Octavius 4D characterization for flattened and flattening filter free rotational deliveries

Purpose: In this study the Octavius detector 729 ionization chamber (IC) array with the Octavius 4D phantom was characterized for flattening filter (FF) and flattening filter free (FFF) static and rotational beams. The device was assessed for verification with FF and FFF RapidArc treatment plans.

Methods: The response of the detectors to field size, dose linearity, and dose rate were assessed for 6 MV FF beams and also 6 and 10 MV FFF beams. Dosimetric and mechanical accuracy of the detector array within the Octavius 4D rotational phantom was evaluated against measurements made using semiflex and pinpoint ionization chambers, and radiochromic film. Verification FF and FFF RapidArc plans were assessed using a gamma function with 3%/3 mm tolerances and 2%/2 mm tolerances and further analysis of these plans was undertaken using film and a second detector array with higher spatial resolution.

Results: A warm-up dose of >6 Gy was required for detector stability. Dose-rate measurements were stable across a range from 0.26 to 15 Gy/min and dose response was linear, although the device overestimated small doses compared with pinpoint ionization chamber measurements. Output factors agreed with ionization chamber measurements to within 0.6% for square fields of side between 3 and 25 cm and within 1.2% for 2 x 2 cm(2) fields. The Octavius 4D phantom was found to be consistent with measurements made with radiochromic film, where the gantry angle was found to be within 0.4. of that expected during rotational deliveries. RapidArc FF and FFF beams were found to have an accuracy of >97.9% and >90% of pixels passing 3%/3 mm and 2%/2 mm, respectively. Detector spatial resolution was observed to be a factor in determining the accurate delivery of each plan, particularly at steep dose gradients. This was confirmed using data from a second detector array with higher spatial resolution and with radiochromic film.

Conclusions: The Octavius 4D phantom with associated Octavius detector 729 ionization chamber array is a dosimetrically and mechanically stable device for pretreatment verification of FF and FFF RapidArc treatments. Further improvements may be possible through use of a detector array with higher spatial resolution (detector size and/or detector spacing). (C) 2013 American Association of Physicists in Medicine.

Implementation of phantom-less IMRT delivery verification using Varian DynaLog files and R/V output

This study aims to evaluate the use of Varian radiotherapy dynamic treatment log (DynaLog) files to verify IMRT plan delivery as part of a routine quality assurance procedure. Delivery accuracy in terms of machine performance was quantified by multileaf collimator (MLC) position errors and fluence delivery accuracy for patients receiving intensity modulated radiation therapy (IMRT) treatment. The relationship between machine performance and plan complexity, quantified by the modulation complexity score (MCS) was also investigated. Actual MLC positions and delivered fraction of monitor units (MU), recorded every 50 ms during IMRT delivery, were extracted from the DynaLog files. The planned MLC positions and fractional MU were taken from the record and verify system MLC control file. Planned and delivered beam data were compared to determine leaf position errors with and without the overshoot effect. Analysis was also performed on planned and actual fluence maps reconstructed from the MLC control file and delivered treatment log files respectively. This analysis was performed for all treatment fractions for 5 prostate, 5 prostate and pelvic node (PPN) and 5 head and neck (H&N) IMRT plans, totalling 82 IMRT fields in ∼5500 DynaLog files. The root mean square (RMS) leaf position errors without the overshoot effect were 0.09, 0.26, 0.19 mm for the prostate, PPN and H&N plans respectively, which increased to 0.30, 0.39 and 0.30 mm when the overshoot effect was considered. Average errors were not affected by the overshoot effect and were 0.05, 0.13 and 0.17 mm for prostate, PPN and H&N plans respectively. The percentage of pixels passing fluence map gamma analysis at 3%/3 mm was 99.94 ± 0.25%, which reduced to 91.62 ± 11.39% at 1%/1 mm criterion. Leaf position errors, but not gamma passing rate, were directly related to plan complexity as determined by the MCS. Site specific confidence intervals for average leaf position errors were set at -0.03-0.12 mm for prostate and -0.02-0.28 mm for more complex PPN and H&N plans. For all treatment sites confidence intervals for RMS errors with the overshoot was set at 0-0.50 mm and for the percentage of pixels passing a gamma analysis at 1%/1 mm a confidence interval of 68.83% was set also for all treatment sites. This work demonstrates the successful implementation of treatment log files to validate IMRT deliveries and how dynamic log files can diagnose delivery errors not possible with phantom based QC. Machine performance was found to be directly related to plan complexity but this is not the dominant determinant of delivery accuracy.

Toward Unconstrained Ear Recognition From Two-Dimensional Images

Ear recognition, as a biometric, has several advantages. In particular, ears can be measured remotely and are also relatively static in size and structure for each individual. Unfortunately, at present, good recognition rates require controlled conditions. For commercial use, these systems need to be much more robust. In particular, ears have to be recognized from different angles ( poses), under different lighting conditions, and with different cameras. It must also be possible to distinguish ears from background clutter and identify them when partly occluded by hair, hats, or other objects. The purpose of this paper is to suggest how progress toward such robustness might be achieved through a technique that improves ear registration. The approach focuses on 2-D images, treating the ear as a planar surface that is registered to a gallery using a homography transform calculated from scale-invariant feature-transform feature matches. The feature matches reduce the gallery size and enable a precise ranking using a simple 2-D distance algorithm. Analysis on a range of data sets demonstrates the technique to be robust to background clutter, viewing angles up to +/- 13 degrees, and up to 18% occlusion. In addition, recognition remains accurate with masked ear images as small as 20 x 35 pixels.

