Fast Multiscale Neighbor Embedding.

Dimension reduction (DR) computes faithful low-dimensional (LD) representations of high-dimensional (HD) data. Outstanding performances are achieved by recent neighbor embedding (NE) algorithms such as t -SNE, which mitigate the curse of dimensionality. The single-scale or multiscale nature of NE schemes drives the HD neighborhood preservation in the LD space (LDS). While single-scale methods focus on single-sized neighborhoods through the concept of perplexity, multiscale ones preserve neighborhoods in a broader range of sizes and account for the global HD organization to define the LDS. For both single-scale and multiscale methods, however, their time complexity in the number of samples is unaffordable for big data sets. Single-scale methods can be accelerated by relying on the inherent sparsity of the HD similarities they involve. On the other hand, the dense structure of the multiscale HD similarities prevents developing fast multiscale schemes in a similar way. This article addresses this difficulty by designing randomized accelerations of the multiscale methods. To account for all levels of interactions, the HD data are first subsampled at different scales, enabling to identify small and relevant neighbor sets for each data point thanks to vantage-point trees. Afterward, these sets are employed with a Barnes-Hut algorithm to cheaply evaluate the considered cost function and its gradient, enabling large-scale use of multiscale NE schemes. Extensive experiments demonstrate that the proposed accelerations are, statistically significantly, both faster than the original multiscale methods by orders of magnitude, and better preserving the HD neighborhoods than state-of-the-art single-scale schemes, leading to high-quality LD embeddings. Public codes are freely available at https://github.com/cdebodt.

Improvement of kilovoltage intrafraction monitoring accuracy through gantry angles selection.

Kilovoltage intrafraction monitoring (KIM) is a method allowing to precisely infer the tumour trajectory based on cone beam computed tomography (CBCT) 2D-projections. However, its accuracy is deteriorated in the case of highly mobile tumours involving hysteresis. A first adaptation of KIM consisting of a prior amplitude based binning step has been developed in order to minimize the errors of the original model (phase-KIM). In this work, we propose enhanced methods (KIMsub-arc optimand phase-KIMsub-arc optim) to improve the accuracy of KIM and phase-KIM which relies on the selection of the optimal starting CBCT gantry angle. Aiming at demonstrating the interest of our approach, we carried out a simulation study and an experimental study: we compared the accuracy of the conventional versus sub-arc optim methods on simulated realistic tumour motions with amplitudes ranging from 5 to 30 mm in 1 mm increments. The same approach was performed using a lung dynamic phantom generating a 30 mm amplitude sinusoidal motion. The results show that for in-silico simulated motions of 10, 20 and 30 mm amplitude, the three-dimensional root mean square error (3D-RMSE) can be reduced by 0.67 mm, 0.91 mm, 0.94 mm and 0.18 mm, 0.25 mm, 0.28 mm using KIMsub-arc optimand phase-KIMsub-arc optimrespectively. Considering all in-silico simulated trajectories, the percentage of errors larger than 1 mm decreases from 21.9% down to 1.6% for KIM (p < 0.001) and from 6.6% down to 1.2% for phase-KIM (p < 0.001). Experimentally, the 3D-RMSE is lowered by 0.5732 mm for KIM and by 0.1 mm for phase-KIM. The percentage of errors larger than 1 mm falls from 39.7% down to 18.5% for KIM and from 23.2% down to 11.1% for phase-KIM. In conclusion, our method efficiently anticipates CBCT gantry angles associated with a significantly better accuracy by using KIM and phase-KIM.

Towards fast and robust 4D optimization for moving tumors with scanned proton therapy.

PURPOSE: Robust optimization is becoming the gold standard for generating robust plans against various kinds of treatment uncertainties. Today, most robust optimization strategies use a pragmatic set of treatment scenarios (the so-called uncertainty set) consisting of combinations of maximum errors, of each considered uncertainty source (such as tumor motion, setup and image-conversion errors). This approach presents two key issues. First, a subset of considered scenarios is unnecessarily improbable which could potentially compromise the plan quality. Second, the resulting large uncertainty set leads to long plan computation times, which limits the potential for robust optimization as a standard clinical tool. In order to address these issues, a method is introduced which is able to preselect a limited set of relevant treatment error scenarios. METHODS: Uncertainties due to systematic setup errors, image-conversion errors and respiratory tumor motion are considered. A four-dimensional (4D)-equiprobability hypersurface is defined, which takes into account the joint probabilities of the above-mentioned uncertainty sources. Only scenarios that lie on the predefined 4D hypersurface are considered, guaranteeing statistical consistency of the uncertainty set. In this regard, twelve scenarios are selected that cover maximum spatial displacements of the tumor during breathing. Subsequently, additional scenarios are considered (sampled from the aforementioned 4D hypersurface) in order to cover any estimated residual range errors. Two different scenario-selection procedures were tested: (a) the maximum displacements (MD) method that only considers twelve scaled maximum displacement scenarios and (b) maximum displacements and residual range (MDR) method which, in addition to the scaled maximum displacement scenarios, considers additional maximum range uncertainty scenarios. The methods were tested for five lung cancer patients by performing comprehensive Monte Carlo robustness evaluations. RESULTS: A plan computation time gain of 78% is achieved by applying the MD method, whilst obtaining a target robustness of D 95 larger than 95% of the prescribed dose, for the worst-case scenario. Additionally, the MD method has the potential to be fully automatic which makes it a promising candidate for fast automatic planning workflows. The MDR method produced plans with excellent target robustness (D 99 larger than 95% of the prescribed dose, even for the worst-case scenario), whilst still obtaining a significant plan computation time gain of 57%. CONCLUSIONS: Two scenario-selection procedures were developed which achieved significant reduction of plan computation time and memory consumption, without compromising plan quality or robustness.

