Liyun Chang

Using Multivariate Regression Model with Least Absolute Shrinkage and Selection Operator (LASSO) to Predict the Incidence of Xerostomia after Intensity-Modulated Radiotherapy for Head and Neck Cancer

Purpose The aim of this study was to develop a multivariate logistic regression model with least absolute shrinkage and selection operator (LASSO) to make valid predictions about the incidence of moderate-to-severe patient-rated xerostomia among head and neck cancer (HNC) patients treated with IMRT. Methods and Materials Quality of life questionnaire datasets from 206 patients with HNC were analyzed. The European Organization for Research and Treatment of Cancer QLQ-H&N35 and QLQ-C30 questionnaires were used as the endpoint evaluation. The primary endpoint (grade 3+ xerostomia) was defined as moderate-to-severe xerostomia at 3 (XER3m) and 12 months (XER12m) after the completion of IMRT. Normal tissue complication probability (NTCP) models were developed. The optimal and suboptimal numbers of prognostic factors for a multivariate logistic regression model were determined using the LASSO with bootstrapping technique. Statistical analysis was performed using the scaled Brier score, Nagelkerke R2, chi-squared test, Omnibus, Hosmer-Lemeshow test, and the AUC. Results Eight prognostic factors were selected by LASSO for the 3-month time point: Dmean-c, Dmean-i, age, financial status, T stage, AJCC stage, smoking, and education. Nine prognostic factors were selected for the 12-month time point: Dmean-i, education, Dmean-c, smoking, T stage, baseline xerostomia, alcohol abuse, family history, and node classification. In the selection of the suboptimal number of prognostic factors by LASSO, three suboptimal prognostic factors were fine-tuned by Hosmer-Lemeshow test and AUC, i.e., Dmean-c, Dmean-i, and age for the 3-month time point. Five suboptimal prognostic factors were also selected for the 12-month time point, i.e., Dmean-i, education, Dmean-c, smoking, and T stage. The overall performance for both time points of the NTCP model in terms of scaled Brier score, Omnibus, and Nagelkerke R2 was satisfactory and corresponded well with the expected values. Conclusions Multivariate NTCP models with LASSO can be used to predict patient-rated xerostomia after IMRT.

Calibration of EBT2 film by the PDD method with scanner non-uniformity correction.

The EBT2 film together with a flatbed scanner is a convenient dosimetry QA tool for verification of clinical radiotherapy treatments. However, it suffers from a relatively high degree of uncertainty and a tedious film calibration process for every new lot of films, including cutting the films into several small pieces, exposing with different doses, restoring them back and selecting the proper region of interest (ROI) for each piece for curve fitting. In this work, we present a percentage depth dose (PDD) method that can accurately calibrate the EBT2 film together with the scanner non-uniformity correction and provide an easy way to perform film dosimetry. All films were scanned before and after the irradiation in one of the two homemade 2 mm thick acrylic frames (one portrait and the other landscape), which was located at a fixed position on the scan bed of an Epson 10 000XL scanner. After the pre-irradiated scan, the film was placed parallel to the beam central axis and sandwiched between six polystyrene plates (5 cm thick each), followed by irradiation of a 20 × 20 cm² 6 MV photon beam. Two different beams on times were used on two different films to deliver a dose to the film ranging from 32 to 320 cGy. After the post-irradiated scan, the net optical densities for a total of 235 points on the beam central axis on the films were auto-extracted and compared with the corresponding depth doses that were calculated through the measurement of a 0.6 cc farmer chamber and the related PDD table to perform the curve fitting. The portrait film location was selected for routine calibration, since the central beam axis on the film is parallel to the scanning direction, where non-uniformity correction is not needed (Ferreira et al 2009 Phys. Med. Biol. 54 1073-85). To perform the scanner non-uniformity calibration, the cross-beam profiles of the film were analysed by referencing the measured profiles from a Profiler™. Finally, to verify our method, the films were exposed to 60° physical wedge fields and the compositive fields, and their relative dose profiles were compared with those from the water phantom measurement. The fitting uncertainty was less than 0.5% due to the many calibration points, and the overall calibration uncertainty was within 3% for doses above 50 cGy, when the average of four films were used for the calibration. According to our study, the non-uniformity calibration factor was found to be independent of the given dose for the EBT2 film and the relative dose differences between the profiles measured by the film and the Profiler were within 1.5% after applying the non-uniformity correction. For the verification tests, the relative dose differences between the measurements by films and in the water phantom, when the average of three films were used, were generally within 3% for the 60° wedge fields and compositive fields, respectively. In conclusion, our method is convenient, time-saving and cost-effective, since no film cutting is needed and only two films with two exposures are needed.

