Objective To explore the sensitivity of volumetric modulated arc therapy(VMAT)plan to setup errors in radiotherapy for brain metastases of various sizes.Methods Five spheres were set to simulate multiple brain metastases in the CT image,with the volumes of 0.5,1.7,4.0,8.1,and 14.0 cm3,respectively.A single isocenter 5-arc non-coplanar VMAT plan was generated,and the distance from isocenter to each target was 5 cm.The rotational setup errorθ around the Z axis was simulated by rotating the treatment couch(clockwise from-3.0°to 3.0°,with a step length of 0.5°);and the translational setup error(from-3.0 mm to 3.0 mm,with a step length of 0.5 mm)was simulated by changing the position of the isocenter on the three-dimensional axis(X/Y/Z).The dose distribution was recalculated in the optimization conditions remaining unchanged.The target coverage after rotation and translation was recorded and normalized to the relative coverage without setup error.The correlation between the target coverage of brain metastases of various sizes and the setup error was analyzed with linear regression method.Results The relative coverage decreased gradually as rotational setup error increased.With the same rotational setup error,the relative coverage was in linear correlation with target volume(θ=0.5°,P=0.006;θ=1.0°,P=0.024;θ=1.5°,P=0.028;θ=2.0°,P=0.019;θ=2.5°,P=0.014;θ=3.0°,P=0.007),and the relative coverage increased with increasing target volume.When the rotational setup error was less than 0.5°,the relative coverage was less than 2%regardless of whether the target area was large(14.0 cm3)or small(0.5 cm3).The relative coverage decreased gradually as translational setup error increased.Compared with large target area,small target area was more sensitive to translational setup error.When the translational setup error was less than 0.5 mm,the relative coverage of the large target area varied less than 2%,while the relative coverage of the small target area(0.5 cm3)decreased by nearly 5%.Conclusion When the rotational setup error is constant,the relative coverage of the target has linear relationship with target volume.Small target area is more sensitive to translational setup error than large target area.The rotational setup error within 0.5°and the translational setup error within 0.5 mm are recommended when implementing SRS VMAT plan for small target area.

Objective On the premise of meeting the image quality requirements of simulated location for pediatric radiotherapy,the simulated localizable CT parameters are optimized through phantom scanning to reduce the radiation dose.Methods CatPhan700 phantom was used to simulate the child's body,Philip 24-row large-aperture spiral simulated localizable CT was performed,and the CT images were obtained by scanning the phantom at different mAs and tube voltages.The mAs range was set at 60-400 mAs,the scanning was performed every 20 mAs interval,and the kV was set at 80,100,and 120 kV.Image evaluation was carried out using parameters such as image noise(N10 and mean SD),uniformity,low contrast resolution,high contrast resolution,and the stabilities of HU values of Air,Acrylic,50%bone,LDPE,20%bone,Teflon,Polystyrene,DelrinTM,Lung,PMP and Water.The CTDIVol and DLP automatically calculated by the simulated localizable CT system were read to evaluate the radiation dose.Results At 100 kV,as mAs increased,both CTDI and DLP showed upward trends,and the fitting results were linear correlated,with slopes of 0.034 5 and 0.932 4.Image noise was decreased nonlinearly with the increasing mAs.When mAs increased from 60 to 140 mAs,N10 decreased from 0.25%to 0.14%,and SD reduced from 3.74 HU to 2.54 HU.When mAs reached 180 mAs or higher,N10 fluctuated between 0.1%and 0.12%,the mean SD fluctuated between 2.0 and 2.5 HU,and the downward trends obviously slowed down.When mAs increased from 60 to 200 mAs,the low contrast resolution of the image dropped from 0.53 to 0.29.The image uniformity,high contrast resolution and HU values of different substances were less affected by mAs.The image quality of 100 kV and 200 mAs scanning was close to that of 120 kV scanning,but the image quality of 80 kV scanning failed to meet the clinical requirements.Conclusion In order to reduce the radiation dose as much as possible,the mAs should be set at 200 mAs when the tube voltage is set at 100 kV for a simulated cylinder with a diameter of 20 cm.In the actual simulation scanning for pediatric radiotherapy,the scanning parameters should be fine-tuned according to the phantom results and the actual physical characteristics of children to satisfy the optimization principle for radiation protection.

