.5 and 89.8 +/- 3.4. The largest mean lung dose (12.5 Gy) was observed for P4 plan as expected, showing the necessity of using lung shielding. Conclusion: We are able to achieve target coverage of D98% >80%, keeping the mean lung and liver doses <90% of prescription using optimal arm positioning and subfields. This approach is easy to implement without the complexity of introducing lung shielding required with the use of 6X AP-PAbeams

Purpose: Esophageal cancer patients undergoing radiotherapy usually receive daily or weekly cone-beam CT (CBCT) imaging to verify positioning before treatment. The purpose of this study is to evaluate the reproducibility of texture features extracted from CBCT and its correlation with CT features for their potential use as early predictors of esophageal cancer response during the course of RT. Methods: Ten patients treated for esophageal cancer that received daily CBCT were retrospectively evaluated (Varian TrueBeam with same Thorax imaging technique: half-fan, full-trajectory, 125 kVp, 270 mAs). The planning CT (CTP) and the two initial CBCTs (day 1 and 2 of treatment) were exported from Eclipse TPS to an in-house processing pipeline. This included edge detection for couch removal, CBCT resampling and automatic 3D rigid-registration of CBCT to CTP using Mattes mutual information metric. GTVs for each patient were exported and texture features were extracted from CTP and the registered CBCTs using the Imaging Biomarker Explorer (IBEX) software. Thirty-three texture features using cooccurrence and run-length matrices were extracted. Texture reproducibility between consecutive CBCTs was evaluated using intraclass correlation (ICC). CTP and CBCTs were evaluated using Pearson's correlation. Significance level was corrected for multiple testing using Bonferroni adjustment. Results: Registration results were deemed satisfactory using mutual information as well as visual inspection within the volume that encompassed the GTV. Out of the initial 33 texture features considered, 13 presented excellent (ICC > 0.9) reproducibility between the two initial CBCTs. Out of these 13 features, 5 presented statistically significant Pearson's correlation adjusted for multiple statistical tests (P < 0.004) ranging from 0.86 to 0.89 (Table). Conclusion: Several texture features from CBCT showed reproducibility between the first two days of treatment and strongly correlated between CTP and CBCT. Therefore, derived texture features could be investigated as predictors of treatment response during the course of treatment

OBJECTIVE:To explore the relationship between relative enhanced diffusivity (RED) and intravoxel incoherent motion (IVIM), as well as the impact of noise and the choice of intermediate diffusion weighting (b value) on the RED parameter. MATERIALS AND METHODS/METHODS:A mathematical derivation was performed to cast RED in terms of the IVIM parameters. Noise analysis and b value optimization was conducted by using Monte Carlo calculations to generate diffusion-weighted imaging data appropriate to breast and liver tissue at three different signal-to-noise ratios. RESULTS:RED was shown to be approximately linearly proportional to the IVIM parameter f, inversely proportional to D and to follow an inverse exponential decay with respect to D*. The choice of intermediate b value was shown to be important in minimizing the impact of noise on RED and in maximizing its discriminatory power. RED was shown to be essentially a reparameterization of the IVIM estimates for f and D obtained with three b values. CONCLUSION/CONCLUSIONS:, respectively. Future clinical studies involving RED should also estimate the IVIM parameters f and D using three b values for comparison.

PURPOSE:inhomogeneity-induced geometric distortion in echo-planar diffusion-weighted imaging on quantitative apparent diffusion coefficient (ADC) analysis in multiparametric prostate MRI. METHODS:-weighted images were mapped to the nondistortion-corrected and distortion-corrected data sets by resampling with and without spatial coregistration. The ADC values were calculated on the volume and voxel level. The effect of distortion correction on ADC quantification and tissue classification was evaluated using linear-mixed models and logistic regression, respectively. RESULTS:(voxel level)) between the nondistortion-corrected and distortion-corrected were significantly associated (P < 0.05) with distortion distance (mean: 1.4 ± 1.3 mm; range: 0.3-5.3 mm). No significant associations were found upon coregistration; however, in patients with high rectal gas residue, distortion correction resulted in improved spatial representation and significantly better classification of healthy versus tumor voxels (P < 0.05). CONCLUSIONS:Geometric distortion correction in DWI could improve quantitative ADC analysis in multiparametric prostate MRI. Magn Reson Med 79:2524-2532, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

BACKGROUND:Diffusion-weighted MRI (DWI) is currently one of the fastest developing MRI-based techniques in oncology. Histogram properties from model fitting of DWI are useful features for differentiation of lesions, and classification can potentially be improved by machine learning. PURPOSE:To evaluate classification of malignant and benign tumors and breast cancer subtypes using support vector machine (SVM). STUDY TYPE:Prospective. SUBJECTS:Fifty-one patients with benign (n = 23) and malignant (n = 28) breast tumors (26 ER+, whereof six were HER2+). FIELD STRENGTH/SEQUENCE:Patients were imaged with DW-MRI (3T) using twice refocused spin-echo echo-planar imaging with echo time / repetition time (TR/TE) = 9000/86 msec, 90 × 90 matrix size, 2 × 2 mm in-plane resolution, 2.5 mm slice thickness, and 13 b-values. ASSESSMENT:Apparent diffusion coefficient (ADC), relative enhanced diffusivity (RED), and the intravoxel incoherent motion (IVIM) parameters diffusivity (D), pseudo-diffusivity (D*), and perfusion fraction (f) were calculated. The histogram properties (median, mean, standard deviation, skewness, kurtosis) were used as features in SVM (10-fold cross-validation) for differentiation of lesions and subtyping. STATISTICAL TESTS:Accuracies of the SVM classifications were calculated to find the combination of features with highest prediction accuracy. Mann-Whitney tests were performed for univariate comparisons. RESULTS:For benign versus malignant tumors, univariate analysis found 11 histogram properties to be significant differentiators. Using SVM, the highest accuracy (0.96) was achieved from a single feature (mean of RED), or from three feature combinations of IVIM or ADC. Combining features from all models gave perfect classification. No single feature predicted HER2 status of ER + tumors (univariate or SVM), although high accuracy (0.90) was achieved with SVM combining several features. Importantly, these features had to include higher-order statistics (kurtosis and skewness), indicating the importance to account for heterogeneity. DATA CONCLUSION:Our findings suggest that SVM, using features from a combination of diffusion models, improves prediction accuracy for differentiation of benign versus malignant breast tumors, and may further assist in subtyping of breast cancer. LEVEL OF EVIDENCE:3 Technical Efficacy: Stage 3 J. Magn. Reson. Imaging 2018;47:1205-1216.