Patient- and therapy-related factors associated with the incidence of xerostomia in nasopharyngeal carcinoma patients receiving parotid-sparing helical tomotherapy

We investigated the incidence of moderate to severe patient-reported xerostomia among nasopharyngeal carcinoma (NPC) patients treated with helical tomotherapy (HT) and identified patient- and therapy-related factors associated with acute and chronic xerostomia toxicity. The least absolute shrinkage and selection operator (LASSO) normal tissue complication probability (NTCP) models were developed using quality-of-life questionnaire datasets from 67 patients with NPC. For acute toxicity, the dosimetric factors of the mean doses to the ipsilateral submandibular gland (Dis) and the contralateral submandibular gland (Dcs) were selected as the first two significant predictors. For chronic toxicity, four predictive factors were selected: age, mean dose to the oral cavity (Doc), education, and T stage. The substantial sparing data can be used to avoid xerostomia toxicity. We suggest that the tolerance values corresponded to a 20% incidence of complications (TD20) for Dis = 39.0 Gy, Dcs = 38.4 Gy, and Doc = 32.5 Gy, respectively, when mean doses to the parotid glands met the QUANTEC 25 Gy sparing guidelines. To avoid patient-reported xerostomia toxicity, the mean doses to the parotid gland, submandibular gland, and oral cavity have to meet the sparing tolerance, although there is also a need to take inherent patient characteristics into consideration.

Developing Multivariable Normal Tissue Complication Probability Model to Predict the Incidence of Symptomatic Radiation Pneumonitis among Breast Cancer Patients

Purpose Symptomatic radiation pneumonitis (SRP), which decreases quality of life (QoL), is the most common pulmonary complication in patients receiving breast irradiation. If it occurs, acute SRP usually develops 4–12 weeks after completion of radiotherapy and presents as a dry cough, dyspnea and low-grade fever. If the incidence of SRP is reduced, not only the QoL but also the compliance of breast cancer patients may be improved. Therefore, we investigated the incidence SRP in breast cancer patients after hybrid intensity modulated radiotherapy (IMRT) to find the risk factors, which may have important effects on the risk of radiation-induced complications. Methods In total, 93 patients with breast cancer were evaluated. The final endpoint for acute SRP was defined as those who had density changes together with symptoms, as measured using computed tomography. The risk factors for a multivariate normal tissue complication probability model of SRP were determined using the least absolute shrinkage and selection operator (LASSO) technique. Results Five risk factors were selected using LASSO: the percentage of the ipsilateral lung volume that received more than 20-Gy (IV20), energy, age, body mass index (BMI) and T stage. Positive associations were demonstrated among the incidence of SRP, IV20, and patient age. Energy, BMI and T stage showed a negative association with the incidence of SRP. Our analyses indicate that the risk of SPR following hybrid IMRT in elderly or low-BMI breast cancer patients is increased once the percentage of the ipsilateral lung volume receiving more than 20-Gy is controlled below a limitation. Conclusions We suggest to define a dose-volume percentage constraint of IV20< 37% (or AIV20< 310cc) for the irradiated ipsilateral lung in radiation therapy treatment planning to maintain the incidence of SPR below 20%, and pay attention to the sequelae especially in elderly or low-BMI breast cancer patients. (AIV20: the absolute ipsilateral lung volume that received more than 20 Gy (cc).

