A fully automated machine learning-based methodology for personalized radiation dose assessment in thoracic and abdomen CT.

PURPOSE: To develop a machine learning-based methodology for patient-specific radiation dosimetry in thoracic and abdomen CT. METHODS: Three hundred and thirty-one thoracoabdominal radiotherapy-planning CT examinations with the respective organ/patient contours were collected retrospectively for the development and validation of segmentation 3D-UNets. Moreover, 97 diagnostic thoracic and 89 diagnostic abdomen CT examinations were collected retrospectively. For each of the diagnostic CT examinations, personalized MC dosimetry was performed. The data derived from MC simulations along with the respective CT data were used for the training and validation of a dose prediction deep neural network (DNN). An algorithm was developed to utilize the trained models and perform patient-specific organ dose estimates for thoracic and abdomen CT examinations. The doses estimated with the DNN were compared with the respective doses derived from MC simulations. A paired t-test was conducted between the DNN and MC results. Furthermore, the time efficiency of the proposed methodology was assessed. RESULTS: The mean percentage differences (range) between DNN and MC dose estimates for the lungs, liver, spleen, stomach, and kidneys were 7.2 % (0.2-24.1 %), 5.5 % (0.4-23.0 %), 7.9 % (0.6-22.3 %), 6.9 % (0.0-23.0 %) and 6.7 % (0.3-22.6 %) respectively. The differences between DNN and MC dose estimates were not significant (p-value = 0.12). Moreover, the mean processing time of the proposed workflow was 99 % lower than the respective time needed for MC-based dosimetry. CONCLUSIONS: The proposed methodology can be used for rapid and accurate patient-specific dosimetry in chest and abdomen CT.

A neural network-enhanced methodology for the rapid establishment of local DRLs in interventional radiology: EVAR as a case example.

PURPOSE: To develop a neural network-enhanced workflow for the automatic and rapid establishment/update of local diagnostic reference levels (DRLs) in interventional radiology (IR) using endovascular aneurysm repair (EVAR) procedures as a case example. METHODS: Radiation dose reports were collected retrospectively for 46 consecutive EVAR procedures. These reports served as demonstrative data for the development of the proposed methodology. An algorithm was developed to receive multiple dose reports, automatically extract the kerma area product (KAP), air kerma (Ka,r), number of exposure images, and fluoroscopy time (FT) from each report and calculate the first, second, third quartiles as well as the maximum and minimum values of the extracted parameters. To extract the values of interest from the dose reports, Tesseract, an open-source optical character recognition (OCR) engine was employed. Furthermore, the accuracy and time efficiency of the proposed methodology were assessed. Specifically, the values extracted from the algorithm were compared with the ground truth values and the algorithm's processing time was compared with the respective time needed to manually extract and process the values of interest. RESULTS: The OCR-based algorithm managed to correctly recognize 182 from the 184 target values, resulting in an accuracy of 99%. Moreover, the proposed pipeline reduced the processing time for the establishment of DRLs by 98%. DRL value for EVAR procedures, set as the third quartile of KAP was found to be 551 Gy*cm2. CONCLUSION: An accurate and time-efficient workflow was developed for the establishment of local DRLs in interventional radiology.

Automatic Radiobiological Comparison of Radiation Therapy Plans: An Application to Gastric Cancer.

(1) Aim: This study was conducted to radiobiologically compare radiotherapy plans for gastric cancer with a newly developed software tool. (2) Methods: Treatment planning was performed on two computational phantoms simulating adult male and female patients. Three-dimensional conformal radiotherapy (3D-CRT), intensity-modulated radiation therapy (IMRT) and volumetric modulated arc therapy (VMAT) plans for gastric cancer were generated with three-photon beam energies. The equivalent uniform dose (EUD), tumor control probability (TCP) of the target and normal tissue control probability (NTCP) of eight different critical organs were calculated. A new software was employed for these calculations using the EUD-based model and dose-volume-histogram data. (3) Results: The IMRT and VMAT plan led to TCPs of 51.3-51.5%, whereas 3D-CRT gave values up to 50.2%. The intensity-modulated techniques resulted in NTCPs of (5.3 × 10-6-3.3 × 10-1)%. The corresponding NTCPs from 3D-CRT were (3.4 × 10-7-7.4 × 10-1)%. The above biological indices were automatically calculated in less than 40 s with the software. (4) Conclusions: The direct and quick radiobiological evaluation of radiotherapy plans is feasible using the new software tool. The IMRT and VMAT reduced the probability of the appearance of late effects in most of the surrounding critical organs and slightly increased the TCP compared to 3D-CRT.

A novel methodology to train and deploy a machine learning model for personalized dose assessment in head CT.

