Hasan Sari1, Ja Reaungamornrat2, Onofrio A Catalano1, Javier Vera-Olmos3, David Izquierdo-Garcia1, Manuel A Morales1, Angel Torrado-Carvajal1,3, Thomas S C Ng4, Norberto Malpica3, Ali Kamen2, Ciprian Catana5. 1. Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital and Harvard Medical School, Charlestown, Massachusetts. 2. Siemens Corporate Research, Princeton, New Jersey. 3. Medical Image Analysis and Biometry Lab, Universidad Rey Juan Carlos, Madrid, Spain; and. 4. Department of Radiology, Massachusetts General Hospital, Boston, Massachusetts. 5. Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital and Harvard Medical School, Charlestown, Massachusetts; ccatana@mgh.harvard.edu.
Abstract
Attenuation correction remains a challenge in pelvic PET/MRI. In addition to the segmentation/model-based approaches, deep learning methods have shown promise in synthesizing accurate pelvic attenuation maps (μ-maps). However, these methods often misclassify air pockets in the digestive tract, potentially introducing bias in the reconstructed PET images. The aims of this work were to develop deep learning-based methods to automatically segment air pockets and generate pseudo-CT images from CAIPIRINHA-accelerated MR Dixon images. Methods: A convolutional neural network (CNN) was trained to segment air pockets using 3-dimensional CAIPIRINHA-accelerated MR Dixon datasets from 35 subjects and was evaluated against semiautomated segmentations. A separate CNN was trained to synthesize pseudo-CT μ-maps from the Dixon images. Its accuracy was evaluated by comparing the deep learning-, model-, and CT-based μ-maps using data from 30 of the subjects. Finally, the impact of different μ-maps and air pocket segmentation methods on the PET quantification was investigated. Results: Air pockets segmented using the CNN agreed well with semiautomated segmentations, with a mean Dice similarity coefficient of 0.75. The volumetric similarity score between 2 segmentations was 0.85 ± 0.14. The mean absolute relative changes with respect to the CT-based μ-maps were 2.6% and 5.1% in the whole pelvis for the deep learning-based and model-based μ-maps, respectively. The average relative change between PET images reconstructed with deep learning-based and CT-based μ-maps was 2.6%. Conclusion: We developed a deep learning-based method to automatically segment air pockets from CAIPIRINHA-accelerated Dixon images, with accuracy comparable to that of semiautomatic segmentations. The μ-maps synthesized using a deep learning-based method from CAIPIRINHA-accelerated Dixon images were more accurate than those generated with the model-based approach available on integrated PET/MRI scanners.
Attenuation correction remains a challenge in pelvic PET/MRI. In addition to the segmentation/model-based approaches, deep learning methods have shown promise in synthesizing accurate pelvic attenuation maps (μ-maps). However, these methods often misclassify air pockets in the digestive tract, potentially introducing bias in the reconstructed PET images. The aims of this work were to develop deep learning-based methods to automatically segment air pockets and generate pseudo-CT images from CAIPIRINHA-accelerated MR Dixon images. Methods: A convolutional neural network (CNN) was trained to segment air pockets using 3-dimensional CAIPIRINHA-accelerated MR Dixon datasets from 35 subjects and was evaluated against semiautomated segmentations. A separate CNN was trained to synthesize pseudo-CT μ-maps from the Dixon images. Its accuracy was evaluated by comparing the deep learning-, model-, and CT-based μ-maps using data from 30 of the subjects. Finally, the impact of different μ-maps and air pocket segmentation methods on the PET quantification was investigated. Results: Air pockets segmented using the CNN agreed well with semiautomated segmentations, with a mean Dice similarity coefficient of 0.75. The volumetric similarity score between 2 segmentations was 0.85 ± 0.14. The mean absolute relative changes with respect to the CT-based μ-maps were 2.6% and 5.1% in the whole pelvis for the deep learning-based and model-based μ-maps, respectively. The average relative change between PET images reconstructed with deep learning-based and CT-based μ-maps was 2.6%. Conclusion: We developed a deep learning-based method to automatically segment air pockets from CAIPIRINHA-accelerated Dixon images, with accuracy comparable to that of semiautomatic segmentations. The μ-maps synthesized using a deep learning-based method from CAIPIRINHA-accelerated Dixon images were more accurate than those generated with the model-based approach available on integrated PET/MRI scanners.
Authors: Andrew P Leynes; Jaewon Yang; Florian Wiesinger; Sandeep S Kaushik; Dattesh D Shanbhag; Youngho Seo; Thomas A Hope; Peder E Z Larson Journal: J Nucl Med Date: 2017-10-30 Impact factor: 10.057
Authors: Guha Balakrishnan; Amy Zhao; Mert R Sabuncu; John Guttag; Adrian V Dalca Journal: IEEE Trans Med Imaging Date: 2019-02-04 Impact factor: 10.048
Authors: Angel Torrado-Carvajal; Javier Vera-Olmos; David Izquierdo-Garcia; Onofrio A Catalano; Manuel A Morales; Justin Margolin; Andrea Soricelli; Marco Salvatore; Norberto Malpica; Ciprian Catana Journal: J Nucl Med Date: 2018-08-30 Impact factor: 10.057
Authors: Onofrio Antonio Catalano; Vincent Wu; Umar Mahmood; Alberto Signore; Mark Vangel; Andrea Soricelli; Marco Salvatore; Debra Gervais; Bruce R Rosen Journal: Am J Nucl Med Mol Imaging Date: 2018-02-05
Authors: Andrii Pozaruk; Kamlesh Pawar; Shenpeng Li; Alexandra Carey; Jeremy Cheng; Viswanath P Sudarshan; Marian Cholewa; Jeremy Grummet; Zhaolin Chen; Gary Egan Journal: Eur J Nucl Med Mol Imaging Date: 2020-05-11 Impact factor: 9.236
Authors: Nicholas J Tustison; Brian B Avants; Philip A Cook; Yuanjie Zheng; Alexander Egan; Paul A Yushkevich; James C Gee Journal: IEEE Trans Med Imaging Date: 2010-04-08 Impact factor: 10.048
Authors: Matteo Maspero; Mark H F Savenije; Anna M Dinkla; Peter R Seevinck; Martijn P W Intven; Ina M Jurgenliemk-Schulz; Linda G W Kerkmeijer; Cornelis A T van den Berg Journal: Phys Med Biol Date: 2018-09-10 Impact factor: 3.609
Authors: Konstantinos Kamnitsas; Christian Ledig; Virginia F J Newcombe; Joanna P Simpson; Andrew D Kane; David K Menon; Daniel Rueckert; Ben Glocker Journal: Med Image Anal Date: 2016-10-29 Impact factor: 8.545
Authors: Hasan Sari; Mohammadreza Teimoorisichani; Clemens Mingels; Ian Alberts; Vladimir Panin; Deepak Bharkhada; Song Xue; George Prenosil; Kuangyu Shi; Maurizio Conti; Axel Rominger Journal: Eur J Nucl Med Mol Imaging Date: 2022-07-19 Impact factor: 10.057