Early experience with four-dimensional computed tomography for observing velopharyngeal closure function in pediatric patients and the evaluation of the exposure dose

a of with clinical using 320-ADCT evaluate the in We the organ-absorbed Our evaluation for refuse to a conventional evaluation. We further will apply these to the quality of and management of


Abstract Background
The aims of this study were to perform a four-dimensional assessment of velopharyngeal closure function in pediatric patients with cleft palate using 320-row area detector computed tomography (CT), and to estimate the organ-absorbed doses using Monte Carlo simulation.

Methods
We evaluated CT image data obtained between July 2018 and August 2019 from ve pediatric patients with cleft palate (four boys and one girl; age range, 4-7 years) at Fujita Health University Hospital. The presence of velopharyngeal insu ciency (VPI), patterns of velopharyngeal closure (VPC), and cross-sectional area of VPI were evaluated. In addition, organ-absorbed doses were assumed in the Monte Carlo simulation. However, we did not perform statistical analysis because of the insu cient number of patients enrolled in this study.

Results
The existence of VPI and hypernasality were completely concordant. The VPC patterns were circular (two patients), circular with Passavant's ridge (one patient), and unevaluable (two patients). The organ-absorbed doses were relatively lower than those in past reports.

Conclusions
Our method could be an alternative for patients who refuse the conventional nasopharyngoscopic evaluation.

Background
Anatomical abnormalities of the nasopharyngeal space that are associated with cleft palate result in velopharyngeal insu ciency (VPI), which may cause dysarthria and dysphagia [1]. If VPI remains despite attempts to improve function with surgery and speech pathology evaluation, a secondary surgery is indicated to achieve function. Various examinations are used to assess velopharyngeal function (VPF) to determine the surgical or nonsurgical approach to improve VPF. The gold standard for assessing VPF is a speech pathology evaluation by a speech-language pathologists (SLPs). However, to determine the application of a secondary surgery and the choice of the procedure, an examination of anatomical abnormalities is necessary. Two main areas of dysfunction are (1) between the soft palate and posterior pharyngeal wall and (2) the lateral walls of the pharynx. Nasopharyngoscopy is the only procedure that allows the observation of both areas; however, it requires the insertion of an endoscope, which may result in much discomfort that often makes an examination di cult to perform.
In 2019, we reported a four-dimensional (4D) evaluation method that appends the time phase image to threedimensional (3D) images for the assessment of velopharyngeal closure (VPC) by using 320-row area detector computed tomography (320-ADCT) images acquired from adult volunteers; this method was named "virtual nasopharyngoscopy" [2]. When using this method, the time and dose of exposure were reduced, compared to those of a previously reported method [3]. In this study, we report the application of our method to pediatric patients and estimate the organ-absorbed doses in these patients by using Monte Carlo (MC) simulation.

Patients
We evaluated the computed tomography (CT) image data of ve pediatric patients (four boys and one girl; age range, 4-7 years); the CT images were obtained between July 2018 and August 2019. The participants consisted of three preoperative patients with submucosal cleft palate and two postoperative patients with unilateral cleft lip and palate. All participants had received dysarthria training for more than 6 months; however, they all presented with persistent hypernasality in speech pathology evaluations by SLPs. In addition, nasopharyngoscopic evaluations could not be performed in these patients (except for two patients) because of persistent crying. The study was conducted in accordance with the Declaration of Helsinki (1964) and was approved by the Fujita Health University Ethics Review Committee (Aichi, Japan; reference number: HM18-431). All parents or legal guardians of the participants were su ciently informed about the purpose of this study, and they provided written informed consent for their child's personal or clinical details to be published in this study.
Imaging procedure CT imaging using a 320-ADCT scanner (Aquilion ONE/Genesis Edition; Toshiba Medical Systems Corp., Tochigi, Japan) was conducted, using the method described in our previous report [2]. Each participant lay horizontally on the bed, and performed two successive tasks following the signal of the examiner: nasal inspiration, followed by oral expiration through a catheter into a water-lled cup. Scanning began immediately after the onset of inspiration. Expiration continued 1 second after starting and until the end of the scanning. We used a 14-French gauge catheter (4.7 mm outer diameter, 47 cm long; Nurvie; Covidien Japan, Tokyo, Japan) and water-lled disposable clear polyethylene terephthalate cups (370 mL) (Fig. 1).
The scanning parameters were set as follows: eld of view (FOV), 240 mm; scanning length, 80-120 mm; tube voltage, 120 kV; tube current, 30 mA; and exposure time, 3.30 s (0.275 s/rotation × 12 rotations). For each child, the scanning length was set between the inferior border of the mandible and the infraorbital border to reduce exposure to the thyroid gland and eyeball. Two patients performed the tasks twice because they failed to execute the protocol. To improve temporal resolution, we implemented a half-reconstruction technique by which images were generated from 0.138-second data for image generation. The medians of the volume of the CT dose index (CTDI vol ) and the dose length product (DLP) for each scan were 1.02 mGy (12.20 mGy/12 rotations) and 119.9 mGy × cm, respectively.