Dependency of EBT2 film calibration curve on postirradiation time

PURPOSE The Ashland Inc. product EBT2 film model is a widely used quality assurance tool, especially for verification of 2-dimensional dose distributions. In general, the calibration film and the dose measurement film are irradiated, scanned, and calibrated at the same postirradiation time (PIT), 1-2 days after the films are irradiated. However, for a busy clinic or in some special situations, the PIT for the dose measurement film may be different from that of the calibration film. In this case, the measured dose will be incorrect. This paper proposed a film calibration method that includes the effect of PIT. METHODS The dose versus film optical density was fitted to a power function with three parameters. One of these parameters was PIT dependent, while the other two were found to be almost constant with a standard deviation of the mean less than 4%. The PIT-dependent parameter was fitted to another power function of PIT. The EBT2 film model was calibrated using the PDD method with 14 different PITs ranging from 1 h to 2 months. Ten of the fourteen PITs were used for finding the fitting parameters, and the other four were used for testing the model. RESULTS The verification test shows that the differences between the delivered doses and the film doses calculated with this modeling were mainly within 2% for delivered doses above 60 cGy, and the total uncertainties were generally under 5%. The errors and total uncertainties of film dose calculation were independent of the PIT using the proposed calibration procedure. However, the fitting uncertainty increased with decreasing dose or PIT, but stayed below 1.3% for this study. CONCLUSIONS The EBT2 film dose can be modeled as a function of PIT. For the ease of routine calibration, five PITs were suggested to be used. It is recommended that two PITs be located in the fast developing period (1 ∼ 6 h), one in 1 ∼ 2 days, one around a week, and one around a month.

Face Description with Local Binary Patterns and Local Ternary Patterns: Improving Face Recognition Performance Using Similarity Feature-Based Selection and Classification Algorithm

In recent years, the binary coding of face image features, such as local binary patterns (LBP) and local ternary patterns (LTP) have become popular in face recognition systems. These local feature descriptors provide a simple and powerful means for texture description. In this paper, we present a novel approach, which uses these descriptors to represent face images, and a similarity feature-based selection and classification algorithm to improve recognition rate. The face image is first divided into small regions from which LBP and LTP histograms are extracted and concatenated into a single feature vector. The proposed algorithm is used to select the similarity features of training set and classify the face image. The experiments are conducted on the ORL Database of Faces and the Extended Yale Face Database B. The results clearly show the superiority of the proposed algorithm.

Geometric error of cervical point A calculated through traditional reconstruction procedures for brachytherapy treatment.

Brachytherapy used in local cervical cancer is still widely based on 2D standard dose planning with the prescription to point A, which is invisible on imaging and located at a high-dose gradient. In this study, the geometric location error of point A was investigated. It is traditionally reconstructed in the treatment planning system after carefully digitizing the point marks that were previously drawn on the orthogonal radiographs into the system. Two Cartesian coordinates of point A were established and compared. One was built up based on the geometric definition of point A and would be taken as the true coordinate, while the other was built up through traditional clinical treatment procedures and named as the practical coordinate. The orthogonal film reconstruction technique was used and the location error between the practical and the true coordinate introduced from the variations of, first, the angle between the tandem and the simulator gantry rotation axis, and second, the interval between the tandem flange and the simulator isocenter, was analyzed. The location error of point A was higher if the tandem was rotated away from the gantry rotation axis or if the location of the tandem flange was set away from the isocenter. If a tandem with a 30° curvature was rotated away from the gantry rotation axis 10° in the anterior-posterior (AP) view, and there was an 8.7 cm interval between the flange and the isocenter, the location error of point A would be 3 mm without including other errors from simulator calibration, data input, patient setup, and movements. To reduce the location error of point A calculated for traditional reconstruction procedures, it is suggested to move the couch or patient to make the mid-point of two points A near the isocenter and the tandem in the AP view parallel to the gantry rotation axis as much as possible. PACS number: 87.55.km.Brachytherapy used in local cervical cancer is still widely based on 2D standard dose planning with the prescription to point A, which is invisible on imaging and located at a high‐dose gradient. In this study, the geometric location error of point A was investigated. It is traditionally reconstructed in the treatment planning system after carefully digitizing the point marks that were previously drawn on the orthogonal radiographs into the system. Two Cartesian coordinates of point A were established and compared. One was built up based on the geometric definition of point A and would be taken as the true coordinate, while the other was built up through traditional clinical treatment procedures and named as the practical coordinate. The orthogonal film reconstruction technique was used and the location error between the practical and the true coordinate introduced from the variations of, first, the angle between the tandem and the simulator gantry rotation axis, and second, the interval between the tandem flange and the simulator isocenter, was analyzed. The location error of point A was higher if the tandem was rotated away from the gantry rotation axis or if the location of the tandem flange was set away from the isocenter. If a tandem with a 30° curvature was rotated away from the gantry rotation axis 10° in the anterior–posterior (AP) view, and there was an 8.7 cm interval between the flange and the isocenter, the location error of point A would be 3 mm without including other errors from simulator calibration, data input, patient setup, and movements. To reduce the location error of point A calculated for traditional reconstruction procedures, it is suggested to move the couch or patient to make the mid‐point of two points A near the isocenter and the tandem in the AP view parallel to the gantry rotation axis as much as possible. PACS number: 87.55.km