Mechanically-assisted and non-invasive ventilation for radiation therapy: A safe technique to regularize and modulate internal tumour motion.

BACKGROUND AND PURPOSE: Current motion mitigation strategies, like margins, gating, and tracking, deal with geometrical uncertainties in the tumour position, induced by breathing during radiotherapy (RT). However, they often overlook motion variability in amplitude, respiratory rate, or baseline position, when breathing spontaneously. Consequently, this may negatively affect the delivered dose conformality in comparison to the plan. We previously demonstrated on volunteers that 3 different modes of mechanically-assisted and non-invasive ventilation (MANIV) may reduce variability in breathing motion. The volume-controlled mode (VC) constraints the amplitude and respiratory rate (RR) in physiologic condition. The shallow-controlled mode (SH), derived from VC, increases the RR and decreases amplitude. The slow-controlled mode (SL) induces repeated breath holds with constrained ventilation pressure. In this study, we compared these mechanical ventilation modes to spontaneous breathing or breath hold and assessed their tolerance and effects on internal tumour motion in patients receiving RT. MATERIAL AND METHODS: The VC and SH modes were evaluated in ten patients with lung or liver cancers (cohort A). The SL mode was evaluated in 12 left breast cancer patients (cohort B). After a training and simulation session, the patients underwent 2 MRI sessions to analyze the internal motion of breast and tumour. RESULTS: MANIV was well tolerated, without any adverse events or oxymetric changes, even in patients with respiratory comorbidities. In cohort A, when compared to spontaneous breathing (SP), VC reduced significantly inter-session variations of the tumour motion amplitude (p = 0.01), as well as intra- and inter-session variations of the RR (p < 0.05). As to SH, the RR increased, while its variations within and across sessions decreased when compared to SP (p < 0.001). SH reduced the median amplitude of the tumour motion by 6.1 mm or 38.2% (p ≤ 0.01) compared to VC. In cohort B, breast position stability over the end-inspiratory plateaus obtained spontaneously or with SL remained similar. Median duration of the plateaus in SL was 16.6 s. CONCLUSION: MANIV is a safe and well tolerated ventilation technique for patients receiving radiotherapy. MANIV could thus make current motion mitigation strategies less critical and more robust. Clinical implementation might be considered, provided the ventilation mode is carefully selected with respect to the treatment indication and patient individualities.

Nonlinear Dimensionality Reduction With Missing Data Using Parametric Multiple Imputations.

Dimensionality reduction (DR) aims at faithfully and meaningfully representing high-dimensional (HD) data into a low-dimensional (LD) space. Recently developed neighbor embedding DR methods lead to outstanding performances, thanks to their ability to foil the curse of dimensionality. Unfortunately, they cannot be directly employed on incomplete data sets, which become ubiquitous in machine learning. Discarding samples with missing features prevents their LD coordinates computation and deteriorates the complete samples treatment. Common missing data imputation schemes are not appropriate in the nonlinear DR context either. Indeed, even if they model the data distribution in the feature space, they can, at best, enable the application of a DR scheme on the expected data set. In practice, one would, instead, like to obtain the LD embedding with the closest cost function value on average with respect to the complete data case. As the state-of-the-art DR techniques are nonlinear, the latter embedding results from minimizing the expected cost function on the incomplete database, not from considering the expected data set. This paper addresses these limitations by developing a general methodology for nonlinear DR with missing data, being directly applicable with any DR scheme optimizing some criterion. In order to model the feature dependences, an HD extension of Gaussian mixture models is first fitted on the incomplete data set. It is afterward employed under the multiple imputation paradigms to obtain a single relevant LD embedding, thus minimizing the cost function expectation. Extensive experiments demonstrate the superiority of the suggested framework over alternative approaches.