Image analysis
The images were drawn with 0.5-mm slice thickness and transferred to a medical imaging workstation (Ziostation2; Ziosoft, Inc., Tokyo, Japan), where 4D images of airway mobility (i.e., airway-mode), 4D images in virtual endoscopy mode (Additional le 1, Video 1), and multiplanar reconstruction (MPR) images were generated (Fig. 2). The airway-mode, virtual endoscopy mode, and MPR images had a window length of 500 HU, 75 HU, and 40 HU, respectively, and a window width of 500 HU, 500 HU, and 400 HU, respectively. The movement of the velopharyngeal structures was recorded during the examination, and the presence of VPI and VPC patterns were estimated. The VPC patterns were categorized into four groups: coronal, sagittal, circular, and circular with Passavant's ridge, based on earlier reports [4] (Fig. 3). The minimum crosssectional area (CSA) of the nasopharynx was measured parallel to the palatal plane, using the MPR images obtained during blowing.

Exposure dose calculation
Based on the protocol described in a previous report [5], we estimated organ-absorbed doses by using MC simulation with ImpactMC (CT imaging GmbH, Erlangen, Germany). With this software, the air kerma, bowtie lter shape data, X-ray spectrum data, CT-to-density conversion coe cients, and scanning conditions of Digital Imaging and Communications in Medicine (DICOM) images can be input to calculate the absorbed doses for individual voxels while taking into account the photon interaction process. Air kerma for 4D-CT imaging was measured with an ionization chamber dosimeter (Model 9015; Radcal, Monrovia, CA, USA) by placing a pencil-type ionization chamber dosimeter (10X-3CT; Radcal, Monrovia, CA, USA) at the isocenter of the X-ray CT apparatus. The shape of the bow-tie lter in the X-ray CT apparatus was evaluated, based on Xray attenuation.
In the present study, the bow-tie lter shape was determined by measuring air kerma by moving the penciltype ionization chamber along the X-axis for 180 mm in 10-mm intervals (Fig. 4). Directly measuring the primary energy spectrum of the X-ray CT apparatus is di cult because of the structure of the system. Therefore, a carbon rod was placed at the isocenter, and the energy spectrum of scattered radiation scattered at 90° was measured using a spectrum analyzer (RAMTEC413; Toyo Medic Co., Ltd., Tokyo, Japan) and a detector (XR-100T-CdTe; Amptek, Bedford, MA, USA) (Fig. 5). The energy spectrum of the scattered radiation was converted to the primary ray energy spectrum, based on the Klein-Nishina formula. A software standard conversion table was used, based on the CT-to-density conversion coe cients of water (0 HU: 1 g/cm 2 ) and air (-1000 HU: 0 g/cm 2 ) (Fig. 6). MC simulation was conducted by inputting the scanning conditions of the DICOM images from 4D-CT imaging and other required items (Table 1) to obtain the absorbed dose distributions (Fig. 7). For each patient's absorbed dose distribution, 5-pixel square regions of interest (ROIs) were randomly selected at ve locations in each target organ (i.e., brain, tongue, vocal cords, esophagus, thyroid, nasopharynx, oropharynx, hypopharynx, crystalline, and salivary gland) to evaluate each of the organ-absorbed doses. We were unable to perform statistical analysis because of the insu cient number of patients enrolled in this study.