Relationships among patient characteristics, irradiation treatment planning parameters, and treatment toxicity of acute radiation dermatitis after breast hybrid intensity modulation radiation therapy

To evaluate the relationships among patient characteristics, irradiation treatment planning parameters, and treatment toxicity of acute radiation dermatitis (RD) after breast hybrid intensity modulation radiation therapy (IMRT). The study cohort consisted of 95 breast cancer patients treated with hybrid IMRT. RD grade ≥2 (2+) toxicity was defined as clinically significant. Patient characteristics and the irradiation treatment planning parameters were used as the initial candidate factors. Prognostic factors were identified using the least absolute shrinkage and selection operator (LASSO)-based normal tissue complication probability (NTCP) model. A univariate cut-off dose NTCP model was developed to find the dose-volume limitation. Fifty-two (54.7%) of ninety-five patients experienced acute RD grade 2+ toxicity. The volume of skin receiving a dose >35 Gy (V35) was the most significant dosimetric predictor associated with RD grade 2+ toxicity. The NTCP model parameters for V35Gy were TV50 = 85.7 mL and γ50 = 0.77, where TV50 was defined as the volume corresponding to a 50% incidence of complications, and γ50 was the normalized slope of the volume-response curve. Additional potential predictive patient characteristics were energy and surgery, but the results were not statistically significant. To ensure a better quality of life and compliance for breast hybrid IMRT patients, the skin volume receiving a dose >35 Gy should be limited to <85.7 mL to keep the incidence of RD grade 2+ toxicities below 50%. To avoid RD toxicity, the volume of skin receiving a dose >35 Gy should follow sparing tolerance and the inherent patient characteristics should be considered.

Local intensity area descriptor for facial recognition in ideal and noise conditions

Abstract. We propose a local texture descriptor, local intensity area descriptor (LIAD), which is applied for human facial recognition in ideal and noisy conditions. Each facial image is divided into small regions from which LIAD histograms are extracted and concatenated into a single feature vector to represent the facial image. The recognition is performed using a nearest neighbor classifier with histogram intersection and chi-square statistics as dissimilarity measures. Experiments were conducted with LIAD using the ORL database of faces (Olivetti Research Laboratory, Cambridge), the Face94 face database, the Georgia Tech face database, and the FERET database. The results demonstrated the improvement in accuracy of our proposed descriptor compared to conventional descriptors [local binary pattern (LBP), uniform LBP, local ternary pattern, histogram of oriented gradients, and local directional pattern]. Moreover, the proposed descriptor was less sensitive to noise and had low histogram dimensionality. Thus, it is expected to be a powerful texture descriptor that can be used for various computer vision problems.

An innovative method to acquire the location of point A for cervical cancer treatment by HDR brachytherapy

Brachytherapy of local cervical cancer is generally accomplished through film‐based treatment planning with the prescription directed to point A, which is invisible on images and is located at a high‐dose gradient area. Through a standard reconstruction method by digitizing film points, the location error for point A would be 3 mm with a condition of 30° curvature tandem, which is 10° away from the gantry rotation axis of a simulator, and has an 8.7 cm interval between the flange and the isocenter. To reduce the location error of the reconstructed point A, this paper proposes a method and demonstrates its accuracy. The Cartesian coordinates of point A were derived by acquiring the locations of the cervical os (tandem flange) and a dummy seed located in the tandem above the flange. To verify this analytical method, ball marks in a commercial “Isocentric Beam Checker” were selected to simulate the two points A, the os, and the dummies. The Checker was placed on the simulator couch with its center ball coincident with the simulator isocenter and its rotation axis perpendicular to the gantry rotation axis. With different combinations of the Checker and couch rotation angles, the orthogonal films were shot and all coordinates of the selected points were reconstructed through the treatment planning system and compared with that calculated through the analytical method. The position uncertainty and the deviation prediction of point A were also evaluated. With a good choice of the reference dummy point, the position deviations of point A obtained through this analytical method were found to be generally within 1 mm, with the standard uncertainty less than 0.5 mm. In summary, this new method is a practical and accurate tool for clinical usage to acquire the accurate location of point A for the treatment of cervical cancer patient. PACS number(s): 87.55.km