Nanoflare Activity In the Solar Chromosphere

We use ground-based images of high spatial and temporal resolution to search for evidence of nanoflare activity in the solar chromosphere. Through close examination of more than 1 x 10(9) pixels in the immediate vicinity of an active region, we show that the distributions of observed intensity fluctuations have subtle asymmetries. A negative excess in the intensity fluctuations indicates that more pixels have fainter-than-average intensities compared with those that appear brighter than average. By employing Monte Carlo simulations, we reveal how the negative excess can be explained by a series of impulsive events, coupled with exponential decays, that are fractionally below the current resolving limits of low-noise equipment on high-resolution ground-based observatories. Importantly, our Monte Carlo simulations provide clear evidence that the intensity asymmetries cannot be explained by photon-counting statistics alone. A comparison to the coronal work of Terzo et al. suggests that nanoflare activity in the chromosphere is more readily occurring, with an impulsive event occurring every similar to 360 s in a 10,000 km(2) area of the chromosphere, some 50 times more events than a comparably sized region of the corona. As a result, nanoflare activity in the chromosphere is likely to play an important role in providing heat energy to this layer of the solar atmosphere.

Region-based Mixture of Gaussians Modelling for Foreground Detection in Dynamic Scenes

One of the most widely used techniques in computer vision for foreground detection is to model each background pixel as a Mixture of Gaussians (MoG). While this is effective for a static camera with a fixed or a slowly varying background, it fails to handle any fast, dynamic movement in the background. In this paper, we propose a generalised framework, called region-based MoG (RMoG), that takes into consideration neighbouring pixels while generating the model of the observed scene. The model equations are derived from Expectation Maximisation theory for batch mode, and stochastic approximation is used for online mode updates. We evaluate our region-based approach against ten sequences containing dynamic backgrounds, and show that the region-based approach provides a performance improvement over the traditional single pixel MoG. For feature and region sizes that are equal, the effect of increasing the learning rate is to reduce both true and false positives. Comparison with four state-of-the art approaches shows that RMoG outperforms the others in reducing false positives whilst still maintaining reasonable foreground definition. Lastly, using the ChangeDetection (CDNet 2014) benchmark, we evaluated RMoG against numerous surveillance scenes and found it to amongst the leading performers for dynamic background scenes, whilst providing comparable performance for other commonly occurring surveillance scenes.

Impact of prosthodontic rehabilitation on the chewing efficiency of partially dentate older patients

Objectives: This study compared two tooth replacement strategies for partially dentate older patients namely; removable partial dentures (RPDs) and functionally orientated treatment based on the shortened dental arch (SDA) concept. Patients were compared in terms of chewing efficiency after prosthodontic rehabilitation. Methods: Chewing efficiency was assessed electronically by a two-colour gum-mixing test. Specimens were assembled from two different colours of chewing gums with a size of 30 x 18 x 3 mm. After participants chewed for 20 cycles, the gum was retrieved, flattened to a 1-mm-thick wafer, and digitized with a flatbed image scanner. The pixels of unmixed colour in the specimen were counted by means of Adobe Photoshop 2.0R software (Adobe Systems, San Jose, CA, USA), and the ratio to the pixels of the entire frame was computed. This ratio is called the Unmixed Fraction (UF). The more efficiently the specimen is chewed, the less unmixed colour remains, and the smaller the gum becomes. Consequently, a low unmixed fraction corresponds to good chewing efficiency. Results: 32 patients completed the chewing efficiency test (17 RPDs and 15 SDA). The mean UF recorded for the SDA group was not significantly different to that recorded for the RPD group (p>0.05, unpaired t-test). Conclusion: These results indicate that prosthodontic rehabilitation according to the principles of the SDA is equivalent to RPDs in terms of restoration of chewing ability for partially dentate older patients.

Machine learning for transient discovery in Pan-STARRS1 difference imaging

Efficient identification and follow-up of astronomical transients is hindered by the need for humans to manually select promising candidates from data streams that contain many false positives. These artefacts arise in the difference images that are produced by most major ground-based time-domain surveys with large format CCD cameras. This dependence on humans to reject bogus detections is unsustainable for next generation all-sky surveys and significant effort is now being invested to solve the problem computationally. In this paper, we explore a simple machine learning approach to real-bogus classification by constructing a training set from the image data of similar to 32 000 real astrophysical transients and bogus detections from the Pan-STARRS1 Medium Deep Survey. We derive our feature representation from the pixel intensity values of a 20 x 20 pixel stamp around the centre of the candidates. This differs from previous work in that it works directly on the pixels rather than catalogued domain knowledge for feature design or selection. Three machine learning algorithms are trained (artificial neural networks, support vector machines and random forests) and their performances are tested on a held-out subset of 25 per cent of the training data. We find the best results from the random forest classifier and demonstrate that by accepting a false positive rate of 1 per cent, the classifier initially suggests a missed detection rate of around 10 per cent. However, we also find that a combination of bright star variability, nuclear transients and uncertainty in human labelling means that our best estimate of the missed detection rate is approximately 6 per cent.

A Novel Semi-Automated Method of Tracking Fetal Movements

We have designed software that can â€â€™look’’ at recorded ultrasound sequences. We analyzed fifteen video sequences representing recorded ultrasound scans of nine fetuses. Our method requires a small amount of user labelled pixels for processing the first frame. These initialize GrowCut 1 , a background removal algorithm, which was used for separating the fetus from its surrounding environment (segmentation). For each subsequent frame, user input is no longer necessary as some of the pixels will inherit labels from the previously processed frame. This results in our software’s ability to track movement. Two sonographers rated the results of our computer’s â€vision’ on a scale from 1 (poor fit) to 10 (excellent fit). They assessed tracking accuracy for the entire video as well as segmentation accuracy (the ability to identify fetus from non-fetus) for every 100th processed frame. There was no appreciable deterioration in the software’s ability to track the fetus over time. I