Objective To determine the applicable conditions and the film reading time of Gafchromic EBT3 film dosimeter through the research and test of the measurement performance such as blackening time,dosimetry range,dosimetry resolution and dose response non-uniformity.Methods Three Gafchromic films were exposure to different doses,and the optical densities were read at different moments after irradiation for analyzing and confirming the effects of blackening time on dose measurement and different doses on blackening time.The absorbed dose-optical density curve was used to determine Gafchromic film dosimetry range.The Gafchromic film dosimetry resolution was obtained by calculating the average optical densities of two groups with a dose difference of 0.01 Gy.The dose response non-uniformity of Gafchromic film was evaluated by irradiating the whole film and the cutting film.Results Relative to the optical density at the termination moment of irradiation,the optical density changes at 4 h were 1.69%(2 Gy),2.53%(4 Gy)and 2.13%(8 Gy);the changes at 24 h were 2.91%(2 Gy),3.31%(4 Gy)and 3.20%(8 Gy);and the changes at 48 h were 2.91%(2 Gy),3.31%(4 Gy)and 3.96%(8 Gy).The optical density given by Gafchromic film dosimeter within 0.1 to 10.0 Gy had a linear relationship with the absorbed dose,while the optical density of Gafchromic film dosimeter within 10.0 to 30.0 Gy had a nonlinear relationship with the absorbed dose,but there was still a one-to-one correspondence.The difference of the average optical densities of two groups was 0.004 when the inter-group dose difference was 0.01 Gy.The non-uniformity was 2.2%for the whole film and 2.8%for the cutting film.Conclusion Gafchromic EBT3 film reading time can be set at 24 h after irradiation,and the dose is independent of blackening time.Gafchromic film dosimeter can be used for point dosimetry and dose distribution measurement within 0.1 to 10.0 Gy,and point dosimetry within 10.0 to 30.0 Gy.The dosimetry resolution of the Gafchromic film dosimeter is better than 0.01 Gy.To avoid the effect of scattered rays,the cutting film is recommended for testing dose response non-uniformity.

Objective To evaluate and analyze the accuracies of 3 software solutions based on deep learning techniques in the automatic segmentation of head and neck organs at risk(OAR).Methods The automatic segmentation accuracies of 3 software(PV-iCurve,RT-Mind,and AccuContour)were evaluated with Dice similarity coefficient(DSC),Hausdorff distance(HD),center of mass deviation(COMD),false negative rate(FNR),false positive rate(FPR),Jaccard coefficient(JC),sensitivity index(SI),and inclusive index(II)using the manual contours of head and neck OAR as the gold standard.Results The FNR,JC,SI of brain,the FPR,II of brainstem,the FPR,FNR,JC,SI of eye_L,the FPR,FNR,SI,II of mandible,the FPR,FNR,SI,II of parotid_L,and the DSC,FPR,JC,II of spinal cord manifested significant differences among the 3 software(P<0.05);but the HD,FNR,SI of brainstem,and the HD of spinal cord revealed trivial differences among the 3 software(P>0.05).Conclusion Through the comparison of multiple parameters,it is found that the accuracies of 3 software are different in OAR segmentation,which makes it difficult to make overall horizontal comparisons.Therefore,these parameters are for reference only and cannot be used as criteria for evaluating the segmentation results in clinic.Although all 3 software achieve preferable segmentation outcomes,scrutiny and manual modifications before clinical practice are still necessary.

To enhance the spatial resolution of vibro-acoustography,an improved vibro-acoustography system is constructed using a large-angle dual-frequency focused transducer.Based on this system,mimicking phantoms are used as the experimental objects to explore the spatial resolution of vibro-acoustography,and test the vibro-acoustography results of mimicking phantoms with different acoustic characteristics.Additionally,the effects of the position of hydrophone and difference frequency on the vibro-acoustography results are studied,and the effects of transducer component distribution on vibro-acoustography results are analyzed with a simulation model established with k-Wave.The research results indicate that the vibro-acoustography system with a large-angle dual-frequency focused transducer has higher spatial resolution,contrast and specificity in imaging mimicking phantoms with different acoustic characteristics.The choice of difference frequency is critical for the imaging quality,but the position of the hydrophone and transducer component distribution have trivial effects on vibro-acoustography results.The study is expected to provide new ideas for the development of vibro-acoustography technology,and to promote its further practical application.

