Translational and rotational localization errors in cone-beam CT based image-guided lung stereotactic radiotherapy

PurposeAccurate localization is crucial in delivering safe and effective stereotactic body radiation therapy (SBRT). The aim of this study was to analyse the accuracy of image-guidance using the cone-beam computed tomography (CBCT) of the VERO system in 57 patients treated for lung SBRT and to calculate the treatment margins.Materials and methodsThe internal target volume (ITV) was obtained by contouring the tumor on maximum and mean intensity projection CT images reconstructed from a respiration correlated 4D-CT. Translational and rotational tumor localization errors were identified by comparing the manual registration of the ITV to the motion-blurred tumor on the CBCT and they were corrected by means of the robotic couch and the ring rotation. A verification CBCT was acquired after correction in order to evaluate residual errors.ResultsThe mean 3D vector at initial set-up was 6.6 ± 2.3 mm, which was significantly reduced to 1.6 ± 0.8 mm after 6D automatic correction. 94% of the rotational errors were within 3°. The PTV margins used to compensate for residual tumor localization errors were 3.1, 3.5 and 3.3 mm in the LR, SI and AP directions, respectively.ConclusionsOn-line image guidance with the ITV–CBCT matching technique and automatic 6D correction of the VERO system allowed a very accurate tumor localization in lung SBRT.     

Heart position variability during voluntary moderate deep inspiration breath-hold radiotherapy for breast cancer determined by repeat CBCT scans

Voluntary moderate deep inspiration breath hold (vmDIBH) in left-sided breast cancer radiotherapy reduces cardiac dose. The aim of this study was to investigate heart position variability in vmDIBH using CBCT and to compare this variability with differences in heart position between vmDIBH and free breathing (FB).For 50 patients initial heart position with respect to the field edge (HP-FE) was measured on a vmDIBH planning CT scan. Breath-hold was monitored using an in-house developed vertical plastic stick. On pre-treatment CBCT scans, heart position variability with respect to the field edge (Δ_HP-FE) was measured, reflecting heart position variability when using an offline correction protocol. After registering the CBCT scan to the planning CT, heart position variability with respect to the chest wall (Δ_HP-CW) was measured, reflecting heart position variability when using an online correction protocol. As a control group, vmDIBH and FB computed tomography (CT) scans were acquired for 30 patients and registering both scans on the chest wall.For 34 out of 50 patients, the average HP-FE and HP-CW increased over the treatment course in comparison to the planning CT. Averaged over all patients and all treatment fractions, the Δ_HP-FE and the Δ_HP-CW was 0.8 ± 4.2 mm (range −9.4–+10.6 mm) and 1.0 ± 4.4 mm (range −8.3–+10.4 mm) respectively. The average gain in heart to chest wall distance was 11.8 ± 4.6 mm when using vmDIBH instead of FB. In conclusion, substantial variability in heart position using vmDIBH was observed during the treatment course.     

Effect of diagnostic cone-beam computed tomography protocols on image quality,patient dose,and lesion detection

ObjectiveTo evaluate the effect of cone-beam computed tomography (CBCT) image acquisition protocols on image quality, lesion detection, delineation, and patient dose.Methods100-patients and a CTDI phantom combined with an electron density phantom were examined using four different CBCT-image acquisition protocols during image-guided transarterial chemoembolization (TACE). Protocol-1 (time: 6 s, tube rotation: 360°), protocol-2 (5 s, 300°), protocol-3 (4 s, 240°) and protocol-4 (3 s, 180°) were used. The protocols were first investigated using a phantom. The protocols that were found to be clinically appropriate in terms of image quality and radiation dose were then assessed on patients. A higher radiation dose and/or a poor image quality were inappropriate for the patient imaging. Patient dose (patient-entrance dose and dose-area product), image quality (Hounsfield Unit, noise, signal-to-noise ratio and contrast-to-noise ratio), and lesion delineation (tumor-liver contrast) were assessed and compared using appropriate statistical tests. Lesion detectability, sensitivity, and predictive values were estimated for CBCT-image data using pre-treatment patient magnetic resonance imaging.ResultsThe estimated patient dose showed no statistical significance (p > 0.05) between protocols-2 and -3; the assessed image quality between these protocols manifested insignificant difference (p > 0.05). Two other phantom protocols were not considered for patient imaging due to significantly higher dose (protocols-1) and poor image quality (protocol-4). Lesion delineation and detection were insignificant (p > 0.05) between protocols-2 and -3. Lesion sensitivities generated were 81–89% (protocol-2) and 81–85% (protocol-3) for different lesion types.ConclusionData acquisition using protocols-2 and -3 provided good image quality, lesion detection and delineation with acceptable patient dose during CBCT-imaging mainly due to similar frame numbers acquired.     

