Combining distance and anatomical information for deep-learning based dose distribution predictions for nasopharyngeal cancer radiotherapy planning

Purpose

Deep-learning effectively predicts dose distributions in knowledge-based radiotherapy planning. Using anatomical information that includes a structure map and computed tomography (CT) data as input has been proven to work well. The minimum distance from each voxel in normal structures to planning target volume (DPTV) closely affects each voxel's dose. In this study, we combined DPTV and anatomical information as input for a deep-learning-based dose-prediction network to improve performance.

Materials and methods

One hundred patients who underwent volumetric-modulated arc therapy for nasopharyngeal cancer were selected in this study. The prediction model based on a residual network had DPTV maps, structure maps, and CT as inputs and the corresponding dose distribution maps as outputs. The performances of the combined distance and anatomical information (COM) model and the traditional anatomical (ANAT) model with two-channel inputs (structure maps and CT) were compared. A 10-fold cross validation was performed to separately train and test the COM and ANAT models. The voxel-based mean error (ME), mean absolute error (MAE), dosimetric parameters, and dice similarity coefficient (DSC) of isodose volumes were used for modeling evaluation.

Results

The mean MAE of the body volume of the COM model were 4.89 ± 1.35%, highly significantly lower than those for the ANAT model of 5.07 ± 1.37% (p<0.001). The ME values of the body for the 2-type models were similar (p >0.05). The mean DSC values of the isodose volumes in the range of 60 Gy were all better in the COM model (p<0.05), and there were highly significant differences between 10 Gy and 55 Gy (p<0.001). For most organs at risk, the ME, MAE, and dosimetric parameters predicted by both models were concurrent with the ground truth values except the MAE values of the pituitary and optic chiasm in the ANAT model and the average mean dose of the right parotid in the ANAT model.

Conclusions

The COM model outperformed the ANAT model and could improve automated planning with statistically highly significant differences.

Lengths for Which Fourth Degree PP Interleavers Lead to Weaker Performances Compared to Quadratic and Cubic PP Interleavers

In this paper, we obtain upper bounds on the minimum distance for turbo codes using fourth degree permutation polynomial (4-PP) interleavers of a specific interleaver length and classical turbo codes of nominal 1/3 coding rate, with two recursive systematic convolutional component codes with generator matrix G=[1,15/13]. The interleaver lengths are of the form 16Ψ or 48Ψ, where Ψ is a product of different prime numbers greater than three. Some coefficient restrictions are applied when for a prime pi∣Ψ, condition 3∤(pi−1) is fulfilled. Two upper bounds are obtained for different classes of 4-PP coefficients. For a 4-PP f4x4+f3x3+f2x2+f1x(mod16kLΨ), kL∈{1,3}, the upper bound of 28 is obtained when the coefficient f3 of the equivalent 4-permutation polynomials (PPs) fulfills f3∈{0,4Ψ} or when f3∈{2Ψ,6Ψ} and f2∈{(4kL−1)·Ψ,(8kL−1)·Ψ}, kL∈{1,3}, for any values of the other coefficients. The upper bound of 36 is obtained when the coefficient f3 of the equivalent 4-PPs fulfills f3∈{2Ψ,6Ψ} and f2∈{(2kL−1)·Ψ,(6kL−1)·Ψ}, kL∈{1,3}, for any values of the other coefficients. Thus, the task of finding out good 4-PP interleavers of the previous mentioned lengths is highly facilitated by this result because of the small range required for coefficients f4,f3 and f2. It was also proven, by means of nonlinearity degree, that for the considered inteleaver lengths, cubic PPs and quadratic PPs with optimum minimum distances lead to better error rate performances compared to 4-PPs with optimum minimum distances.

A superellipsoid-plane model for simulating foot-ground contact during human gait

Musculoskeletal models and forward dynamics simulations of human movement often include foot-ground interactions, with the foot-ground contact forces often determined using a constitutive model that depends on material properties and contact kinematics. When using soft constraints to model the foot-ground interactions, the kinematics of the minimum distance between the foot and planar ground needs to be computed. Due to their geometric simplicity, a considerable number of studies have used point-plane elements to represent these interacting bodies, but few studies have provided comparisons between point contact elements and other geometrically based analytical solutions. The objective of this work was to develop a more general-purpose superellipsoid-plane contact model that can be used to determine the three-dimensional foot-ground contact forces. As an example application, the model was used in a forward dynamics simulation of human walking. Simulation results and execution times were compared with a point-like viscoelastic contact model. Both models produced realistic ground reaction forces and kinematics with similar computational efficiency. However, solving the equations of motion with the surface contact model was found to be more efficient (~18% faster), and on average numerically ~37% less stiff. The superellipsoid-plane elements are also more versatile than point-like elements in that they allow for volumetric contact during three-dimensional motions (e.g. rotating, rolling, and sliding). In addition, the superellipsoid-plane element is geometrically accurate and easily integrated within multibody simulation code. These advantages make the use of superellipsoid-plane contact models in musculoskeletal simulations an appealing alternative to point-like elements.

Semiparametric inference of proportional odds model based on randomly truncated data

This paper studies the estimation in the proportional odds model based on randomly truncated data. The proposed estimators for the regression coefficients include a class of minimum distance estimators defined through weighted empirical odds function. We have investigated the asymptotic properties like the consistency and the limiting distribution of the proposed estimators under mild conditions. The finite sample properties were investigated through simulation study making comparison of some of the estimators in the class. We conclude with an illustration of our proposed method to a well-known AIDS data.