Performance of a hybrid Monte Carlo-Pencil Beam dose algorithm for proton therapy inverse planning.

PURPOSE: Analytical algorithms have a limited accuracy when modeling very heterogeneous tumor sites. This work addresses the performance of a hybrid dose optimizer that combines both Monte Carlo (MC) and pencil beam (PB) dose engines to get the best trade-off between speed and accuracy for proton therapy plans. METHODS: The hybrid algorithm calculates the optimal spot weights (w) by means of an iterative optimization process where the dose at each iteration is computed by using a precomputed dose influence matrix based on the conventional PB plus a correction term c obtained from a MC simulation. Updates of c can be triggered as often as necessary by calling the MC dose engine with the last corrected values of w as input. In order to analyze the performance of the hybrid algorithm against dose calculation errors, it was applied to a simplistic water phantom for which several test cases with different errors were simulated, including proton range uncertainties. Afterwards, the algorithm was used in three clinical cases (prostate, lung, and brain) and benchmarked against full MC-based optimization. The influence of different stopping criteria in the final results was also investigated. RESULTS: The hybrid algorithm achieved excellent results provided that the estimated range in a homogeneous material is the same for the two dose engines involved, i.e., PB and MC. For the three patient cases, the hybrid plans were clinically equivalent to those obtained with full MC-based optimization. Only a single update of c was needed in the hybrid algorithm to fulfill the clinical dose constraints, which represents an extra computation time to obtain c that ranged from 1 (brain) to 4 min (lung) with respect to the conventional PB-based optimization, and an estimated average gain factor of 14 with respect to full MC-based optimization. CONCLUSION: The hybrid algorithm provides an improved trade-off between accuracy and speed. This algorithm can be immediately considered as an option for improving dose calculation accuracy of commercial analytical treatment planning systems, without a significant increase in the computation time (≪5 min) with respect to current PB-based optimization.

Molecular Imaging-Guided Radiotherapy for the Treatment of Head-and-Neck Squamous Cell Carcinoma: Does it Fulfill the Promises?

With the routine use of intensity modulated radiation therapy for the treatment of head-and-neck squamous cell carcinoma allowing highly conformed dose distribution, there is an increasing need for refining both the selection and the delineation of gross tumor volumes (GTV). In this framework, molecular imaging with positron emission tomography and magnetic resonance imaging offers the opportunity to improve diagnostic accuracy and to integrate tumor biology mainly related to the assessment of tumor cell density, tumor hypoxia, and tumor proliferation into the treatment planning equation. Such integration, however, requires a deep comprehension of the technical and methodological issues related to image acquisition, reconstruction, and segmentation. Until now, molecular imaging has had a limited value for the selection of nodal GTV, but there are increasing evidences that both FDG positron emission tomography and diffusion-weighted magnetic resonance imaging has a potential value for the delineation of the primary tumor GTV, effecting on dose distribution. With the apprehension of the heterogeneity in tumor biology through molecular imaging, growing evidences have been collected over the years to support the concept of dose escalation/dose redistribution using a planned heterogeneous dose prescription, the so-called "dose painting" approach. Validation trials are ongoing, and in the coming years, one may expect to position the dose painting approach in the armamentarium for the treatment of patients with head-and-neck squamous cell carcinoma.

An individualized radiation dose escalation trial in non-small cell lung cancer based on FDG-PET imaging.

AIM: The aim of the study was to assess the feasibility of an individualized 18F fluorodeoxyglucose positron emission tomography (FDG-PET)-guided dose escalation boost in non-small cell lung cancer (NSCLC) patients and to assess its impact on local tumor control and toxicity. PATIENTS AND METHODS: A total of 13 patients with stage II-III NSCLC were enrolled to receive a dose of 62.5 Gy in 25 fractions to the CT-based planning target volume (PTV; primary turmor and affected lymph nodes). The fraction dose was increased within the individual PET-based PTV (PTVPET) using intensity modulated radiotherapy (IMRT) with a simultaneous integrated boost (SIB) until the predefined organ-at-risk (OAR) threshold was reached. Tumor response was assessed during follow-up by means of repeat FDG-PET/computed tomography. Acute and late toxicity were recorded and classified according to the CTCAE criteria (Version 4.0). Local progression-free survival was determined using the Kaplan-Meier method. RESULTS: The average dose to PTVPET reached 89.17 Gy for peripheral and 75 Gy for central tumors. After a median follow-up period of 29 months, seven patients were still alive, while six had died (four due to distant progression, two due to grade 5 toxicity). Local progression was seen in two patients in association with further recurrences. One and 2-year local progression free survival rates were 76.9% and 52.8%, respectively. Three cases of acute grade 3 esophagitis were seen. Two patients with central tumors developed late toxicity and died due to severe hemoptysis. CONCLUSION: These results suggest that a non-uniform and individualized dose escalation based on FDG-PET in IMRT delivery is feasible. The doses reached were higher in patients with peripheral compared to central tumors. This strategy enables good local control to be achieved at acceptable toxicity rates. However, dose escalation in centrally located tumors with direct invasion of mediastinal organs must be performed with great caution in order to avoid severe late toxicity.

