Abstract:BACKGROUND CT imaging is one of the most important contributors to medical radiation exposure(1). The frequency of CT scans and radiation doses accepted by patients attracted serious concerns for health physics researchers. The utilization of advanced technology ATCM has the potentials to reduce CT radiation doses while diagnostic image quality is maintained (2-7). As ATCM adjusted tube currents slice by slice it brought challenges to organ dose estimation using conversion factors derived from fixed tube current. Cross-system communication with hospital Picture Archive and Communication System (PACS),made it possible to read massive data automatically like the scanning parameters of each slice in each case. Monte Carlo simulations are probably the most reliable techniques which could be used for accurate dose assessment. [8-11]. However, specific patient model development and specific patient dose simulations are computationally demanding and may require dedicated hardware resources, this limitation constrained its application in large scale investigation. As an alternative method, patient specific organ doses could be calculated using the patient specific scan parameters and the Monte Carlo simulated organ doses with reference human phantom, and then correct the results with patient size factors. Dw is referred as the preferred patient size metric that determined the patient group and affected organ dose. The distance of the pathway traversed by the X-ray beam could provide the best approximation of tissue length traversed during the examination (12, 13),as CT image is a cross-sectional map normalized to the linear attenuation of water (14). The purpose of current study was to establish a method to access patient-specific organ dose associated with ATCM in chest computed tomography (CT) scans by combining Monte Carlo simulation with parameters contracted from clinical CT images of each patient underwent chest CT scan with ATCM. OBJECTIVE To explore a method to access patient-specific organ dose associated with automatic tube current modulation (ATCM) in chest computed tomography (CT) scans based on the information extracted from PACS automatically. METHODS 176cases of chest CT scans were read through cross-system communication with hospital PACS. A total of 8468 images were collected and analyzed automatically using in-house software. The scanning parameters (kVp, tube current, collimation width, etc.) of each CT examination were collected in real time, and a middle CT image of each case was collected for patient size(water equivalent diameter, Dw) calculation. Based on the reference human phantom, organ doses were simulated slice by slice using Monte Carlo method. The patient specific organ doses were calculated by combining tube currents of each patient slice with the simulated results, and doses were revised by correction factors that related to patient size. RESULTS A sum of 8468 slice of tube currents were extracted and analyzed in this study, the average mAs for large size patient group was about 1.6 times to the small size patient group. For organs that covered in the scan range like lung, breast, heart, the dose values were 18.30±2.91mGy, 15.13 ±2.75mGy and 17.87±2.96mGy in small size patients(Dw smaller than 22cm).The dose values of lung, breast, heart, in medium-sized patients (Dw from 22cm to 25cm) were 21.89±4.60mGy, 18.16 ±4.13mGy and 21.46±4.60mGy, while the values were 24.98±4.40mGy, 20.81±3.66mGy and 24.77±4.46mGy respectively in large size patients(Dw larger than 25cm). The organ doses increase with the patient size due to the increase of mAs. CONCLUSIONS The PACS-based method of large batch organ dose calculation to patients undergoing chest CT with ATCM was established. The methods and results may provide guidance to the design and optimization of chest CT protocols with ATCM.

Automated size‐specific dose estimates framework in thoracic CT using convolutional neural network based on U‐Net model

Automated size-specific dose estimates using deep learning image processing

Automated Clinical Target Volume Delineation Using Deep 3D Neural Networks in Radiation Therapy of Non-small Cell Lung Cancer

Estimation of Size-specific Dose Estimate (SSDE) of CT Scans using an Effective Diameter Electron Density

Development of a novel automated algorithm for patient dosimetry in computed tomography: a step towards the facilitation of size-specific dose estimates and organs dosimetry estimations in a busy clinical workflow

Automated segmentation of an intensity calibration phantom in clinical CT images using a convolutional neural network

A Method of Rapid Quantification of Patient-Specific Organ Doses for CT Using Deep-Learning-based Multi-Organ Segmentation and GPU-accelerated Monte Carlo Dose Computing.

Automatic Calcium Scoring in Low-Dose Chest CT Using Deep Neural Networks With Dilated Convolutions

Deep Learning Renal Segmentation for Fully Automated Radiation Dose Estimation in Unsealed Source Therapy

A Method of Rapid Quantification of Patient-Specific Organ Dose for CT Using Coupled Deep-Learning based Multi-Organ Segmentation and GPU-accelerated Monte Carlo Dose Computing

Dose-volume-based evaluation of convolutional neural network-based auto-segmentation of thoracic organs at risk

Fast dose calculation in x-ray guided interventions by using deep learning

Deep learning architecture for scatter estimation in cone-beam computed tomography head imaging with varying field-of-measurement settings

Introduction and evaluation of size-specific DLP for radiation dose estimation in CT examinations

Deep Learning-Based Low Dose Ct Imaging

Deep Learning-based calculation of patient size and attenuation surrogates from localizer Image: Toward personalized chest CT protocol optimization

Unsupervised learning for deformable registration of thoracic CT and cone‐beam CT based on multiscale features matching with spatially adaptive weighting

Automatic Segmentation of Thoracic Aorta Segments in Low-Dose Chest CT

PACS-based Calculation of Organ Dose to Patients undergoing Chest Computed Tomography with Automatic Tube Current Modulation (Preprint)

Automatic airway segmentation from Computed Tomography using robust and efficient 3-D convolutional neural networks

Cone Beam CT (CBCT) Based Synthetic CT Generation Using Deep Learning Methods for Dose Calculation of Nasopharyngeal Carcinoma Radiotherapy