Image analysis
The CT images revealed VPI in all patients (Table 2). In the patients' assessment, VPI and hypernasality were completely concordant. The VPC patterns were evaluated as circular (two patients) and circular with Passavant's ridge (one patient); the other patterns (two patients) were unevaluable because of poor mobility of the structures in the velopharyngeal spaces. While blowing, some participants sealed the nostril with the upper lip or blew air that was stored in the oral vestibule (Fig. 8). The CSA of VPI ranged 3.42-271 mm 2 . The absorbed doses of each organ are shown in Table 3. The scanning length was set between the inferior border of the mandible and the infraorbital border; however, radiation exposure to the thyroid gland and eyeball occurred in some patients because of body movement during imaging. Each absorbed dose was nevertheless relatively lower than the dose reported in past reports in which swallowing was evaluated with CT examinations in adult patients [5,6].

Discussion
In this study, we evaluated the reliability of "virtual nasopharyngoscopy," for assessing VPF in pediatric patients with cleft palate. We compared the ndings of 320-ADCT versus those of the speech pathology evaluation. In general, the gold standard for estimating VPF is the detection of hypernasality in a speech pathology evaluation by a SLP. However, an imaging examination is essential for choosing the appropriate plan for treating VPI. Fiberoptic nasopharyngoscopy, which can anatomically identify the region and severity of VPI with 3D images, has an important role; however, it may cause pain, distress, or the vomiting re ex.
Thus, pediatric patients often refuse to undergo this examination or fail to show su cient ability to undergo an examination. Hence, an incomplete treatment plan or an inadequate surgical procedure for VPI may result.
In their 2015 study, Sakamoto et al. [3] reported the rst 4D assessment of VPF in a patient with cleft palate by using 320-ADCT. They evaluated VPF in ve pediatric patients during an exposure time of approximately 10 seconds and concluded that 4D-CT produces clear and detailed images with less stress and pain; it also had a shorter examination duration than that of other imaging modalities, and allows a quantitative VPC evaluation [3]. However, they reported that a major drawback of their procedure was an average radiation exposure of 4.44 mSv, which is larger than the level required in cephalometry, video uoroscopy, and conventional CT [3].
In 2019, we reported a modi ed scan protocol that shortened the exposure time during VPF assessment [2].
In that study, we evaluated 10 adult volunteers, which included ve patients with postoperative cleft. We concluded that nasopharyngoscopy and 4D evaluation with 320-ADCT had a high concordance rate for evaluating VPI (8/10 patients) and VPC patterns (9/10 patients) [2].
In the current study, we evaluated ve pediatric patients with hypernasality. The presence of hypernasality and VPI in the CT images were completely concordant. The VPC patterns were evaluated as circular and circular with Passavant's ridge; however, two patients had unevaluable images. This nding was similar to the assessment of adult postoperative cleft patients reported in our past study: four patients had the circular pattern and one patient had the circular with Passavant's ridge pattern [2]. However, some participants in the current study demonstrated compensatory behavior during blowing to overcome the malfunction. This behavior was not observed in our past study, which assessed patients without VPI. Therefore, preparing an alternative task (e.g., sustained phonation) may be important when evaluating patients with VPI.
The CSA among patients with VPI ranged widely and was not associated with cleft type or whether a primary or secondary operation had been scheduled. We reported that the CSA among postoperative adult patients with VPI was at most 16.28 mm 2 and they did not present with hypernasality. However, Patients #2 and #3 in this study had a smaller CSA and presented with hypernasality. The difference in CSA size among the same symptomatic patients remains unclear; however, future research will hopefully clarify this point.
In radiographic imaging of pediatric patients, the most concerning problem is exposure dose. Several investigators have estimated VPI or VPF by using dynamic magnetic resonance imaging [7,8].This is a noninvasive technique that can be repeated without the risk of radiation exposure; however, compared to CT, magnetic resonance imaging's low imaging speed, narrow spatial coverage, and noise may interfere with the examination of pediatric patients.
We calculated the organ-absorbed dose with MC simulation to assess the radiation exposure in this examination as another goal in this study. Three types of dose estimation methods primarily exist: the use of an anthropomorphic phantom, a simulated reference man [9], or actual clinical images with MC simulation.
Each method has advantages and disadvantages. The strongest advantage of MC simulation is the arbitrariness in setting the ROI size and place. We could then evaluate the details of the organ-absorbed dose by using MC simulation. However, this simulation cannot be used to assess the effective dose because of the lack of exposure dosage data outside of the coverage area.
The International Commission on Radiological Protection (ICRP) and many other international organizations have cooperated to optimize medical exposure. The ICRP recommended using the diagnostic reference level (DRL) to propel optimizing protection in the radiological diagnosis in its ICRP Publication 73 [10]. As mentioned in ICRP Publication 135 [11], DRLs are individually established in each country or region because equipment and procedure protocols can vary between different facilities. In general, the CTDI vol or the DLP is described as the DRL. The CTDI vol is used to evaluate the performance of CT apparatuses. In 2015, the Japan Network for Research and Information on Medical Exposure published the report Diagnostic Reference Levels Based on Latest Surveys in Japan [12]. In the report, DRLs for pediatric head CT in children 1-5 years and 6-10 years were set at 47 mGy CTDI vol and 60 mGy CTDI vol , respectively. In our study, the CTDI vol of 320-ADCT was 12.20 mGy/12 rotations. The coverage area does not correspond to the head completely; however, we believe that this amount of exposure is su ciently optimized.
This study has some limitations. First, the short-duration protocol limits the number of feasible tasks.
Therefore, the obtained results may not su ciently re ect the actual VPI or hypernasality. In particular, the difference in the size of VPI or the in uence of gravity is not su ciently explained. In some patients, alternative procedures such as sustained pronunciation should be considered rather than blowing.
Second, the exposure dose remained much higher than that of conventional radiographic inspections. We hope future technological innovations will enhance the sensitivity of detectors and reduce the exposure dose. However, we believe that this amount of radiation exposure is acceptable with proper use because much information can be acquired with this method.
The greatest advantages of CT evaluation, compared to nonirradiating methods, are reproducibility, standardization, and the ability to allow for quantitative analysis. In this study, we did not prove the reproducibility of this method in the single examination of each participant. However, in future research, we will accumulate a number of cases and improve the reliability of our method.

