-
Something wrong with this record ?
Development and validation of pan-cancer lesion segmentation AI-model for whole-body 18F-FDG PET/CT in diverse clinical cohorts
F. Haque, A. Chen, N. Lay, J. Carrasquillo, E. Mena, L. Lindenberg, JE. Segal, PC. Eclarinal, S. Talvacchio, A. Derkyi, PL. Choyke, K. Pacak, RN. Kaplan, FI. Lin, B. Turkbey, SA. Harmon
Language English Country United States
Document type Journal Article, Validation Study
Grant support
Z99 CA999999
Intramural NIH HHS - United States
- MeSH
- Whole Body Imaging * methods MeSH
- Deep Learning * MeSH
- Child MeSH
- Adult MeSH
- Fluorodeoxyglucose F18 * MeSH
- Cohort Studies MeSH
- Middle Aged MeSH
- Humans MeSH
- Adolescent MeSH
- Neoplasms * diagnostic imaging MeSH
- Positron Emission Tomography Computed Tomography * methods MeSH
- Image Processing, Computer-Assisted * methods MeSH
- Check Tag
- Child MeSH
- Adult MeSH
- Middle Aged MeSH
- Humans MeSH
- Adolescent MeSH
- Male MeSH
- Female MeSH
- Publication type
- Journal Article MeSH
- Validation Study MeSH
BACKGROUND: This study develops a deep learning-based automated lesion segmentation model for whole-body 3D18F-fluorodeoxyglucose (FDG)-Position emission tomography (PET) with computed tomography (CT) images agnostic to disease location and site. METHOD: A publicly available lesion-annotated dataset of 1014 whole-body FDG-PET/CT images was used to train, validate, and test (70:10:20) eight configurations with 3D U-Net as the backbone architecture. The best-performing model on the test set was further evaluated on 3 different unseen cohorts consisting of osteosarcoma or neuroblastoma (OS cohort) (n = 13), pediatric solid tumors (ST cohort) (n = 14), and adult Pheochromocytoma/Paraganglioma (PHEO cohort) (n = 40). Both lesion-level and patient-level statistical analyses were conducted to validate the performance of the model on different cohorts. RESULTS: The best performing 3D full resolution nnUNet model achieved a lesion-level sensitivity and DISC of 71.70 % and 0.40 for the test set, 97.83 % and 0.73 for ST, 40.15 % and 0.36 for OS, and 78.37 % and 0.50 for the PHEO cohort. For the test set and PHEO cohort, the model has missed small volume and lower uptake lesions (p < 0.01), whereas no statistically significant differences (p > 0.05) were found in the false positive (FP) and false negative lesions volume and uptake for the OS and ST cohort. The predicted total lesion glycolysis is slightly higher than the ground truth because of FP calls, which experts can easily check and reject. CONCLUSION: The developed deep learning-based automated lesion segmentation AI model which utilizes 3D_FullRes configuration of the nnUNet framework showed promising and reliable performance for the whole-body FDG-PET/CT images.
References provided by Crossref.org
- 000
- 00000naa a2200000 a 4500
- 001
- bmc25015899
- 003
- CZ-PrNML
- 005
- 20250731091336.0
- 007
- ta
- 008
- 250708e20250323xxu f 000 0|eng||
- 009
- AR
- 024 7_
- $a 10.1016/j.compbiomed.2025.110052 $2 doi
- 035 __
- $a (PubMed)40127518
- 040 __
- $a ABA008 $b cze $d ABA008 $e AACR2
- 041 0_
- $a eng
- 044 __
- $a xxu
- 100 1_
- $a Haque, Fahmida $u Artificial Intelligence Resource, National Cancer Institute, National Institute of Health, Bethesda, MD, 20814, USA
- 245 10
- $a Development and validation of pan-cancer lesion segmentation AI-model for whole-body 18F-FDG PET/CT in diverse clinical cohorts / $c F. Haque, A. Chen, N. Lay, J. Carrasquillo, E. Mena, L. Lindenberg, JE. Segal, PC. Eclarinal, S. Talvacchio, A. Derkyi, PL. Choyke, K. Pacak, RN. Kaplan, FI. Lin, B. Turkbey, SA. Harmon
- 520 9_
- $a BACKGROUND: This study develops a deep learning-based automated lesion segmentation model for whole-body 3D18F-fluorodeoxyglucose (FDG)-Position emission tomography (PET) with computed tomography (CT) images agnostic to disease location and site. METHOD: A publicly available lesion-annotated dataset of 1014 whole-body FDG-PET/CT images was used to train, validate, and test (70:10:20) eight configurations with 3D U-Net as the backbone architecture. The best-performing model on the test set was further evaluated on 3 different unseen cohorts consisting of osteosarcoma or neuroblastoma (OS cohort) (n = 13), pediatric solid tumors (ST cohort) (n = 14), and adult Pheochromocytoma/Paraganglioma (PHEO cohort) (n = 40). Both lesion-level and patient-level statistical analyses were conducted to validate the performance of the model on different cohorts. RESULTS: The best performing 3D full resolution nnUNet model achieved a lesion-level sensitivity and DISC of 71.70 % and 0.40 for the test set, 97.83 % and 0.73 for ST, 40.15 % and 0.36 for OS, and 78.37 % and 0.50 for the PHEO cohort. For the test set and PHEO cohort, the model has missed small volume and lower uptake lesions (p < 0.01), whereas no statistically significant differences (p > 0.05) were found in the false positive (FP) and false negative lesions volume and uptake for the OS and ST cohort. The predicted total lesion glycolysis is slightly higher than the ground truth because of FP calls, which experts can easily check and reject. CONCLUSION: The developed deep learning-based automated lesion segmentation AI model which utilizes 3D_FullRes configuration of the nnUNet framework showed promising and reliable performance for the whole-body FDG-PET/CT images.