Consistency in quality correction factors for ionization chamber dosimetry in scanned proton beam therapy.

PURPOSE: The IAEA TRS-398 code of practice details the reference conditions for reference dosimetry of proton beams using ionization chambers and the required beam quality correction factors (kQ ). Pencil beam scanning (PBS) systems cannot approximate reference conditions using a single spot. However, dose distributions requested in TRS-398 can be reproduced with PBS using a combination of spots. This study aims to demonstrate, using Monte Carlo (MC) simulations, that kQ factors computed/measured for broad beams can be used with scanned beams for similar reference dose distributions with no additional significant uncertainty. METHODS: We consider the Alfonso formalism13 usually employed for nonstandard photon beams. To approach reference conditions similar as IAEA TRS-398 and the associated dose distributions, PBS must combine many pencil beams with range or energy modulation and shaping techniques that differ from those used in passive systems (broad beams). In order to evaluate the impact of these differences on kQ factors, ionization chamber responses are computed with MC (Geant4 9.6) in three different proton beams, with their corresponding quality factors (Q), producing a 10 × 10 cm2 field with a flat dose distribution for (a) a dedicated scanned pencil beam (Qpbs ), (b) a hypothetical proton source (Qhyp ), and (c) a double-scattering beam (Qds ). The tested ionization chamber cavities are a 2 × 2 × 0.2 mm³ air cavity, a Roos-type ionization chamber, and a Farmer-type ionization chamber. RESULTS AND DISCUSSION: Ranges of Qpbs , Qhyp , and Qds are consistent within 0.4 mm. Flatnesses of dose distributions are better than 0.5%. Calculated kQpbs,Qhypfpbs,fref is 0.999 ± 0.002 for the air cavity and the Farmer-type ionization chamber and 1.001 ± 0.002 for the Roos-type ionization chamber. The quality correction factors kQpbs,Qdsfpbs,fref is 0.999 ± 0.002 for the Farmer-type and Roos-type ionization chambers and 1.001 ± 0.001 for the Roos-type ionization chamber. CONCLUSION: The Alfonso formalism was applied to scanned proton beams. In our MC simulations, neither the difference in the beam profiles (scanned beam vs hypothetical beam) nor the different incident beam energies influenced significantly the beam correction factors. This suggests that ionization chamber quality correction factors in scanned or broad proton beams are indistinguishable within the calculation uncertainties provided dose distributions achieved by both modalities are similar and compliant with the TRS-398 reference conditions.

Radiation dose escalation based on FDG-PET driven dose painting by numbers in oropharyngeal squamous cell carcinoma: a dosimetric comparison between TomoTherapy-HA and RapidArc.

PURPOSE: Validation of dose escalation through FDG-PET dose painting (DP) for oropharyngeal squamous cell carcinoma (SCC) requires randomized clinical trials with large sample size, potentially involving different treatment planning and delivery systems. As a first step of a joint clinical study of DP, a planning comparison was performed between Tomotherapy HiArt® (HT) and Varian RapidArc® (RA). METHODS: The planning study was conducted on five patients with oropharyngeal SCC. Elective and therapeutic CTVs were delineated based on anatomic information, and the respective PTVs (CTVs + 4 mm) were prescribed a dose of 56 (PTV56) and 70 Gy (PTV70). A gradient-based method was used to delineate automatically the external contours of the FDG-PET volume (GTVPET). Variation of the FDG uptake within the GTVPET was linearly converted into a prescription between 70 and 86 Gy. A dilation of the voxel-by-voxel prescription of 2.5 mm was applied to account for geometric errors in dose delivery (PTVPET). The study was divided in two planning phases aiming at maximizing target coverage (phase I) and lowering doses to OAR (phase II). A Quality-Volume Histogram (QVH) assessed conformity with the DP prescription inside the PTVPET. RESULTS: In phase I, for both HT and RA, all plans achieved comparable target coverage for PTV56 and PTV70, respecting the planning objectives. A median value of 99.9 and 97.2% of all voxels in the PTVPET received at least 95% of the prescribed dose for RA and HT, respectively. A median value of 0.0% and 3.7% of the voxels in the PTVPET received 105% or more of prescribed dose for RA and HT, respectively. In phase II, no significant differences were found in OAR sparing. Median treatment times were 13.7 min for HT and 5 min for RA. CONCLUSIONS: Both HT and RA can generate similar dose distributions for FDG-PET based dose escalation and dose painting in oropharyngeal SCC patients.

