1
|
Vellini L, Quaranta F, Menna S, Pilloni E, Catucci F, Lenkowicz J, Votta C, Aquilano M, D’Aviero A, Iezzi M, Preziosi F, Re A, Boschetti A, Piccari D, Piras A, Di Dio C, Bombini A, Mattiucci GC, Cusumano D. A deep learning algorithm to generate synthetic computed tomography images for brain treatments from 0.35 T magnetic resonance imaging. Phys Imaging Radiat Oncol 2025; 33:100708. [PMID: 39958708 PMCID: PMC11830347 DOI: 10.1016/j.phro.2025.100708] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/30/2024] [Revised: 01/16/2025] [Accepted: 01/21/2025] [Indexed: 02/18/2025] Open
Abstract
Background and Purpose The development of Magnetic Resonance Imaging (MRI)-only Radiotherapy (RT) represents a significant advancement in the field. This study introduces a Deep Learning (DL) algorithm designed to quickly generate synthetic CT (sCT) images from low-field MR images in the brain, an area not yet explored. Methods Fifty-six patients were divided into training (32), validation (8), and test (16) groups. A conditional Generative Adversarial Network (cGAN) was trained on pre-processed axial paired images. sCTs were validated using mean absolute error (MAE) and mean error (ME) calculated within the patient body. Intensity Modulated Radiation Therapy (IMRT) plans were optimised on simulation MRI and calculated considering sCT and original CT as electron density (ED) map. Dose distributions using sCT and CT were compared using global gamma analysis at different tolerance criteria (2 %/2mm and 3 %/3mm) and evaluating the difference in estimating different Dose Volume Histogram (DVH) parameters for target and organs at risk (OARs). Results The network generated sCTs of each single patient in less than two minutes (mean time = 103 ± 41 s). For test patients, the MAE was 62.1 ± 17.7 HU, and the ME was -7.3 ± 13.4 HU. Dose parameters on sCTs were within 0.5 Gy of those on original CTs. Gamma passing rates 2 %/2mm, and 3 %/3mm criteria were 99.5 %±0.5 %, and 99.7 %±0.3 %, respectively. Conclusion The proposed DL algorithm generates in less than 2 min accurate sCT images in the brain for online adaptive radiotherapy, potentially eliminating the need for CT simulation in MR-only workflows for brain treatments.
Collapse
Affiliation(s)
| | | | | | | | | | - Jacopo Lenkowicz
- Fondazione Policlinico Gemelli Agostino Gemelli IRCCS Roma Italy
| | - Claudio Votta
- Fondazione Policlinico Gemelli Agostino Gemelli IRCCS Roma Italy
| | | | - Andrea D’Aviero
- Department of Medical, Oral and Biotechnological Sciences, “Gabriele D’Annunzio” Università di Chieti, Italy
- Department of Radiation Oncology, “S.S. Annunziata”, Chieti Hospital, Italy
| | | | | | - Alessia Re
- Mater Olbia Hospital Olbia Sassari Italy
| | | | | | - Antonio Piras
- UO Radioterapia Oncologica, Villa Santa Teresa Bagheria Palermo Italy
| | | | - Alessandro Bombini
- Istituto Nazionale di Fisica Nucleare (INFN) Sesto Fiorentino FI Italy
- ICSC - Centro Nazionale di Ricerca in High Performance Computing, Big Data & Quantum Computing Casalecchio di Reno BO Italy
| | - Gian Carlo Mattiucci
- Mater Olbia Hospital Olbia Sassari Italy
- Università Cattolica del Sacro Cuore Rome Italy
| | | |
Collapse
|
2
|
Zaffino P, Raggio CB, Thummerer A, Marmitt GG, Langendijk JA, Procopio A, Cosentino C, Seco J, Knopf AC, Both S, Spadea MF. Toward Closing the Loop in Image-to-Image Conversion in Radiotherapy: A Quality Control Tool to Predict Synthetic Computed Tomography Hounsfield Unit Accuracy. J Imaging 2024; 10:316. [PMID: 39728213 DOI: 10.3390/jimaging10120316] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/13/2024] [Revised: 12/03/2024] [Accepted: 12/07/2024] [Indexed: 12/28/2024] Open
Abstract
In recent years, synthetic Computed Tomography (CT) images generated from Magnetic Resonance (MR) or Cone Beam Computed Tomography (CBCT) acquisitions have been shown to be comparable to real CT images in terms of dose computation for radiotherapy simulation. However, until now, there has been no independent strategy to assess the quality of each synthetic image in the absence of ground truth. In this work, we propose a Deep Learning (DL)-based framework to predict the accuracy of synthetic CT in terms of Mean Absolute Error (MAE) without the need for a ground truth (GT). The proposed algorithm generates a volumetric map as an output, informing clinicians of the predicted MAE slice-by-slice. A cascading multi-model architecture was used to deal with the complexity of the MAE prediction task. The workflow was trained and tested on two cohorts of head and neck cancer patients with different imaging modalities: 27 MR scans and 33 CBCT. The algorithm evaluation revealed an accurate HU prediction (a median absolute prediction deviation equal to 4 HU for CBCT-based synthetic CTs and 6 HU for MR-based synthetic CTs), with discrepancies that do not affect the clinical decisions made on the basis of the proposed estimation. The workflow exhibited no systematic error in MAE prediction. This work represents a proof of concept about the feasibility of synthetic CT evaluation in daily clinical practice, and it paves the way for future patient-specific quality assessment strategies.
