Deep Learning-Based Automatic Assessment of Radiation Dermatitis in Patients With Nasopharyngeal Carcinoma.

PURPOSE: Radiation dermatitis (RD) is a common, unpleasant side effect of patients receiving radiation therapy. In clinical practice, the severity of RD is graded manually through visual inspection, which is labor intensive and often leads to large interrater variations. To overcome these shortcomings, this study aimed to develop an automatic RD assessment based on deep learning (DL) techniques that could efficiently assist the RD severity classification in clinical application. METHODS AND MATERIALS: A total of 1205 photographs of the head and neck region were collected from patients with nasopharyngeal carcinoma (NPC) undergoing radiation therapy. The severity of RD in these photographs was graded by 5 qualified assessors based on the Radiation Therapy Oncology Group guidance. An end-to-end RD grading framework was developed by combining a DL-based segmentation network and a DL-based RD severity classifier, which are used for segmenting the neck region from the camera-captured photographs and grading, respectively. U-Net was used for segmentation and another convolutional neural network classifier (DenseNet-121) was applied to RD severity classification. Dice similarity coefficient was used to evaluate the performance of segmentation. Severity classification was evaluated by several metrics, including overall accuracy, precision, recall, and F1 score. RESULTS: Results of segmentation showed that the averaged dice similarity coefficients were 91.2% and 90.8% for front and side view, respectively. For RD severity classification, the overall accuracy of test photographs was 83.0%. Our method accurately classified 90.5% of grade 0, 67.2% of grade 1, 93.8% of grade 2, and 100% of above grade 2 cases. The overall prediction performance was comparable with human assessors. There was no significant difference in accuracy when using manually or automatically segmented regions (P = .683). CONCLUSIONS: We have successfully demonstrated a DL-based method for automatic assessment of RD severity in patients with NPC. This method holds great potential for efficient and effective assessing and monitoring of RD in patients with NPC.

Effects of Performing Applied Muscle Tension during Recovery after Phlebotomy in Young, First-Time Donors: A Pilot Study.

Vasovagal reaction (VVR) compromises donor safety and reduces the subsequent return rates. Performing applied muscle tension (AMT) during phlebotomy may reduce the incidence of VVR. However, the effectiveness of performing AMT after phlebotomy to reduce delayed VVR remains unclear. With ethics approval, 12 young, first-time donors (YFTD) were recruited to study the effects on stroke volume (SV), cardiac output (CO) and systemic vascular resistance (SVR) while performing AMT from needle insertion to end of recovery. Measurements from 12 matched control YFTD were used for comparison. Pre-donation anxiety and VVR severity were assessed. Compared to controls, donors who performed AMT had higher SV (Control: 57 mL vs. AMT: 69 mL, p = 0.045), higher CO (Control: 3.7 L·min-1 vs. AMT: 5.2 L·min-1, p = 0.006) and lower SVR (Control: 1962 dyn·s·cm-5 vs. AMT: 1569 dyn·s·cm-5, p = 0.032) during mid-phlebotomy. During recovery, the AMT group retained higher SV, higher CO and lower SVR than the control, but not reaching statistical significance. Practicing AMT during recovery resulted in sustained haemodynamic improvements beyond the donation period, despite the reduction in delayed VVR was insignificant compared to the control group. A larger sample size is needed to validate the effectiveness of performing AMT after donation to mitigate delayed VVR.

A dosimetric comparison of the use of equally spaced beam (ESB), beam angle optimization (BAO), and volumetric modulated arc therapy (VMAT) in head and neck cancers treated by intensity modulated radiotherapy.

INTRODUCTION: Previous studies have shown that the beam arrangement had significant influence on plan quality in intensity modulated radiotherapy (IMRT). This study aimed to evaluate the dosimetric performance of beam arrangement methods by employing equally spaced beams (ESB), beam angle optimization (BAO), and volumetric modulated arc therapy (VMAT) in the planning of five types of head and neck (H&N) cancers treated by IMRT. METHODS: Five plans of different beam arrangement methods were optimized for 119 H&N cancer patients with the prescription of 66-70 Gy for high-risk planning target volume (PTV), 60 Gy for intermediate risk PTV, 54 Gy for low-risk PTV using a simultaneously integrated boost method. The five-beam arrangement methods were: ESB, coplanar BAO (BAOc), noncoplanar BAO (BAOnc), two-arc VMAT (VMAT2), and three-arc VMAT (VMAT3). The H&N cancers included cancers of nasopharynx, oral cavity, larynx, maxillary sinus, and parotid. Although the partial arc VMAT could be used in cases where the PTVs were situated at one side of the head such as the parotid, this arrangement was not included because it was intended to include only the beam arrangements that were applicable to all the types of head and neck cancers in the study. The plans were evaluated using a "figure-of-merit" known as uncomplicated target conformity index (UTCI). In addition, PTV conformation number and homogeneity index, normal tissue integral dose, and organ at risk (OAR) doses were also used. The mean values of these parameters were compared among the five plans. RESULTS: All treatment plans met the preset dose requirements for the target volumes and OARs. For nasopharyngeal cancer, VMAT3 and BAOnc demonstrated significantly higher UTCI. For cancer of oral cavity, most beam arrangement showed similar UTCI except ESB, which was relatively lower. For cancer of larynx, there was no significant difference in UTCI among the five-beam arrangement methods. For cancers of maxillary sinus and parotid gland, the two BAO methods showed marginally higher UTCI among all the five methods. CONCLUSION: Individual methods showed dosimetric advantages on certain aspects, and the UTCI of the BAO treatment plans are marginally greater in the case of maxillary sinus and parotid gland. However, if treatment time was included into consideration, VMAT plans would be recommended for cancers of the nasopharynx, oral cavity, and larynx.