The effect of tumour geometry on the quantification accuracy of planar 123I phantom images

IntroductionAccurate activity quantification is applied in radiation dosimetry. Planar images are important for quantification of whole-body images, enabling assessment of biodistribution from radionuclide administrations. We evaluated the effect of tumour geometry on quantification accuracy of ¹²³I planar phantom studies, including various tumour sizes, tumour-liver distances and two tumour-background ratios.Methods and materialsAn in-house manufactured abdominal phantom was equipped with a liver, different size cylindrical tumours, and a rod for tumour-liver distance variation. The geometric mean method with scatter and attenuation corrections was used for image processing. Scatter and attenuation corrections were made using the triple energy window scatter correction technique and a printed transmission sheet source, respectively. Region definitions for tumour activity distribution compensated for the partial volume effect (PVE). Activity measured in the dose calibrator served as reference for determining quantification accuracy.ResultsThe smallest tumour had the largest percentage deviation with an average activity underestimation of 34.6 ± 1.2%. Activity values for the largest tumour were overestimated by 3.1 ± 3.0%. PVE compensation improved quantification accuracy for all tumour sizes yielding accuracies of <12.4%. Scatter contribution to the tumours from the liver had minimal effect on quantification accuracy at tumour-liver distances >3 cm. With PVE compensation, increased tumour-background ratio resulted in a percentage increase of up to 26.3%.ConclusionWhen applying relevant corrections for scatter, attenuation and PVE without background activity, quantification accuracy of <13% was obtained. We demonstrated the successful implementation of a practical technique to obtain quantitative information from ¹²³I planar images.     

Two examples of indication specific radiation dose calculations in dental CBCT and Multidetector CT scanners

PurposeTo calculate organ doses and estimate the effective dose for justification purposes in patients undergoing orthognathic treatment planning purposes and temporal bone imaging in dental cone beam CT (CBCT) and Multidetector CT (MDCT) scanners.MethodsThe radiation dose to the ICRP reference male voxel phantom was calculated for dedicated orthognathic treatment planning acquisitions via Monte Carlo simulations in two dental CBCT scanners, Promax 3D Max (Planmeca, FI) and NewTom VGi evo (QR s.r.l, IT) and in Somatom Definition Flash (Siemens, DE) MDCT scanner. For temporal bone imaging, radiation doses were calculated via MC simulations for a CBCT protocol in NewTom 5G (QR s.r.l, IT) and with the use of a software tool (CT-expo) for Somatom Force (Siemens, DE). All procedures had been optimized at the acceptance tests of the devices.ResultsFor orthognathic protocols, dental CBCT scanners deliver lower doses compared to MDCT scanners. The estimated effective dose (ED) was 0.32 mSv for a normal resolution operation mode in Promax 3D Max, 0.27 mSv in VGi-evo and 1.18 mSv in the Somatom Definition Flash. For temporal bone protocols, the Somatom Force resulted in an estimated ED of 0.28 mSv while for NewTom 5G the ED was 0.31 and 0.22 mSv for monolateral and bilateral imaging respectively.ConclusionsTwo clinical exams which are carried out with both a CBCT or a MDCT scanner were compared in terms of radiation dose. Dental CBCT scanners deliver lower doses for orthognathic patients whereas for temporal bone procedures the doses were similar.     