Fast multipurpose Monte Carlo simulation for proton therapy using multi- and many-core CPU architectures.

PURPOSE: Accuracy in proton therapy treatment planning can be improved using Monte Carlo (MC) simulations. However the long computation time of such methods hinders their use in clinical routine. This work aims to develop a fast multipurpose Monte Carlo simulation tool for proton therapy using massively parallel central processing unit (CPU) architectures. METHODS: A new Monte Carlo, called MCsquare (many-core Monte Carlo), has been designed and optimized for the last generation of Intel Xeon processors and Intel Xeon Phi coprocessors. These massively parallel architectures offer the flexibility and the computational power suitable to MC methods. The class-II condensed history algorithm of MCsquare provides a fast and yet accurate method of simulating heavy charged particles such as protons, deuterons, and alphas inside voxelized geometries. Hard ionizations, with energy losses above a user-specified threshold, are simulated individually while soft events are regrouped in a multiple scattering theory. Elastic and inelastic nuclear interactions are sampled from ICRU 63 differential cross sections, thereby allowing for the computation of prompt gamma emission profiles. MCsquare has been benchmarked with the gate/geant4 Monte Carlo application for homogeneous and heterogeneous geometries. RESULTS: Comparisons with gate/geant4 for various geometries show deviations within 2%-1 mm. In spite of the limited memory bandwidth of the coprocessor simulation time is below 25 s for 10(7) primary 200 MeV protons in average soft tissues using all Xeon Phi and CPU resources embedded in a single desktop unit. CONCLUSIONS: MCsquare exploits the flexibility of CPU architectures to provide a multipurpose MC simulation tool. Optimized code enables the use of accurate MC calculation within a reasonable computation time, adequate for clinical practice. MCsquare also simulates prompt gamma emission and can thus be used also for in vivo range verification.

Methodology for adaptive and robust FDG-PET escalated dose painting by numbers in head and neck tumors.

OBJECTIVE: To develop a methodology for using FDG PET/CT in adaptive dose painting by numbers (DPBN) in head and neck squamous cell carcinoma (HNSCC) patients. Issues related to noise in PET and treatment robustness against geometric errors are addressed. METHODS: Five patients with locally advanced HNSCC scheduled for chemo-radiotherapy were imaged with FDG-PET/CT at baseline and 2-3 times during radiotherapy (RT). The GTVPET was segmented with a gradient-based method. A double median filter reduces the impact of noise in the PET uptake-to-dose conversion. Filtered FDG uptake values were linearly converted into a voxel-by-voxel prescription from 70 (median uptake) to 86 Gy (highest uptake). A PTVPET was obtained by applying a dilation of 2.5 mm to the entire prescription. Seven iso-uptake thresholds led to seven sub-levels compatible with the Tomotherapy HiArt(®) Treatment Planning System. Planning aimed to deliver a median dose of 56 Gy and 70 Gy in 35 fractions on the elective and therapeutic PTVs, respectively. Plan quality was assessed with quality volume histogram (QVH). At each time point, plans were generated with a total of 3-4 plans for each patient. Deformable image registration was used for automatic contour propagation and dose summation of the 3 or 4 treatment plans (MIMvista(®)). RESULTS: GTVPET segmentations were performed successfully until week 2 of RT but failed in two patients at week 3. QVH analysis showed high conformity for all plans (mean VQ = 0.95 93%; mean VQ = 1.05 3.9%; mean QF 2.2%). Good OAR sparing was achieved while keeping high plan quality. CONCLUSION: Our results show that adaptive FDG-PET-based escalated dose painting in patients with locally advanced HNSCC is feasible while respecting strict dose constraints to organs at risk. Clinical studies must be conducted to evaluate toxicities and tumor response of such a strategy.

Validation of the mid-position strategy for lung tumors in helical TomoTherapy.

PURPOSE: To compare the mid-position (MidP) strategy to the conventional internal target volume (ITV) for lung tumor management in helical TomoTherapy, using 4D Monte Carlo (MC) plan simulations. MATERIALS AND METHODS: For NSCLC patients treated by SBRT (n = 8) or SIB-IMRT (n = 7), target volumes and OARs were delineated on a contrast-enhanced CT, while 4D-CT was used to generate either ITV or MidP volumes with deformable registrations. PTV margins were added. Conformity indexes, volumetric and dosimetric parameters were compared for both strategies. Dose distributions were also computed using a 4D MC model (TomoPen) to assess how intra-fraction tumor motion affects tumor coverage, with and without interplay effect. RESULTS: PTVs derived from MidP were on average 1.2 times smaller than those from ITV, leading to lower doses to OARs. Planned dose conformity to TVs was similar for both strategies. 4D MC computation showed that ITV ensured adequate TV coverage (D95 within 1% of clinical requirements), while MidP failed in 3 patients of the SBRT group (D95 to the TV lowered by 4.35%, 2.16% and 2.61%) due to interplay effect in one case and to breathing motion alone in the others. CONCLUSIONS: Compared to the ITV, the MidP significantly reduced PTV and doses to OARs. MidP is safe for helical delivery except for very small tumors (<5 cc) with large-amplitude motion (>10mm) where the ITV might remain the most adequate approach.