The impact of positron emission tomography on primary tumour delineation and dosimetric outcome in intensity modulated radiotherapy of early T-stage nasopharyngeal carcinoma.

BACKGROUND: In intensity modulated radiotherapy (IMRT) of nasopharyngeal carcinoma (NPC), accurate delineation of the gross tumour volume (GTV) is important. Image registration of CT and MRI has been routinely used in treatment planning. With recent development of positron emission tomography (PET), the aims of this study were to evaluate the impact of PET on GTV delineation and dosimetric outcome in IMRT of early stage NPC patients. METHODS: Twenty NPC patients with T1 or T2 disease treated by IMRT were recruited. For each patient, 2 sets of NP GTVs were delineated separately, in which one set was performed using CT and MRI registration only (GTVCM), while the other set was carried out using PET, CT and MRI information (GTVCMP). A 9-field IMRT plan was computed based on the target volumes generated from CT and MRI (PTVCM). To assess the geometric difference between the GTVCM and GTVCMP, GTV volumes and DICE similarity coefficient (DSC), which measured the geometrical similarity between the two GTVs, were recorded. To evaluate the dosimetric impact, the Dmax, Dmin, Dmean and D95 of PTVs were obtained from their dose volume histograms generated by the treatment planning system. RESULTS: The overall mean volume of GTVCMP was greater than GTVCM by 4.4 %, in which GTVCMP was slightly greater in the T1 group but lower in the T2 group. The mean DSC of the whole group was 0.79 ± 0.05. Similar mean DSC values were also obtained from the T1 and T2 groups separately. The dosimetric parameters of PTVCM fulfilled the planning requirements. When applying this plan to the PTVCMP, the average Dmin (56.9 Gy) and D95 (68.6 Gy) of PTVCMP failed to meet the dose requirements and demonstrated significant differences from the PTVCM (p = 0.001 and 0.016 respectively), whereas the doses to GTVCMP did not show significant difference with the GTVCM. CONCLUSION: In IMRT of early stage NPC, PET was an important imaging modality in radiotherapy planning so as to avoid underdosing the PTV, although its effect on GTV delineation was not significant. It was recommended that PET images should be included in the treatment planning of NPC patients.

A comparison between electronic portal imaging device and cone beam CT in radiotherapy verification of nasopharyngeal carcinoma.

The demand of greater accuracy in intensity-modulated radiotherapy (IMRT) has driven the development of more advanced verification systems. The purpose of this study is to investigate the differences in verification accuracy in terms of the position error detected between cone-beam computed tomography (CBCT) and electronic portal imaging device (EPID) in the IMRT of nasopharyngeal carcinoma (NPC). Two groups of NPC patients (n = 22 and n = 28) verified by CBCT (G1-CB), EPID (G1-EP), and EPID (G2-EP) only, respectively, were recruited. The positional errors between the G1-CB group and the G2-EP group were compared. In addition, the magnitudes of the position errors of EPID taken in the same session of the CBCT, but after necessary corrections (G1-EP), were analyzed. In the CBCT group, 455 CBCT images (G1-CB) and 206 EPID images (G1-EP) were collected, whereas 319 EPID images (G2-EP) for the EPID group, were recorded. The median position errors detected in CBCT were between 0.80 and 0.90 mm in the antero-posterior (A-P), left-right (L-R), and supero-inferior (S-I) directions, whereas those of the EPID were all 0.50 mm. The magnitude of position deviation detected by the CBCT was higher than that of the EPID and their differences were extremely significant (p < 0.001). The frequencies in the G2-EP group with position errors greater than the tolerance (2 mm) were 32, 42, and 27 in the A-P, L-R, and S-I directions, respectively, which accounted for 16.5%, 21.6%, and 13.9% of the total number of EPID. There was difference in verification capability between the CBCT and EPID when applied to IMRT of NPC patients. Because an average of 1 of 6 verifications in EPID was inferior to that of the CBCT, verification by CBCT is recommended.