Collapse
Affiliation(s)
- Paolo Zaffino
- Department of Experimental and Clinical Medicine, Magna Graecia University, viale Europa, 88100 Catanzaro, Italy
- Department of Radiation Oncology, University Medical Center Groningen, University of Groningen, 9712 CP Groningen, The Netherlands
| | - Ciro Benito Raggio
- Institute of Biomedical Engineering, Karlsruhe Institute of Technology (KIT), 76131 Karlsruhe, Germany
| | - Adrian Thummerer
- Department of Radiation Oncology, University Medical Center Groningen, University of Groningen, 9712 CP Groningen, The Netherlands
- Department of Radiation Oncology, University Hospital, LMU Munich, 81377 Munich, Germany
| | - Gabriel Guterres Marmitt
- Department of Radiation Oncology, University Medical Center Groningen, University of Groningen, 9712 CP Groningen, The Netherlands
| | - Johannes Albertus Langendijk
- Department of Radiation Oncology, University Medical Center Groningen, University of Groningen, 9712 CP Groningen, The Netherlands
| | - Anna Procopio
- Department of Experimental and Clinical Medicine, Magna Graecia University, viale Europa, 88100 Catanzaro, Italy
| | - Carlo Cosentino
- Department of Experimental and Clinical Medicine, Magna Graecia University, viale Europa, 88100 Catanzaro, Italy
| | - Joao Seco
- Department of Biomedical Physics in Radiation Oncology, Deutsches Krebsfoschungszentrum (DKFZ), 69120 Heidelberg, Germany
- Department of Physics and Astronomy, Heidelberg University, 69120 Heidelberg, Germany
| | - Antje Christin Knopf
- Institute for Medical Engineering and Medical Informatics, School of Life Science FHNW, 4132 Muttenz, Switzerland
- Department of Radiotherapy and Radiation Oncology, Faculty of Medicine, University Hospital Carl Gustav Carus, Technische Universität Dresden, 01307 Dresden, Germany
| | - Stefan Both
- Department of Radiation Oncology, University Medical Center Groningen, University of Groningen, 9712 CP Groningen, The Netherlands
| | - Maria Francesca Spadea
- Institute of Biomedical Engineering, Karlsruhe Institute of Technology (KIT), 76131 Karlsruhe, Germany
| |
Collapse
|
3
|
Zhang J, Chen W, Joshi T, Uyanik M, Zhang X, Loh PL, Jog V, Bruce R, Garrett J, McMillan A. RobMedNAS: searching robust neural network architectures for medical image synthesis. Biomed Phys Eng Express 2024; 10:055029. [PMID: 39137798 PMCID: PMC11346166 DOI: 10.1088/2057-1976/ad6e87] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/10/2024] [Revised: 08/07/2024] [Accepted: 08/13/2024] [Indexed: 08/15/2024]
Abstract
Investigating U-Net model robustness in medical image synthesis against adversarial perturbations, this study introduces RobMedNAS, a neural architecture search strategy for identifying resilient U-Net configurations. Through retrospective analysis of synthesized CT from MRI data, employing Dice coefficient and mean absolute error metrics across critical anatomical areas, the study evaluates traditional U-Net models and RobMedNAS-optimized models under adversarial attacks. Findings demonstrate RobMedNAS's efficacy in enhancing U-Net resilience without compromising on accuracy, proposing a novel pathway for robust medical image processing.