Effect of verification imaging on in vivo dosimetry results using Gafchromic EBT3 film

In this work, the apparent treatment dose that kV planar or CBCT imaging contributes to Gafchromic EBT3 film used for in vivo dosimetry, was investigated. Gafchromic EBT3 film pieces were attached to a variety of phantoms and irradiated using the linear accelerator’s built-in kV imaging system, in both kV planar mode and CBCT mode. To evaluate the sensitivity of the film in the clinical scenario where dose contributions are received from both imaging and treatment, additional pieces of film were irradiated using base doses of 50 cGy and then irradiated using selected kV planar and CBCT techniques. For kV planar imaging, apparent treatment doses of up to 3.4 cGy per image pair were seen. For CBCT, apparent treatment doses ranged from 0.22 cGy to 3.78 cGy. These apparent doses were reproducible with and without the inclusion of the 50 cGy base dose. The contribution of apparent treatment dose from both planar kV as well as CBCT imaging can be detected, even in conjunction with an actual treatment dose. The magnitude of the apparent dose was found to be dependent on patient geometry, scanning protocol, and measurement location. It was found that the apparent treatment dose from the imaging could add up to 8% of additional uncertainty to the in vivo dosimetry result, if not taken into account. It is possible for this apparent treatment dose to be accounted for by subtraction of the experimentally determined apparent doses from in vivo measurements, as demonstrated in this work.     

Estimation of optimal matching position for orthogonal kV setup images and minimal setup margins in radiotherapy of whole breast and lymph node areas

AimThe aim was to find an optimal setup image matching position and minimal setup margins to maximally spare the organs at risk in breast radiotherapy.BackgroundRadiotherapy of breast cancer is a routine task but has many challenges. We investigated residual position errors in whole breast radiotherapy when orthogonal setup images were matched to different bony landmarks.Materials and methodsA total of 1111 orthogonal setup image pairs and tangential field images were analyzed retrospectively for 50 consecutive patients. Residual errors in the treatment field images were determined by matching the orthogonal setup images to the vertebrae, sternum, ribs and their compromises. The most important region was the chest wall as it is crucial for the dose delivered to the heart and the ipsilateral lung. Inter-observer variation in online image matching was investigated.ResultsThe best general image matching position was the compromise of the vertebrae, ribs and sternum, while the worst position was the vertebrae alone (p ≤ 0.03). The setup margins required for the chest wall varied from 4.3 mm to 5.5 mm in the lung direction while in the superior–inferior (SI) direction the margins varied from 5.1 mm to 7.6 mm. The inter-observer variation increased the minimal margins by approximately 1 mm. The margin of the lymph node areas should be at least 4.8 mm.ConclusionsSetup margins can be reduced by proper selection of a matching position for the orthogonal setup images. To retain the minimal margins sufficient, systematic error of the chest wall should not exceed 4 mm in the tangential field image.     

Virtual patient 3D dose reconstruction using in air EPID measurements and a back-projection algorithm for IMRT and VMAT treatments

PurposeAt our institute, a transit back-projection algorithm is used clinically to reconstruct in vivo patient and in phantom 3D dose distributions using EPID measurements behind a patient or a polystyrene slab phantom, respectively. In this study, an extension to this algorithm is presented whereby in air EPID measurements are used in combination with CT data to reconstruct ‘virtual’ 3D dose distributions. By combining virtual and in vivo patient verification data for the same treatment, patient-related errors can be separated from machine, planning and model errors.Methods and materialsThe virtual back-projection algorithm is described and verified against the transit algorithm with measurements made behind a slab phantom, against dose measurements made with an ionization chamber and with the OCTAVIUS 4D system, as well as against TPS patient data. Virtual and in vivo patient dose verification results are also compared.ResultsVirtual dose reconstructions agree within 1% with ionization chamber measurements. The average γ-pass rate values (3% global dose/3 mm) in the 3D dose comparison with the OCTAVIUS 4D system and the TPS patient data are 98.5 ± 1.9%(1SD) and 97.1 ± 2.9%(1SD), respectively. For virtual patient dose reconstructions, the differences with the TPS in median dose to the PTV remain within 4%.ConclusionsVirtual patient dose reconstruction makes pre-treatment verification based on deviations of DVH parameters feasible and eliminates the need for phantom positioning and re-planning. Virtual patient dose reconstructions have additional value in the inspection of in vivo deviations, particularly in situations where CBCT data is not available (or not conclusive).     