Conclusions
We reported the clinical application of 4D-CT using 320-ADCT to evaluate the VPC function in pediatric patients with cleft palate. We also evaluated the organ-absorbed doses. Our method can be an alternative evaluation method for patients who refuse to undergo a conventional nasopharyngoscopic evaluation. We believe that further research will apply these merits to improve the quality of treatment and management of cleft patients.
Abbreviations 320-ADCT, 320-row area detector computed tomography; 3D, three-dimensional; 4D, four-dimensional; CSA, The study was conducted in accordance with the Declaration of Helsinki (1964) and was approved by the Fujita Health University Ethics Review Committee (Aichi, Japan; reference number: HM18-431). All guardians of the participants were su ciently informed about the purpose of this study, and they provided written informed consent.

Consent for publication
All parents or legal guardians of the participants gave written informed consent for their child's personal or clinical details, along with any identifying images, to be published in this study.

Availability of data and materials
The datasets supporting the conclusions of this article are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.  The test protocol. Each participant performed two tasks: nasal inspiration, followed by forceful expiration through a catheter into a water-lled cup. The parents of the participant gave written consent for the identifying images to be published in this study. Schematics of velopharyngeal closure. The patterns of velopharyngeal closure are categorized into four groups, as described in previous literature [4]. (This gure is cited in our past report [2].) Page 15/19 The schema indicates the measurement of the energy spectrum in the computed tomography (CT) scanner.

Figure 6
Relationship between density and the computed tomography (CT) value in the Monte Carlo simulation.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.