Jian Peng1, Daniel D Kim2, Jay B Patel3, Xiaowei Zeng1, Jiaer Huang4, Ken Chang3, Xinping Xun1, Chen Zhang1, John Sollee2, Jing Wu5, Deepa J Dalal6, Xue Feng7, Hao Zhou8, Chengzhang Zhu1,4, Beiji Zou4, Ke Jin9, Patrick Y Wen10, Jerrold L Boxerman2, Katherine E Warren11, Tina Y Poussaint12, Lisa J States6, Jayashree Kalpathy-Cramer3, Li Yang1, Raymond Y Huang13, Harrison X Bai2. 1. Department of Neurology, The Second Xiangya Hospital of Central South University, Changsha, Hunan, China. 2. Department of Diagnostic Imaging, Rhode Island Hospital and Alpert Medical School of Brown University, Providence, Rhode Island, USA. 3. Department of Radiology, Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, USA. 4. School of Computer Science and Engineering, Central South University, Changsha, Hunan, China. 5. Department of Radiology, The Second Xiangya Hospital of Central South University, Changsha, Hunan, China. 6. Department of Radiology, Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA. 7. Department of Biomedical Engineering, University of Virginia, Charlottesville, Virginia, USA. 8. Department of Neurology, Xiangya Hospital of Central South University, Changsha, Hunan, China. 9. Department of Radiology, Hunan Children's Hospital, Changsha, Hunan, China. 10. Center for Neuro-Oncology, Dana Farber Cancer Institute, Boston, Massachusetts, USA. 11. Department of Pediatrics, Dana Farber Cancer Institute, Boston, Massachusetts, USA. 12. Department of Radiology, Boston Children's Hospital, Boston, Massachusetts, USA. 13. Department of Radiology, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts, USA.
Abstract
BACKGROUND: Longitudinal measurement of tumor burden with magnetic resonance imaging (MRI) is an essential component of response assessment in pediatric brain tumors. We developed a fully automated pipeline for the segmentation of tumors in pediatric high-grade gliomas, medulloblastomas, and leptomeningeal seeding tumors. We further developed an algorithm for automatic 2D and volumetric size measurement of tumors. METHODS: The preoperative and postoperative cohorts were randomly split into training and testing sets in a 4:1 ratio. A 3D U-Net neural network was trained to automatically segment the tumor on T1 contrast-enhanced and T2/FLAIR images. The product of the maximum bidimensional diameters according to the RAPNO (Response Assessment in Pediatric Neuro-Oncology) criteria (AutoRAPNO) was determined. Performance was compared to that of 2 expert human raters who performed assessments independently. Volumetric measurements of predicted and expert segmentations were computationally derived and compared. RESULTS: A total of 794 preoperative MRIs from 794 patients and 1003 postoperative MRIs from 122 patients were included. There was excellent agreement of volumes between preoperative and postoperative predicted and manual segmentations, with intraclass correlation coefficients (ICCs) of 0.912 and 0.960 for the 2 preoperative and 0.947 and 0.896 for the 2 postoperative models. There was high agreement between AutoRAPNO scores on predicted segmentations and manually calculated scores based on manual segmentations (Rater 2 ICC = 0.909; Rater 3 ICC = 0.851). Lastly, the performance of AutoRAPNO was superior in repeatability to that of human raters for MRIs with multiple lesions. CONCLUSIONS: Our automated deep learning pipeline demonstrates potential utility for response assessment in pediatric brain tumors. The tool should be further validated in prospective studies.
BACKGROUND: Longitudinal measurement of tumor burden with magnetic resonance imaging (MRI) is an essential component of response assessment in pediatric brain tumors. We developed a fully automated pipeline for the segmentation of tumors in pediatric high-grade gliomas, medulloblastomas, and leptomeningeal seeding tumors. We further developed an algorithm for automatic 2D and volumetric size measurement of tumors. METHODS: The preoperative and postoperative cohorts were randomly split into training and testing sets in a 4:1 ratio. A 3D U-Net neural network was trained to automatically segment the tumor on T1 contrast-enhanced and T2/FLAIR images. The product of the maximum bidimensional diameters according to the RAPNO (Response Assessment in Pediatric Neuro-Oncology) criteria (AutoRAPNO) was determined. Performance was compared to that of 2 expert human raters who performed assessments independently. Volumetric measurements of predicted and expert segmentations were computationally derived and compared. RESULTS: A total of 794 preoperative MRIs from 794 patients and 1003 postoperative MRIs from 122 patients were included. There was excellent agreement of volumes between preoperative and postoperative predicted and manual segmentations, with intraclass correlation coefficients (ICCs) of 0.912 and 0.960 for the 2 preoperative and 0.947 and 0.896 for the 2 postoperative models. There was high agreement between AutoRAPNO scores on predicted segmentations and manually calculated scores based on manual segmentations (Rater 2 ICC = 0.909; Rater 3 ICC = 0.851). Lastly, the performance of AutoRAPNO was superior in repeatability to that of human raters for MRIs with multiple lesions. CONCLUSIONS: Our automated deep learning pipeline demonstrates potential utility for response assessment in pediatric brain tumors. The tool should be further validated in prospective studies.
Authors: Benjamin M Ellingson; Martin Bendszus; Jerrold Boxerman; Daniel Barboriak; Bradley J Erickson; Marion Smits; Sarah J Nelson; Elizabeth Gerstner; Brian Alexander; Gregory Goldmacher; Wolfgang Wick; Michael Vogelbaum; Michael Weller; Evanthia Galanis; Jayashree Kalpathy-Cramer; Lalitha Shankar; Paula Jacobs; Whitney B Pope; Dewen Yang; Caroline Chung; Michael V Knopp; Soonme Cha; Martin J van den Bent; Susan Chang; W K Al Yung; Timothy F Cloughesy; Patrick Y Wen; Mark R Gilbert Journal: Neuro Oncol Date: 2015-08-05 Impact factor: 12.300
Authors: M A Deeley; A Chen; R Datteri; J H Noble; A J Cmelak; E F Donnelly; A W Malcolm; L Moretti; J Jaboin; K Niermann; Eddy S Yang; David S Yu; F Yei; T Koyama; G X Ding; B M Dawant Journal: Phys Med Biol Date: 2011-07-01 Impact factor: 3.609
Authors: A L Albright; R Sposto; E Holmes; P M Zeltzer; J L Finlay; J H Wisoff; M S Berger; R J Packer; I F Pollack Journal: Neurosurgery Date: 2000-10 Impact factor: 4.654
Authors: Craig Erker; Benita Tamrazi; Tina Y Poussaint; Sabine Mueller; Daddy Mata-Mbemba; Enrico Franceschi; Alba A Brandes; Arvind Rao; Kellie B Haworth; Patrick Y Wen; Stewart Goldman; Gilbert Vezina; Tobey J MacDonald; Ira J Dunkel; Paul S Morgan; Tim Jaspan; Michael D Prados; Katherine E Warren Journal: Lancet Oncol Date: 2020-06 Impact factor: 54.433
Authors: J L Quon; W Bala; L C Chen; J Wright; L H Kim; M Han; K Shpanskaya; E H Lee; E Tong; M Iv; J Seekins; M P Lungren; K R M Braun; T Y Poussaint; S Laughlin; M D Taylor; R M Lober; H Vogel; P G Fisher; G A Grant; V Ramaswamy; N A Vitanza; C Y Ho; M S B Edwards; S H Cheshier; K W Yeom Journal: AJNR Am J Neuroradiol Date: 2020-08-13 Impact factor: 4.966