Megavoltage 2D topographic imaging: An attractive alternative to megavoltage CT for the localization of breast cancer patients treated with TomoDirect

PurposeTo show the usefulness of topographic 2D megavoltage images (MV2D) for the localization of breast cancer patients treated with TomoDirect (TD), a radiotherapy treatment technique with fixed-angle beams performed on a TomoTherapy system.MethodsA method was developed to quickly localize breast cancer patients treated with TD by registering the MV2D images produced before a TD treatment with reference images reconstructed from a kilovoltage CT simulation scanner and by using the projection of the beam-eye-view TD treatment field. Dose and image quality measurements were performed to determine the optimal parameters for acquiring MV2D images. A TD treatment was simulated on a chest phantom equipped with a breast attachment. MVCT and MV2D images were performed for 7 different shifted positions of the phantom and registered by 10 different operators with the simulation kilovoltage CT images.ResultsCompared to MVCT, MV2D imaging reduces the dose by a factor of up to 45 and the acquisition time by a factor of up to 49. Comparing the registration shift values obtained for the phantom images obtained with MVCT in the coarse mode to those obtained with MV2D, the mean difference is 1.0 ± 1.1 mm, −1.1 mm ± 1.1, and −0.1 ± 2.2 mm, respectively, in the lateral, longitudinal, and vertical directions.ConclusionsWith dual advantages (very fast imaging and a potentially reduced dose to the heart and contralateral organs), MV2D topographic images may be an attractive alternative to MVCT for the localization of breast cancer patients treated with TomoDirect.     

The importance of image guided radiotherapy in small cell lung cancer: Case report and review of literature

AimDescribe the anatomical changes and tumor displacement due to a rapid response of a patient’s small cell lung cancer (SCLC) during definitive chemoradiotherapy (CRT).BackgroundThe treatment for SCLC is based on CRT. If interfractional changes during RT are incorrectly assessed they might compromise adequate coverage of the tumor or increase dose to organs at risk. Image guided RT with cone-beam computed tomography (CBCT) allows to identify daily treatment variations.Material and methodsDescribe a SCLC case with rapid changes in size, shape and location of the primary tumor during RT.Case reportA 62-year-old woman was diagnosed with SCLC with complete obstruction of the anterior and lingular bronchi and incomplete left thorax expansion due to a 12 × 15 cm mass. During CRT (45 Gy in 1.5 Gy per fraction, twice daily) the patient presented rapid tumor response, leading to resolution of bronchi obstruction and hemithorax expansion. Tumor shifted up to 4 cm from its original position. The identification of variations led to two new simulations and planning in a 3-week treatment course.ConclusionsThe complete radiological response was possible due to systematic monitoring of the tumor during CRT. We recommend frequent on-site image verification. Daily CBCT should be considered with pretreatment tumor obstruction, pleural effusion, atelectasis, large volumes or radiosensitive histology that might resolve early and rapidly and could lead to a miss of the tumor or increased toxicity. Further research should be made in replanning effect in coverage of microscopic disease since it increases uncertainty in this scenario.     

Investigating the spatial accuracy of CBCT-guided cranial radiosurgery: A phantom end-to-end test study

PurposeTo evaluate the spatial accuracy of a frameless cone-beam computed tomography (CBCT)-guided cranial radiosurgery (SRS) using an end-to-end (E2E) phantom test methodology.Methods and materialsFive clinical SRS plans were mapped to an acrylic phantom containing a radiochromic film. The resulting phantom-based plans (E2E plans) were delivered four times. The phantom was setup on the treatment table with intentional misalignments, and CBCT-imaging was used to align it prior to E2E plan delivery. Comparisons (global gamma analysis) of the planned and delivered dose to the film were performed using a commercial triple-channel film dosimetry software. The necessary distance-to-agreement to achieve a 95% (DTA95) gamma passing rate for a fixed 3% dose difference provided an estimate of the spatial accuracy of CBCT-guided SRS. Systematic (∑) and random (σ) error components, as well as 95% confidence levels were derived for the DTA95 metric.ResultsThe overall systematic spatial accuracy averaged over all tests was 1.4 mm (SD: 0.2 mm), with a corresponding 95% confidence level of 1.8 mm. The systematic (Σ) and random (σ) spatial components of the accuracy derived from the E2E tests were 0.2 mm and 0.8 mm, respectively.ConclusionsThe E2E methodology used in this study allowed an estimation of the spatial accuracy of our CBCT-guided SRS procedure. Subsequently, a PTV margin of 2.0 mm is currently used in our department.     