Despite significant achievements of deep learning in medical image segmentation,there are challenges for brain tumor segmentation using deep learning,such as insufficient receptive field,excessive redundant information,and information loss.To address these issues,a novel brain tumor segmentation network model(EAU-Net)is proposed based on encoder-decoder structure.EAU-Net incorporates an effective receptive field expansion block and an attention fusion module to minimize the adverse effects caused by insufficient receptive field and excessive redundant information which often occurred in the current brain tumor segmentation network.Additionally,a bottleneck resampling module based on inverted residual structure is introduced to effectively avoid information loss during upsampling and downsampling,while deep convolutions are used to reduce computational complexity.Experimental results on the BraTS2020 dataset reveal that EAU-Net achieves the highest segmentation accuracy,demonstrating its feasibility and effectiveness for brain tumor segmentation.

In view of complex pathogenic factors and diagnostic difficulties of benign paroxysmal positional vertigo,a novel method for diagnosing nystagmus in vertigo based on non-local convolutional and convolutional block attention module(CBAM)is proposed.An object detection model is constructed to locate the pupil,thereby tracking eye movement and extract temporal data of horizontal and vertical motion trajectories.Subsequently,a classification model is employed for detection and classification,utilizing a non-local convolutional module to capture remote dependency relationships in nystagmus data,and introducing CBAM to extract high-and low-level semantic information in the feature layer for enhancing the detection performance.Experiments were conducted on a video nystagmus dataset provided by the Eye,Ear,Nose and Throat Hospital.The results show that compared with the best mainstream method,the proposed method improves precision,recall rate,accuracy,and average F1 score by 1.82%,2.09%,1.62%,and 1.96%,respectively,demonstrating its superiority.

Based on the analysis of heart rate variability(HRV),a prediction method for paroxysmal atrial fibrillation(PAF)attacks is proposed.A new adaptive filtering technique is used for smoothing and coarse graining of HRV,followed by entropy-based quantification of HRV complexity at multiple adaptive scales.After the features are normalized by Min-Max,feature subsets are selected by sequential forward selection method,and then input to support vector machine to identify HRV types and predict PAF attacks.Through 5-fold cross-validation on a set of 50 HRV sequences each lasting 5 minutes,the optimal prediction results are obtained:98%accuracy,100%sensitivity,96%specificity,demonstrating excellent performance.In addition,the experiment shows significant changes(P<0.05)in the complexity eigenvalues of HRV far away from and close to PAF at different frequency bands,reflecting alterations in nervous system regulation of cardiac rhythm and a decline in the ability to adapt to external environmental changes such as stress regulation.

Whether alpha wave synchronization is correlated with stimulus color is explored under the stimulus modes of binocular homo-frequency and binocular dual-frequency in low-,medium-or high-frequency ranges which are(fI±3),(2fI±3),and(4fI±3)Hz,respectively(colors being red,green,blue,where fI represents the spontaneous alpha wave frequency).Normalized Shannon entropy is served as an indicator to evaluate the degree of alpha wave synchronization,and the effects of stimulus modes and color combinations on alpha wave synchronization are investigated.The results indicate that under the stimulus modes of binocular homo-frequency and binocular dual-frequency in the low-,medium-or high-frequency ranges,regardless of red,green or blue light,the degree of alpha wave phase synchronization varies in an"Arnold's tongue"shape,with red>green>blue(in term of synchronization degree).Under the stimulus of binocular dual-frequency and different color lights,a change in the stimulus frequency of one eye affects the alpha wave synchronization of the other eye.These findings offer new insights for the diagnosis and treatment of cognitive impairment,and the combination of stimulus frequency,color and mode can address limitations in stimulus frequency selection for brain-computer interface technology.

To achieve non-invasive detection of hemoglobin concentration,a hemoglobin concentration detection method based on support vector regression is designed.A mathematical model for non-invasive hemoglobin detection is established based on the Beer-Lambert law.After removing the noise and baseline drift from the collected photoplethysmography signals,hemoglobin concentration information is extracted,and a recursive feature elimination algorithm is used to select the extracted information and eliminate redundant features.Finally,29 key features are identified as input to construct a hemoglobin prediction model using support vector regression algorithm.Experimental validation is conducted on 249 clinical data samples(199 cases in training dataset and 50 in test dataset),resulting in a root mean square error of 1.83 g/dL between predicted values and references,with a correlation coefficient of 0.75(P<0.01),demonstrating the high consistency of the proposed method and traditional invasive detection methods.