Nicholas J Tustison1, Brian B Avants2, Zixuan Lin3, Xue Feng3, Nicholas Cullen4, Jaime F Mata3, Lucia Flors5, James C Gee4, Talissa A Altes5, John P Mugler Iii3, Kun Qing3. 1. Department of Radiology and Medical Imaging, University of Virginia, Charlottesville, Virginia. Electronic address: ntustison@virginia.edu. 2. Cingulate, Hampton, New Hampshire. 3. Department of Radiology and Medical Imaging, University of Virginia, Charlottesville, Virginia. 4. Department of Radiology, University of Pennsylvania, Philadelphia, Pennsylvania. 5. Department of Radiology, University of Missouri, Columbia, Missouri.
Abstract
RATIONALE AND OBJECTIVES: We propose an automated segmentation pipeline based on deep learning for proton lung MRI segmentation and ventilation-based quantification which improves on our previously reported methodologies in terms of computational efficiency while demonstrating accuracy and robustness. The large data requirement for the proposed framework is made possible by a novel template-based data augmentation strategy. Supporting this work is the open-source ANTsRNet-a growing repository of well-known deep learning architectures first introduced here. MATERIALS AND METHODS: Deep convolutional neural network (CNN) models were constructed and trained using a custom multilabel Dice metric loss function and a novel template-based data augmentation strategy. Training (including template generation and data augmentation) employed 205 proton MR images and 73 functional lung MRI. Evaluation was performed using data sets of size 63 and 40 images, respectively. RESULTS: Accuracy for CNN-based proton lung MRI segmentation (in terms of Dice overlap) was left lung: 0.93 ± 0.03, right lung: 0.94 ± 0.02, and whole lung: 0.94 ± 0.02. Although slightly less accurate than our previously reported joint label fusion approach (left lung: 0.95 ± 0.02, right lung: 0.96 ± 0.01, and whole lung: 0.96 ± 0.01), processing time is <1 second per subject for the proposed approach versus ∼30 minutes per subject using joint label fusion. Accuracy for quantifying ventilation defects was determined based on a consensus labeling where average accuracy (Dice multilabel overlap of ventilation defect regions plus normal region) was 0.94 for the CNN method; 0.92 for our previously reported method; and 0.90, 0.92, and 0.94 for expert readers. CONCLUSION: The proposed framework yields accurate automated quantification in near real time. CNNs drastically reduce processing time after offline model construction and demonstrate significant future potential for facilitating quantitative analysis of functional lung MRI.
RATIONALE AND OBJECTIVES: We propose an automated segmentation pipeline based on deep learning for proton lung MRI segmentation and ventilation-based quantification which improves on our previously reported methodologies in terms of computational efficiency while demonstrating accuracy and robustness. The large data requirement for the proposed framework is made possible by a novel template-based data augmentation strategy. Supporting this work is the open-source ANTsRNet-a growing repository of well-known deep learning architectures first introduced here. MATERIALS AND METHODS: Deep convolutional neural network (CNN) models were constructed and trained using a custom multilabel Dice metric loss function and a novel template-based data augmentation strategy. Training (including template generation and data augmentation) employed 205 proton MR images and 73 functional lung MRI. Evaluation was performed using data sets of size 63 and 40 images, respectively. RESULTS: Accuracy for CNN-based proton lung MRI segmentation (in terms of Dice overlap) was left lung: 0.93 ± 0.03, right lung: 0.94 ± 0.02, and whole lung: 0.94 ± 0.02. Although slightly less accurate than our previously reported joint label fusion approach (left lung: 0.95 ± 0.02, right lung: 0.96 ± 0.01, and whole lung: 0.96 ± 0.01), processing time is <1 second per subject for the proposed approach versus ∼30 minutes per subject using joint label fusion. Accuracy for quantifying ventilation defects was determined based on a consensus labeling where average accuracy (Dice multilabel overlap of ventilation defect regions plus normal region) was 0.94 for the CNN method; 0.92 for our previously reported method; and 0.90, 0.92, and 0.94 for expert readers. CONCLUSION: The proposed framework yields accurate automated quantification in near real time. CNNs drastically reduce processing time after offline model construction and demonstrate significant future potential for facilitating quantitative analysis of functional lung MRI.
Authors: Leon Lenchik; Laura Heacock; Ashley A Weaver; Robert D Boutin; Tessa S Cook; Jason Itri; Christopher G Filippi; Rao P Gullapalli; James Lee; Marianna Zagurovskaya; Tara Retson; Kendra Godwin; Joey Nicholson; Ponnada A Narayana Journal: Acad Radiol Date: 2019-08-10 Impact factor: 3.173
Authors: Andrew T Grainger; Arun Krishnaraj; Michael H Quinones; Nicholas J Tustison; Samantha Epstein; Daniela Fuller; Aakash Jha; Kevin L Allman; Weibin Shi Journal: Acad Radiol Date: 2020-08-05 Impact factor: 3.173
Authors: Rachel E Lackie; Aline S de Miranda; Mei Peng Lim; Vladislav Novikov; Nimrod Madrer; Nadun C Karunatilleke; Benjamin S Rutledge; Stephanie Tullo; Anne Brickenden; Matthew E R Maitland; David Greenberg; Daniel Gallino; Wen Luo; Anoosha Attaran; Irina Shlaifer; Esther Del Cid Pellitero; Caroline Schild-Poulter; Thomas M Durcan; Edward A Fon; Martin Duennwald; Flavio H Beraldo; M Mallar Chakravarty; Timothy J Bussey; Lisa M Saksida; Hermona Soreq; Wing-Yiu Choy; Vania F Prado; Marco A M Prado Journal: Acta Neuropathol Date: 2022-09-19 Impact factor: 15.887
Authors: Joshua R Astley; Alberto M Biancardi; Paul J C Hughes; Helen Marshall; Laurie J Smith; Guilhem J Collier; James A Eaden; Nicholas D Weatherley; Matthew Q Hatton; Jim M Wild; Bilal A Tahir Journal: Sci Rep Date: 2022-06-22 Impact factor: 4.996
Authors: Nicholas J Tustison; Philip A Cook; Andrew J Holbrook; Hans J Johnson; John Muschelli; Gabriel A Devenyi; Jeffrey T Duda; Sandhitsu R Das; Nicholas C Cullen; Daniel L Gillen; Michael A Yassa; James R Stone; James C Gee; Brian B Avants Journal: Sci Rep Date: 2021-04-27 Impact factor: 4.379