Measurement of coronary bifurcation angle with coronary CT angiography: A phantom study

PurposeAccurate determination of the bifurcation angle and correlation with plaque buildup may lead to the prediction of coronary artery disease (CAD). This work evaluates two techniques to measure bifurcation angles in 3D space using coronary computed tomography angiography (CCTA).Materials and MethodsNine phantoms were fabricated with different bifurcation angles ranging from 55.3° to 134.5°. General X-ray and CCTA were employed to acquire 2D and 3D images of the bifurcation phantoms, respectively. Multiplanar reformation (MPR) and volume rendering technique (VRT) were used to measure the bifurcation angle between the left anterior descending (LAD) and left circumflex arteries (LCx). The measured angles were compared with the true values to determine the accuracy of each measurement technique. Inter-observer variability was evaluated. The two techniques were further applied on 50 clinical CCTA cases to verify its clinical value.ResultsIn the phantom setting, the mean absolute differences calculated between the true and measured angles by MPR and VRT were 2.4° ± 2.2° and 3.8° ± 2.9°, respectively. Strong correlation was found between the true and measured bifurcation angles. Furthermore, no significant differences were found between the bifurcation angles measured using either technique. In clinical settings, large difference of 12.0° ± 10.6° was found between the two techniques.ConclusionIn the phantom setting, both techniques demonstrated a significant correlation to the true bifurcation angle. Despite the lack of agreement of the two techniques in the clinical context, our findings in phantoms suggest that MPR should be preferred to VRT for the measurement of coronary bifurcation angle by CCTA.     

CT dose reduction using Automatic Exposure Control and iterative reconstruction: A chest paediatric phantoms study

PurposeTo evaluate the impact of Automatic Exposure Control (AEC) on radiation dose and image quality in paediatric chest scans (MDCT), with or without iterative reconstruction (IR).MethodsThree anthropomorphic phantoms representing children aged one, five and 10-year-old were explored using AEC system (CARE Dose 4D) with five modulation strength options. For each phantom, six acquisitions were carried out: one with fixed mAs (without AEC) and five each with different modulation strength. Raw data were reconstructed with Filtered Back Projection (FBP) and with two distinct levels of IR using soft and strong kernels. Dose reduction and image quality indices (Noise, SNR, CNR) were measured in lung and soft tissues. Noise Power Spectrum (NPS) was evaluated with a Catphan 600 phantom.ResultsThe use of AEC produced a significant dose reduction (p < 0.01) for all anthropomorphic sizes employed. According to the modulation strength applied, dose delivered was reduced from 43% to 91%. This pattern led to significantly increased noise (p < 0.01) and reduced SNR and CNR (p < 0.01). However, IR was able to improve these indices. The use of AEC/IR preserved image quality indices with a lower dose delivered. Doses were reduced from 39% to 58% for the one-year-old phantom, from 46% to 63% for the five-year-old phantom, and from 58% to 74% for the 10-year-old phantom. In addition, AEC/IR changed the patterns of NPS curves in amplitude and in spatial frequency.ConclusionsIn chest paediatric MDCT, the use of AEC with IR allows one to obtain a significant dose reduction while maintaining constant image quality indices.     