- 650 _2
- $a lidé $7 D006801
- 650 12
- $a fluorodeoxyglukosa F18 $7 D019788
- 650 12
- $a PET/CT $x metody $7 D000072078
- 650 12
- $a nádory $x diagnostické zobrazování $7 D009369
- 650 _2
- $a ženské pohlaví $7 D005260
- 650 12
- $a celotělové zobrazování $x metody $7 D051598
- 650 _2
- $a mužské pohlaví $7 D008297
- 650 _2
- $a dospělí $7 D000328
- 650 12
- $a deep learning $7 D000077321
- 650 _2
- $a dítě $7 D002648
- 650 _2
- $a kohortové studie $7 D015331
- 650 12
- $a počítačové zpracování obrazu $x metody $7 D007091
- 650 _2
- $a lidé středního věku $7 D008875
- 650 _2
- $a mladiství $7 D000293
- 655 _2
- $a časopisecké články $7 D016428
- 655 _2
- $a validační studie $7 D023361
- 700 1_
- $a Chen, Alex $u Artificial Intelligence Resource, National Cancer Institute, National Institute of Health, Bethesda, MD, 20814, USA
- 700 1_
- $a Lay, Nathan $u Artificial Intelligence Resource, National Cancer Institute, National Institute of Health, Bethesda, MD, 20814, USA
- 700 1_
- $a Carrasquillo, Jorge $u Molecular Imaging Branch, National Cancer Institute, National Institute of Health, Bethesda, MD, 20892, USA
- 700 1_
- $a Mena, Esther $u Molecular Imaging Branch, National Cancer Institute, National Institute of Health, Bethesda, MD, 20892, USA
- 700 1_
- $a Lindenberg, Liza $u Molecular Imaging Branch, National Cancer Institute, National Institute of Health, Bethesda, MD, 20892, USA
- 700 1_
- $a Segal, Julia E $u Section on Medical Neuroendocrinology National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, 20892, MD, USA
- 700 1_
- $a Eclarinal, Philip C $u Molecular Imaging Branch, National Cancer Institute, National Institute of Health, Bethesda, MD, 20892, USA
- 700 1_
- $a Talvacchio, Sara $u Section on Medical Neuroendocrinology National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, 20892, MD, USA
- 700 1_
- $a Derkyi, Alberta $u Section on Medical Neuroendocrinology National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, 20892, MD, USA
- 700 1_
- $a Choyke, Peter L $u Molecular Imaging Branch, National Cancer Institute, National Institute of Health, Bethesda, MD, 20892, USA
- 700 1_
- $a Pacak, Karel $u Section on Medical Neuroendocrinology National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, 20892, MD, USA; AKESO, Prague 5, Czech Republic
- 700 1_
- $a Kaplan, Rosandra N $u Pediatric Oncology Branch, National Cancer Institute, National Institute of Health, Bethesda, MD, 20892, USA
- 700 1_
- $a Lin, Frank I $u Molecular Imaging Branch, National Cancer Institute, National Institute of Health, Bethesda, MD, 20892, USA
- 700 1_
- $a Turkbey, Baris $u Artificial Intelligence Resource, National Cancer Institute, National Institute of Health, Bethesda, MD, 20814, USA; Molecular Imaging Branch, National Cancer Institute, National Institute of Health, Bethesda, MD, 20892, USA
- 700 1_
- $a Harmon, Stephanie A $u Artificial Intelligence Resource, National Cancer Institute, National Institute of Health, Bethesda, MD, 20814, USA
- 773 0_
- $w MED00001218 $t Computers in biology and medicine $x 1879-0534 $g Roč. 190 (20250323), s. 110052
- 856 41
- $u https://pubmed.ncbi.nlm.nih.gov/40127518 $y Pubmed
- 910 __
- $a ABA008 $b sig $c sign $y - $z 0
- 990 __
- $a 20250708 $b ABA008
- 991 __
- $a 20250731091330 $b ABA008
- 999 __
- $a ok $b bmc $g 2366617 $s 1253024
- BAS __
- $a 3
- BAS __
- $a PreBMC-MEDLINE
- BMC __
- $a 2025 $b 190 $c - $d 110052 $e 20250323 $i 1879-0534 $m Computers in biology and medicine $n Comput Biol Med $x MED00001218
- GRA __
- $a Z99 CA999999 $p Intramural NIH HHS $2 United States
- LZP __
- $a Pubmed-20250708