Radiotherapy for head and neck tumours in 2012 and beyond: conformal, tailored, and adaptive?

Intensity-modulated radiation therapy (IMRT) is a conformal irradiation technique that enables steep dose gradients. In head and neck tumours this approach spares parotid-gland function without compromise to treatment efficacy. Anatomical and molecular imaging modalities may be used to tailor treatment by enabling proper selection and delineation of target volumes and organs at risk, which in turn lead to dose prescriptions that take into account the underlying tumour biology (eg, human papillomavirus status). Therefore, adaptations can be made throughout the course of radiotherapy, as required. Planned dose increases to parts of the target volumes may also be used to match the radiosensitivity of tumours (so-called dose-painting), assessed by molecular imaging. For swift implementation of tailored and adaptive IMRT, tools and procedures, such as accurate image acquisition and reconstruction, automatic segmentation of target volumes and organs at risk, non-rigid image and dose registration, and dose summation methods, need to be developed and properly validated.

Adaptive functional image-guided IMRT in pharyngo-laryngeal squamous cell carcinoma: is the gain in dose distribution worth the effort?

BACKGROUND AND PURPOSE: The planning process in radiotherapy (RT) typically involves the acquisition of a unique set of CT images - and eventually of functional images - which is used for delineation of target volumes (TV) and organs at risk (OAR) and for dose calculation. Restricting the delineation and dose calculation solely on pre-treatment images is an oversimplification as it is only a snapshot of the patient's anatomy. The objectives of the present study were (1) to assess the consequences of anatomic modification in dose distribution for both TVs and OARs; (2) to assess the potential benefit of adaptive strategies using Helical Tomotherapy (HT); and (3) to compare CT-based and FDG-PET-based adaptive planning strategies. MATERIALS AND METHODS: Ten patients with H&N SCC were imaged before and during concomitant chemo-RT using CT and FDG-PET acquisition after a mean dose of 14.2, 24.5, 35.0 and 44.9 Gy. Simultaneous integrated boost IMRT planning was performed using HT. We compared (1) the planned dose distribution, (2) the delivered dose distributions that took into account impact of anatomical modifications on dose distribution, (3) the adaptive dose distributions after replanning to take into account the anatomic modifications and the anatomic or functional GTV shrinkage. RESULTS: There was an increase between the planned and the delivered high dose volumes, which correlated with the slope of the GTV shrinkage. The adaptive high dose volumes were significantly smaller than the delivered ones. The difference between the adaptive and the delivered high dose volume also correlated with the slope of the GTV shrinkage. For both parotid glands combined, the delivered D(mean) showed a statistical trend for an increase of 4.4% compared to the planned D(mean). For the ipsilateral parotid glands, there was a correlation between the D(mean) gain and the slope of the GTV shrinkage when an adaptive planning was used. For the oral cavity, the adaptive D(mean) was 10% smaller than the delivered ones. For the PRV around the spinal cord, there was an increase of about 4.5% between the delivered and the planned D(2%). The adaptive planning translated into a decrease in D(2%) of 7.2%. The differences between the delivered and planned D(2%) and between the adaptive and the delivered D(2%) were correlated with the slope of the GTV shrinkage. For the CTV(proph) and PTV(proph) coverage, adaptive strategy induced a better dose conformation. No significant difference was observed in the various figures of merit between PET-based plan and CT-based isodose distributions. CONCLUSIONS: The dose distribution that is actually delivered to patients significantly differs from what was planned because of anatomic modifications. Adaptive multi-modality IMRT is feasible in H&N tumors and could compensate and improve dose distribution. Some useful surrogate criteria or "flags" are, however, needed to identify patients who might benefit from an adaptive strategy. The optimal adaptive strategy still needs to be defined and prospective studies will have to be conducted to address the safety and the clinical impact of such approaches on patient outcome.

Gradient-based delineation of the primary GTV on FDG-PET in non-small cell lung cancer: a comparison with threshold-based approaches, CT and surgical specimens.