Exit fluence analysis using portal dosimetry in volumetric modulated arc therapy

AimIn measuring exit fluences, there are several sources of deviations which include the changes in the entrance fluence, changes in the detector response and patient orientation or geometry. The purpose of this work is to quantify these sources of errors.BackgroundThe use of the volumetric modulated arc therapy treatment with the help of image guidance in radiotherapy results in high accuracy of delivering complex dose distributions while sparing critical organs. The transit dosimetry has the potential of Verifying dose delivery by the linac, Multileaf collimator positional accuracy and the calculation of dose to a patient or phantom.Materials and methodsThe quantification of errors caused by a machine delivery is done by comparing static and arc picket fence test for 30 days. A RapidArc plan, created for the pelvis site was delivered without and with Rando phantom and exit portal images were acquired. The day to day dose variation were analysed by comparing the daily exit dose images during the course of treatment. The gamma criterion used for analysis is 3% dose difference and 3 mm distance to agreement with a threshold of 10% of maximum dose.ResultsThe maximum standard deviation for the static and arc picket fence test fields were 0.19 CU and 1.3 CU, respectively. The delivery of the RapidArc plans without a phantom shows the maximum standard deviation of 1.85 CU and the maximum gamma value of 0.59. The maximum gamma value for the RapidArc plan delivered with the phantom was found to be 1.2. The largest observed fluence deviation during the delivery to patient was 5.7% and the maximum standard deviation was 4.1 CU.ConclusionIt is found from this study that the variation due to patient anatomy and interfraction organ motion is significant.     

EPID-based in vivo dosimetry for stereotactic body radiotherapy of non-small cell lung tumors: Initial clinical experience

PurposeEPID-based in vivo dosimetry (IVD) has been implemented for stereotactic body radiotherapy treatments of non-small cell lung cancer to check both isocenter dose and the treatment reproducibility comparing EPID portal images.Methods15 patients with lung tumors of small dimensions and treated with volumetric modulated arc therapy were enrolled for this initial experience. IVD tests supplied ratios R between in vivo reconstructed and planned isocenter doses. Moreover a γ-like analysis between daily EPID portal images and a reference one, in terms of percentage of points with γ-value smaller than 1, P_γ<1, and mean γ-values, γ_mean, using a local 3%–3 mm criteria, was adopted to check the treatment reproducibility. Tolerance levels of 5% for R ratio, P_γ<1 higher than 90% and γ_mean lower than 0.67 were adopted.ResultsA total of 160 EPID images, two images for each therapy session, were acquired during the treatment of the 15 patients. The overall mean of the R ratios was equal to 1.005 ± 0.014 (1 SD), with 96.9% of tests within ± 5%. The 2 D image γ-like analysis showed an overall γ_mean of 0.39 ± 0.12 with 96.1% of tests within the tolerance level, and an average P_γ<1 value equal to 96.4 ± 3.6% with 95.4% of tests with P_γ<1 > 90%. Paradigmatic discrepancies were observed in three patients: a set-up error and a patient morphological change were identified thanks to CBCT image analysis whereas the third discrepancy was not fully justified.ConclusionsThis procedure can provide improved patient safety as well as a first step to integrate IVD and CBCT dose recalculation.     

The interaction between body position and vibration frequency on acute response to whole body vibration

PurposeThe present study was designed to investigate the electromyographic (EMG) response in leg muscles to whole-body vibration while using different body positions and vibration frequencies.MethodsTwenty male sport sciences students voluntarily participated in this single-group, repeated-measures study in which EMG data from the vastus lateralis (VL) and the lateral gastrocnemius (LG) were collected over a total of 36 trials for each subject (4 static positions × 9 frequencies).ResultsWe found that vibration frequency, body position and the muscle stimulated had a significant effect (P-values ranged from 0.001 to 0.031) on the EMG response. Similarly, the muscle × frequency and position × muscle interactions were significant (P < 0.001). Interestingly, the frequency × positions interactions were not significant (P > 0.05).ConclusionsOur results indicate that lower frequencies of vibration (25–35 Hz) result in maximal activation of LG, whereas higher frequencies (45–55 Hz) elicit the highest responses in the VL. In addition, the position P2 (half squat position with the heels raised) is beneficial both for VL and LG, independently of the vibration frequency.     