Collapse
Affiliation(s)
- Jinnian Zhang
- Electrical and Computer Engineering, University of Wisconsin-Madison, Madison, United States of America
| | - Weijie Chen
- Electrical and Computer Engineering, University of Wisconsin-Madison, Madison, United States of America
| | - Tanmayee Joshi
- Electrical and Computer Engineering, University of Wisconsin-Madison, Madison, United States of America
| | - Meltem Uyanik
- Medical Physics, University of Wisconsin-Madison, Madison, United States of America
| | - Xiaomin Zhang
- Computer Science, University of Wisconsin-Madison, Madison, United States of America
| | - Po-Ling Loh
- Pure Mathematics and Mathematical Statistics, University of Cambridge, Cambridge, United Kingdom
| | - Varun Jog
- Pure Mathematics and Mathematical Statistics, University of Cambridge, Cambridge, United Kingdom
| | - Richard Bruce
- Radiology, University of Wisconsin-Madison, Madison, United States of America
| | - John Garrett
- Radiology, University of Wisconsin-Madison, Madison, United States of America
| | - Alan McMillan
- Electrical and Computer Engineering, University of Wisconsin-Madison, Madison, United States of America
- Medical Physics, University of Wisconsin-Madison, Madison, United States of America
- Radiology, University of Wisconsin-Madison, Madison, United States of America
- Biomedical Engineering, University of Wisconsin-Madison, Madison, United States of America
| |
Collapse
|
4
|
Yeap PL, Wong YM, Lee KH, Koh CWY, Lew KS, Chua CGA, Wibawa A, Master Z, Lee JCL, Park SY, Tan HQ. A treatment-site-specific evaluation of commercial synthetic computed tomography solutions for proton therapy. Phys Imaging Radiat Oncol 2024; 31:100639. [PMID: 39297079 PMCID: PMC11407964 DOI: 10.1016/j.phro.2024.100639] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/22/2024] [Revised: 08/25/2024] [Accepted: 08/28/2024] [Indexed: 09/21/2024] Open
Abstract
Background and purpose Despite the superior dose conformity of proton therapy, the dose distribution is sensitive to daily anatomical changes, which can affect treatment accuracy. This study evaluated the dose recalculation accuracy of two synthetic computed tomography (sCT) generation algorithms in a commercial treatment planning system. Materials and methods The evaluation was conducted for head-and-neck, thorax-and-abdomen, and pelvis sites treated with proton therapy. Thirty patients with two cone-beam computed tomography (CBCT) scans each were selected. The sCT images were generated from CBCT scans using two algorithms, Corrected CBCT (corrCBCT) and Virtual CT (vCT). Dose recalculations were performed based on these images for comparison with "ground truth" deformed CTs. Results The choice of algorithm influenced dose recalculation accuracy, particularly in high dose regions. For head-and-neck cases, the corrCBCT method showed closer agreement with the "ground truth", while for thorax-and-abdomen and pelvis cases, the vCT algorithm yielded better results (mean percentage dose discrepancy of 0.6 %, 1.3 % and 0.5 % for the three sites, respectively, in the high dose region). Head-and-neck and pelvis cases exhibited excellent agreement in high dose regions (2 %/2 mm gamma passing rate >98 %), while thorax-and-abdomen cases exhibited the largest differences, suggesting caution in sCT algorithm usage for this site. Significant systematic differences were observed in the clinical target volume and organ-at-risk doses in head-and-neck and pelvis cases, highlighting the importance of using the correct algorithm. Conclusions This study provided treatment site-specific recommendations for sCT algorithm selection in proton therapy. The findings offered insights for proton beam centers implementing adaptive radiotherapy workflows.