PURPOSE: The aim of this study was to validate a gradient-based segmentation method for GTV delineation on FDG-PET in NSCLC through surgical specimen, in comparison with threshold-based approaches and CT. MATERIALS AND METHODS: Ten patients with stage I-II NSCLC were prospectively enrolled. Before lobectomy, all patients underwent contrast enhanced CT and gated FDG-PET. Next, the surgical specimen was removed, inflated with gelatin, frozen and sliced. The digitized slices were used to reconstruct the 3D macroscopic specimen. GTVs were manually delineated on the macroscopic specimen and on CT images. GTVs were automatically segmented on PET images using a gradient-based method, a source to background ratio method and fixed threshold values at 40% and 50% of SUV(max). All images were finally registered. Analyses of raw volumes and logarithmic differences between GTVs and GTV(macro) were performed on all patients and on a subgroup excluding the poorly defined tumors. A matching analysis between the different GTVs was also conducted using Dice's similarity index. RESULTS: Considering all patients, both lung and mediastinal windowed CT overestimated the macroscopy, while FDG-PET provided closer values. Among various PET segmentation methods, the gradient-based technique best estimated the true tumor volume. When analysis was restricted to well defined tumors without lung fibrosis or atelectasis, the mediastinal windowed CT accurately assessed the macroscopic specimen. Finally, the matching analysis did not reveal significant difference between the different imaging modalities. CONCLUSIONS: FDG-PET improved the GTV definition in NSCLC including when the primary tumor was surrounded by modifications of the lung parenchyma. In this context, the gradient-based method outperformed the threshold-based ones in terms of accuracy and robustness. In other cases, the conventional mediastinal windowed CT remained appropriate.

Evaluation of the radiobiological impact of anatomic modifications during radiation therapy for head and neck cancer: can we simply summate the dose?

BACKGROUND AND PURPOSE: Adaptive strategies in radiotherapy (RT) require the knowledge of the total dose given to every organ of the body. Because of anatomical changes and setup errors non-rigid registration is necessary to map the different dose fractions to a common reference. This study evaluates practically if the accumulation of all of these registered dose fractions must take radiobiology into account in a classical clinical setting. MATERIALS AND METHODS: Ten patients with head and neck tumors treated by chemo-RT were used. Contrast-enhanced CT scans were acquired prior and during RT following delivery of mean doses of 14.2, 24.5, 35.0 and 44.9 Gy and the planned pre-treatment helical tomotherapy sinograms were applied on the per-treatment CTs to create a series of per-treatment dose distributions corresponding to each per-treatment CT image. In order to calculate the cumulative dose distribution, the per-treatment dose maps were non-rigidly deformed by using the deformation map computed by a non-rigid registration. The deformed dose maps were then summed in two ways: one while taking radiobiology into account and one without. These two strategies were compared using clinical surrogates in the target volumes (TV) and in surrounding organs at risk (OAR). RESULTS: The differences between the strategies, while statistically significant (p<0.05), are clinically irrelevant. In the OARs, the mean differences stay in the 0.01-0.07 Gy range for the total dose. In the targets, all mean differences stay in the 0.001-0.012 Gy range. However, some local high difference spots appear leading to punctual errors as high as 2.5 Gy. CONCLUSION: If using current radiotherapy practices and clinical recommendations based on dose surrogates computed globally on OARs and TVs, one does not need to take radiobiological effects into account while accumulating total dose as these lead to very small differences compared to a simple accumulation technique consisting of a linear sum of the dose fractions. However, care must be taken if other adaptive strategies, based on local rather than global information, are used.

Assessment by a deformable registration method of the volumetric and positional changes of target volumes and organs at risk in pharyngo-laryngeal tumors treated with concomitant chemo-radiation.