OBJECTIVES: To propose a machine learning-based methodology for the creation of radiation dose maps and the prediction of patient-specific organ/tissue doses associated with head CT examinations. METHODS: CT data were collected retrospectively for 343 patients who underwent standard head CT examinations. Patient-specific Monte Carlo (MC) simulations were performed to determine the radiation dose distribution to patients' organs/tissues. The collected CT images and the MC-produced dose maps were processed and used for the training of the deep neural network (DNN) model. For the training and validation processes, data from 231 and 112 head CT examinations, respectively, were used. Furthermore, a software tool was developed to produce dose maps from head CT images using the trained DNN model and to automatically calculate the dose to the brain and cranial bones. RESULTS: The mean (range) percentage differences between the doses predicted from the DNN model and those provided by MC simulations for the brain, eye lenses, and cranial bones were 4.5% (0-17.7%), 5.7% (0.2-19.0%), and 5.2% (0.1-18.9%), respectively. The graphical user interface of the software offers a user-friendly way for radiation dose/risk assessment. The implementation of the DNN allowed for a 97% reduction in the computational time needed for the dose estimations. CONCLUSIONS: A novel methodology that allows users to develop a DNN model for patient-specific CT dose prediction was developed and implemented. The approach demonstrated herein allows accurate and fast radiation dose estimation for the brain, eye lenses, and cranial bones of patients who undergo head CT examinations and can be used in everyday clinical practice. KEY POINTS: • The methodology presented herein allows fast and accurate radiation dose estimation for the brain, eye lenses, and cranial bones of patients who undergo head CT examinations and can be implemented in everyday clinical practice. • The scripts developed in the current study will allow users to train models for the acquisition protocols of their CT scanners, generate dose maps, estimate the doses to the brain and cranial bones, and estimate the lifetime attributable risk of radiation-induced brain cancer.

A software tool for organ-specific second cancer risk assessment from exposure to therapeutic doses.

The aim of this study was the development of a software tool (SCRcalc) for the automatic estimation of the patient- and organ-specific cancer risk due to radiotherapy. SCRcalc was developed using the Python 3.8.7 programming language. It incorporates equations and parameters of mechanistic models for the calculation of the organ equivalent dose (OED), the excess absolute risk (EA R) and the lifetime attributable risk (LA R) of carcinogenesis for various organs due to radiotherapy. Data from differential dose-volume histograms, as defined by a treatment planning system, could be automatically inserted into the program. Eighteen different cancer risk estimates for various organs were performed of patients subjected to radiation therapy with conventional and modulated techniques. These software estimates were compared with manual calculations. SCRcalc was developed as a standalone executable program without any dependencies. It enables direct estimations of the OED and LAR for various organs at risk. An important aspect of the software is that it does not require pre-processing of the DVH data. No differences were found between the SCRcalc results and those derived from manual calculations. The newly developed software offers the possibility to medical physicists and radiation oncologists to directly estimate the probability of radiotherapy-induced secondary malignancies for various organs at risk.

Radiation exposure to infants undergoing voiding cystourethrography: The importance of the digital imaging technology.

PURPOSE: To determine the radiation burden to infants undergoing voiding cystourethrography (VCUG) in a single institution and investigate the effect of shifting from analogue to digital imaging that allowed the use of a radiography-free examination protocol. METHODS: Anthropometric and exposure data were prospectively collected for 35 consecutive infants undergoing VCUG on a digital system with a standardized examination protocol not including radiographs. Thermoluminescent dosimeters were used to determine entrance-skin dose. Monte Carlo simulations and patient-specific anthropomorphic phantoms were employed to determine organ/tissue doses and effective dose (ED). The associated theoretical risk of radiation-induced cancer was determined and compared to the nominal risk of cancer induction. The radiation burden from VCUG on a modern digital system with a contemporary examination protocol was compared to corresponding data reported previously for an analogue system in the same institution. RESULTS: The median ED from VCUG was found 47 µSv. The associated total life attributable risk of radiation-induced cancer was found 10x10-6 and 13x10-6 for boys and girls, respectively. VCUG was found to increase the nominal risk of cancer by a factor of 1.000025 in boys and 1.000034 in girls. Shifting from analogue to digital imaging system resulted in 89% reduction of the radiation burden from VCUG. CONCLUSION: The theoretical radiation risks for infants undergoing VCUG using a modern digital imaging system and a radiography-free protocol were found to be minor. The transition from analogue to digital equipment resulted in considerable reduction of the radiation burden from VCUG.

A novel personalized dosimetry method for endovascular aneurysm repair (EVAR) procedures.