Few-view CT reconstruction via a novel non-local means algorithm

PurposeNon-local means (NLM) based reconstruction method is a promising algorithm for few-view computed tomography (CT) reconstruction, but often suffers from over-smoothed image edges. To address this problem, an adaptive NLM reconstruction method based on rotational invariance (ART-RIANLM) is proposed.MethodsThe method consists of four steps: 1) Initializing parameters; 2) ART reconstruction using raw data; 3) Positivity constraint of the reconstructed image; 4) Image updating by RIANLM filtering. In RIANLM, two kinds of rotational invariance measures which are average gradient (AG) and region homogeneity (RH) are proposed to calculate the distance between two patches and a novel NLM filter is developed to avoid over-smoothed image. Moreover, the parameter h in RIANLM which controls the decay of the weights is adaptive to avoid over-smoothness, while it is constant in NLM during the whole reconstruction process. The proposed method is validated on two digital phantoms and real projection data.ResultsIn our experiments, the searching neighborhood size is set as 15 × 15 and the similarity window is set as 3 × 3. For the simulated case of Shepp-Logan phantom, ART-RIANLM produces higher SNR (36.23 dB > 24.00 dB) and lower MAE (0.0006 < 0.0024) reconstructed images than ART-NLM. The visual inspection demonstrated that the proposed method could suppress artifacts or noises more effectively and recover image edges better. The result of real data case is also consistent with the simulation result.ConclusionsA RIANLM based reconstruction method for few-view CT is presented. Compared to the traditional ART-NLM method, SNR and MAE from ART-RIANLM increases 51% and decreases 75%, respectively.     

Bone orientation and position estimation errors using Cosserat point elements and least squares methods: Application to gait

The aim of this study was to analyze the accuracy of bone pose estimation based on sub-clusters of three skin-markers characterized by triangular Cosserat point elements (TCPEs) and to evaluate the capability of four instantaneous physical parameters, which can be measured non-invasively in vivo, to identify the most accurate TCPEs. Moreover, TCPE pose estimations were compared with the estimations of two least squares minimization methods applied to the cluster of all markers, using rigid body (RBLS) and homogeneous deformation (HDLS) assumptions. Analysis was performed on previously collected in vivo treadmill gait data composed of simultaneous measurements of the gold-standard bone pose by bi-plane fluoroscopy tracking the subjects' knee prosthesis and a stereophotogrammetric system tracking skin-markers affected by soft tissue artifact. Femur orientation and position errors estimated from skin-marker clusters were computed for 18 subjects using clusters of up to 35 markers. Results based on gold-standard data revealed that instantaneous subsets of TCPEs exist which estimate the femur pose with reasonable accuracy (median root mean square error during stance/swing: 1.4/2.8 deg for orientation, 1.5/4.2 mm for position). A non-invasive and instantaneous criteria to select accurate TCPEs for pose estimation (4.8/7.3 deg, 5.8/12.3 mm), was compared with RBLS (4.3/6.6 deg, 6.9/16.6 mm) and HDLS (4.6/7.6 deg, 6.7/12.5 mm). Accounting for homogeneous deformation, using HDLS or selected TCPEs, yielded more accurate position estimations than RBLS method, which, conversely, yielded more accurate orientation estimations. Further investigation is required to devise effective criteria for cluster selection that could represent a significant improvement in bone pose estimation accuracy.     

A deep learning classifier for digital breast tomosynthesis

PurposeTo develop a computerized detection system for the automatic classification of the presence/absence of mass lesions in digital breast tomosynthesis (DBT) annotated exams, based on a deep convolutional neural network (DCNN).Materials and MethodsThree DCNN architectures working at image-level (DBT slice) were compared: two state-of-the-art pre-trained DCNN architectures (AlexNet and VGG19) customized through transfer learning, and one developed from scratch (DBT-DCNN). To evaluate these DCNN-based architectures we analysed their classification performance on two different datasets provided by two hospital radiology departments. DBT slice images were processed following normalization, background correction and data augmentation procedures. The accuracy, sensitivity, and area-under-the-curve (AUC) values were evaluated on both datasets, using receiver operating characteristic curves. A Grad-CAM technique was also implemented providing an indication of the lesion position in the DBT slice.Results Accuracy, sensitivity and AUC for the investigated DCNN are in-line with the best performance reported in the field. The DBT-DCNN network developed in this work showed an accuracy and a sensitivity of (90% ± 4%) and (96% ± 3%), respectively, with an AUC as good as 0.89 ± 0.04. A k-fold cross validation test (with k = 4) showed an accuracy of 94.0% ± 0.2%, and a F1-score test provided a value as good as 0.93 ± 0.03. Grad-CAM maps show high activation in correspondence of pixels within the tumour regions.Conclusions We developed a deep learning-based framework (DBT-DCNN) to classify DBT images from clinical exams. We investigated also a possible application of the Grad-CAM technique to identify the lesion position.