PURPOSE: Anatomic changes occur during radiation therapy (RT) for head and neck (H&N) tumors. This study aims at quantifying the volumetric and positional changes of gross tumor volumes (GTV), clinical target volumes (CTV), and organs at risk (OAR). Anatomic (CT) and functional (FDG-PET) imaging were used for the delineation of the GTVs. MATERIALS AND METHODS: Ten patients with H&N tumors treated by chemo-RT were used. Contrast-enhanced CT and FDG-PET were acquired prior and during RT following delivery of mean doses of 14.2, 24.5, 35.0, and 44.9 Gy. CT-based GTVs were manually delineated, and PET-based GTVs were segmented using a gradient-based segmentation method. Pre-treatment prophylactic dose CTVs were manually delineated on the pre-treatment CT using consistent and reproducible guidelines. Per-treatment prophylactic CTVs were obtained with an automatic re-contouring method based on deformable registration. For the therapeutic dose CTVs, a 5 mm margin was applied around the corresponding GTVs. OARs such as the parotid glands and the submandibular glands were manually delineated on the pre-treatment CT. OARs on the per-treatment CT were automatically delineated using the method used for prophylactic CTVs. The mean slopes of the relative change in volume over time and the mean displacements of the center of mass after 44.9 Gy were calculated for each volume. RESULTS: Regarding volumetric changes, CT-based and PET-based primary tumor GTVs decreased at a mean rate of 3.2% and 3.9%/treatment day (td), respectively; nodal GTVs decreased at a mean rate of 2.2%/td. This led to a corresponding decrease of the CT-based and PET-based therapeutic CTVs by 2.4% and 2.5%/td, respectively. CT- and PET-based prophylactic tumor CTVs decreased by an average of 0.7% and 0.5%/td, respectively. No difference in volume shrinkage was observed between CT- and PET-based volumes. The ipsilateral and contralateral parotid glands showed a mean decrease of 0.9% and 1.0%/td, respectively. The ipsilateral and contralateral submandibular glands shrank at a mean rate of 1.5% and 1.3%/td, respectively. Regarding positional changes, CT-based GTVs showed a lateral shift of 1.3 mm, PET-based GTVs a posterior shift of 3.4mm and the nodal GTVs a medial shift of 1.0 mm, translating into parallel shifts of the therapeutic CTVs. The ipsilateral prophylactic nodal CTV shifted medially by 1.8 mm. The ipsilateral parotid gland shifted medially by 3.4 mm. The ipsilateral submandibular gland showed a medial shift of 1.7 mm and a superior shift of 2.7 mm. The contralateral submandibular gland only showed a superior shift of 1.7 mm. CONCLUSIONS: Volumetric and positional changes in TVs and OARs were observed during concomitant chemo-RT suggesting that adaptive strategies, where patients are re-imaged and possibly re-planned during treatment, are worth evaluating.

Variance stabilizing transformations in patch-based bilateral filters for poisson noise image denoising.

Denoising is a key step in the processing of medical images. It aims at improving both the interpretability and visual aspect of the images. Yet, designing a robust and efficient denoising tool remains an unsolved challenge and a specific issue concerns the noise model. Many filters typically assume that noise is additive and Gaussian, with uniform variance. In contrast, noise in medical images often has more complex properties. This paper considers images with Poissonian noise and the patch-based bilateral filters, that is, filters that involve a tonal kernel and pair wise comparisons between shifted blocks of the images. The main aim is then to integrate two variance stabilizing transformations that allow the filters to work with Gaussianized noise. The performances of these filters are compared to those of the classical bilateral filter with the same transformations. The experiments include an artificial benchmark as well as a positron emission tomography image.

Evaluation of MVCT protocols for brain and head and neck tumor patients treated with helical tomotherapy.

PURPOSE: Helical tomotherapy is a modality of radiation treatment delivery which is equipped with an on-board imaging device (MVCT) allowing for daily patient set-up verification and correction in the medial-lateral (m-l), cranial-caudal (c-c), anterior-posterior (a-p) and transversal angular (roll) directions. In this study, we measured set-up deviations and evaluated different MVCT protocols for brain and head and neck (H&N) cancer patients. MATERIALS AND METHODS: The daily set-up errors of 75 H&N cancer patients immobilized with 5-point fixation thermoplastic masks and 30 brain cancer patients immobilized with 3-point fixation thermoplastic masks were detected by matching the MVCT with the treatment planning CT images. This co-registration procedure was accomplished automatically by the system's software (automatic deviations), then corrected manually by the radiation therapists (total deviations). Systematic and random errors were analyzed on a patient and a population basis. Moreover, 2 MVCT protocols were retrospectively evaluated; MVCTs were either acquired during the first five fractions (FFFs) or on alternate week (ALT). Systematic deviations were calculated based upon prior "MVCT" fractions and applied during the "non-MVCT" fractions. The resulting residual deviations were then analyzed. RESULTS: The total systematic (and random) deviations reached 1.7mm (1.4mm), 1.6mm (1.5mm), 1.5mm (1.5mm) and 0.6 degrees (0.6 degrees ) for H&N cancer patients and reached 1.6mm (0.9mm), 1.7mm (1.1mm), 1.1mm (0.8mm) and 0.9 degrees (0.6 degrees ) for brain cancer patients in the m-l, c-c, a-p and roll directions, respectively. A t-test detected small but statistically significant differences between the automatic and total deviations. Both MVCT protocols gave rise to similar residual deviations. However, for H&N cancer patients the ALT protocol resulted in smaller residual deviations and CTV-PTV margins, particularly in the a-p direction. CONCLUSION: The total systematic and random deviations were comparable to the previously published data. No clinical difference exists between the automatic and total deviations. Both MVCT protocols were similar. But, for H&N cancer patients, the ALT protocol gave rise to smaller residual deviations and therefore is the correct formula to adopt in order to reduce the frequency of pre-treatment MVCTs.