Collapse
Affiliation(s)
- Ping Lin Yeap
- Division of Radiation Oncology, National Cancer Centre Singapore, Singapore
- Department of Oncology, University of Cambridge, United Kingdom
| | - Yun Ming Wong
- Division of Radiation Oncology, National Cancer Centre Singapore, Singapore
- Division of Physics and Applied Physics, Nanyang Technological University, Singapore
| | - Kang Hao Lee
- Division of Radiation Oncology, National Cancer Centre Singapore, Singapore
| | | | - Kah Seng Lew
- Division of Radiation Oncology, National Cancer Centre Singapore, Singapore
- Division of Physics and Applied Physics, Nanyang Technological University, Singapore
| | - Clifford Ghee Ann Chua
- Division of Radiation Oncology, National Cancer Centre Singapore, Singapore
- Division of Physics and Applied Physics, Nanyang Technological University, Singapore
| | - Andrew Wibawa
- Division of Radiation Oncology, National Cancer Centre Singapore, Singapore
| | - Zubin Master
- Division of Radiation Oncology, National Cancer Centre Singapore, Singapore
| | - James Cheow Lei Lee
- Division of Radiation Oncology, National Cancer Centre Singapore, Singapore
- Division of Physics and Applied Physics, Nanyang Technological University, Singapore
| | - Sung Yong Park
- Division of Radiation Oncology, National Cancer Centre Singapore, Singapore
- Oncology Academic Clinical Programme, Duke-NUS Medical School, Singapore
| | - Hong Qi Tan
- Division of Radiation Oncology, National Cancer Centre Singapore, Singapore
- Division of Physics and Applied Physics, Nanyang Technological University, Singapore
- Oncology Academic Clinical Programme, Duke-NUS Medical School, Singapore
| |
Collapse
|
5
|
Vellini L, Zucca S, Lenkowicz J, Menna S, Catucci F, Quaranta F, Pilloni E, D'Aviero A, Aquilano M, Di Dio C, Iezzi M, Re A, Preziosi F, Piras A, Boschetti A, Piccari D, Mattiucci GC, Cusumano D. A Deep Learning Approach for the Fast Generation of Synthetic Computed Tomography from Low-Dose Cone Beam Computed Tomography Images on a Linear Accelerator Equipped with Artificial Intelligence. APPLIED SCIENCES 2024; 14:4844. [DOI: 10.3390/app14114844] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 11/08/2024]
Abstract
Artificial Intelligence (AI) is revolutionising many aspects of radiotherapy (RT), opening scenarios that were unimaginable just a few years ago. The aim of this study is to propose a Deep Leaning (DL) approach able to quickly generate synthetic Computed Tomography (CT) images from low-dose Cone Beam CT (CBCT) acquired on a modern linear accelerator integrating AI. Methods: A total of 53 patients treated in the pelvic region were enrolled and split into training (30), validation (9), and testing (14). A Generative Adversarial Network (GAN) was trained for 200 epochs. The image accuracy was evaluated by calculating the mean and mean absolute error (ME and ME) between sCT and CT. RT treatment plans were calculated on CT and sCT images, and dose accuracy was evaluated considering Dose Volume Histogram (DVH) and gamma analysis. Results: A total of 4507 images were selected for training. The MAE and ME values in the test set were 36 ± 6 HU and 7 ± 6 HU, respectively. Mean gamma passing rates for 1%/1 mm, 2%/2 mm, and 3%/3 mm tolerance criteria were respectively 93.5 ± 3.4%, 98.0 ± 1.3%, and 99.2 ± 0.7%, with no difference between curative and palliative cases. All the DVH parameters analysed were within 1 Gy of the difference between sCT and CT. Conclusion: This study demonstrated that sCT generation using the DL approach is feasible on low-dose CBCT images. The proposed approach can represent a valid tool to speed up the online adaptive procedure and remove CT simulation from the RT workflow.
Collapse
Affiliation(s)
| | - Sergio Zucca
- Azienda Ospedaliera Brotzu, 09047 Cagliari, Italy
| | - Jacopo Lenkowicz
- Fondazione Policlinico Universitario “Agostino Gemelli” IRCCS, 00168 Roma, Italy
| | | | | | | | | | | | | | | | | | - Alessia Re
- Mater Olbia Hospital, 07026 Olbia, Italy
| | | | - Antonio Piras
- UO Radioterapia Oncologica, Villa Santa Teresa, 90011 Palermo, Italy
| | | | | | - Gian Carlo Mattiucci
- Mater Olbia Hospital, 07026 Olbia, Italy
- Facoltà di Medicina e Chirurgia, Università Cattolica del Sacro Cuore, 00168 Roma, Italy
| | | |
Collapse
|
6
|
Muren LP. Physics and Imaging in Radiation Oncology comes of age. Phys Imaging Radiat Oncol 2024; 29:100559. [PMID: 38434208 PMCID: PMC10906384 DOI: 10.1016/j.phro.2024.100559] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Track Full Text] [Download PDF] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 03/05/2024] Open
Affiliation(s)
- Ludvig P Muren
- Physics and Imaging in Radiation Oncology, Danish Centre for Particle Therapy, Aarhus University Hospital / Aarhus University, Aarhus, Denmark
| |
Collapse
|