OBJECTIVE: To estimate radiation doses for the primarily irradiated organs/tissues of patients subjected to standard endovascular aneurysm repair (EVAR) procedures using a novel personalized dosimetry method. METHODS: Dosimetric and anthropometric data were collected prospectively for eight patients who underwent standard EVAR procedures. Patient-specific Monte Carlo simulations were performed to estimate organ/tissue doses from each of the fluoroscopic and digital subtraction angiography acquisitions involved in EVAR. Individual-specific cumulative absorbed doses were estimated for the skin, spinal bone marrow, heart, kidneys, liver, colon, bladder, pancreas, stomach, and spleen and compared to corresponding values estimated through a commercially available dosimetric software package that employs standardized phantoms. RESULTS: The highest organ/tissue radiation doses from EVAR were found for the skin, spinal bone marrow, kidneys, and spleen as 192.4 mGy, 96.7 mGy, 72.9 mGy, and 33.6 mGy respectively, while the doses to the heart, liver, colon, bladder, pancreas, and stomach were 6.3 mGy, 14.4 mGy, 18.4 mGy, 14.8 mGy, 21.6 mGy, and 11.2 mGy respectively. Corresponding dose values using standardized phantoms were found to differ up to 151%. CONCLUSION: Considerable radiation doses may be received by primarily exposed organs/tissues during standard EVAR. The specific size/anatomy of the patient and the variation in exposure parameters/beam angulation between different projections commonly involved in EVAR procedures should be taken into account if reliable organ dose data are to be derived. KEY POINTS: • A novel patient-specific dosimetry method was utilized to estimate radiation doses to the primarily irradiated organs/tissues of patients subjected to standard endovascular aneurysm repair procedures. • The use of standardized mathematical anthropomorphic phantoms to derive organ dose from fluoroscopically guided procedures may result in considerable inaccuracies due to differences in the assumed organ position/volume/shape compared to patients.

Complexity-based local diagnostic reference levels (DRLs) for standard endovascular aneurysm repair (EVAR) procedures.

PURPOSE: The aim of this study was to establish complexity-based local diagnostic reference levels (DRLs) for standard endovascular aneurysm repair (EVAR) procedures. METHODS: Dosimetric data for 73 consecutive patients were collected prospectively. All procedures were performed on a Siemens Axiom Artis FA angiographic unit (Siemens, Erlangen, Germany). Fluoroscopy time (FT), dose area product (DAP), air kerma (Ka.r) at reference point as well as patient's age, height and weight were recorded for each procedure. Moreover, the complexity level of each procedure was evaluated. Spearman rank correlation tests were used to evaluate the degree of association between variables. RESULTS: 2nd quartiles of DAP for low, medium and high complexity standard EVAR procedures were 144.2 Gycm2, 160.1 Gycm2 and 189.5 Gycm2respectively. The median DAP of the full sample was 153.2 Gycm2. Statistical analysis showed a strong correlation between BMI and DAP (r = 0.68, p-value < 0.0001) and a moderate correlation between BMI and Ka.r (r = 0.52, p-value < 0.0001). Furthermore, a strong correlation was found between Ka.r and FT (r = 0.76, p-value < 0.0001) and a moderate correlation was found between DAP and FT (r = 0.57, p-value < 0.0001). CONCLUSION: The complexity of an EVAR procedure is associated with the total burden of radiation. Establishment of complexity-based DRLs for interventional radiology procedures can contribute to the radiation protection of patients and staff.

Occupational exposure during endovascular aneurysm repair (EVAR) and aortoiliac percutaneous transluminal angioplasty (PTA) procedures.

OBJECTIVES: The purpose of this study was to determine the radiation exposure of primary interventionalist's different body parts during endovascular aneurysm repair (EVAR) procedures and aortoiliac percutaneous transluminal angioplasty (PTA) procedures and to evaluate the efficacy of a radioprotective drape. METHODS: Occupational doses for 36 consecutive aortoiliac PTA procedures and 17 consecutive EVAR procedures were estimated using thermoluminescence dosimetry (TLD) chips (TLD-200, Hashaw, Solon, OH). Effective dose (ED) was calculated using the Niklason algorithm. For the evaluation of a 0.25 mm Pb equivalent drape (Ecolab, Saint Paul, Minnesota, USA), experiments were performed using two physical anthropomorphic phantoms (Rando-Alderson Research Labs, CA, USA). RESULTS: Median ED for a typical EVAR and PTA procedure was 4.7 ± 1.4 µSv and 4.4 ± 3.6 µSv, respectively. The highest radiation doses were measured for the operator's hands in both procedures. Moreover, considerable doses were measured to the operator's head, eye lenses and thyroid. Due to the use of the drape, radiation exposure of primary operator's abdominal area, genitals, thyroid and eye lenses was reduced by an average of 59%, 60%, 65% and 59%, respectively. However, dose area product (DAP) and peak skin dose (PSD) were increased by 20% when part of the drape was placed into the X-ray field. CONCLUSION: During EVAR and PTA procedures, primary operator's organs are exposed to considerable radiation doses. Occupational radiation exposure can be reduced significantly with the proper use of a radioprotective drape.