Deep learning for segmentation of 49 selected bones in CT scans: First step in automated PET/CT-based 3D quantification of skeletal metastases

Sarah Lindgren Belal, May Sadik, Reza Kaboteh, Olof Enqvist, Johannes Ulén, Mads Hvid Poulsen, Jane Angel Simonsen, Poul Flemming Høilund-Carlsen, Lars Edenbrandt, Elin Trägårdh

Publikation: Bidrag til tidsskriftTidsskriftartikelForskningpeer review

99 Downloads (Pure)

Resumé

Purpose: The aim of this study was to develop a deep learning-based method for segmentation of bones in CT scans and test its accuracy compared to manual delineation, as a first step in the creation of an automated PET/CT-based method for quantifying skeletal tumour burden. Methods: Convolutional neural networks (CNNs) were trained to segment 49 bones using manual segmentations from 100 CT scans. After training, the CNN-based segmentation method was tested on 46 patients with prostate cancer, who had undergone 18 F-choline-PET/CT and 18 F-NaF PET/CT less than three weeks apart. Bone volumes were calculated from the segmentations. The network's performance was compared with manual segmentations of five bones made by an experienced physician. Accuracy of the spatial overlap between automated CNN-based and manual segmentations of these five bones was assessed using the Sørensen-Dice index (SDI). Reproducibility was evaluated applying the Bland-Altman method. Results: The median (SD) volumes of the five selected bones were by CNN and manual segmentation: Th7 41 (3.8) and 36 (5.1), L3 76 (13) and 75 (9.2), sacrum 284 (40) and 283 (26), 7th rib 33 (3.9) and 31 (4.8), sternum 80 (11) and 72 (9.2), respectively. Median SDIs were 0.86 (Th7), 0.85 (L3), 0.88 (sacrum), 0.84 (7th rib) and 0.83 (sternum). The intraobserver volume difference was less with CNN-based than manual approach: Th7 2% and 14%, L3 7% and 8%, sacrum 1% and 3%, 7th rib 1% and 6%, sternum 3% and 5%, respectively. The average volume difference measured as ratio volume difference/mean volume between the two CNN-based segmentations was 5–6% for the vertebral column and ribs and ≤3% for other bones. Conclusion: The new deep learning-based method for automated segmentation of bones in CT scans provided highly accurate bone volumes in a fast and automated way and, thus, appears to be a valuable first step in the development of a clinical useful processing procedure providing reliable skeletal segmentation as a key part of quantification of skeletal metastases.

OriginalsprogEngelsk
TidsskriftEuropean Journal of Radiology
Vol/bind113
Sider (fra-til)89-95
ISSN0720-048X
DOI
StatusUdgivet - 2019

Fingeraftryk

Sacrum
Choline
Tumor Burden
Prostatic Neoplasms
Physicians

Citer dette

@article{a21ce228e69b44289d3813d830ec22d8,
title = "Deep learning for segmentation of 49 selected bones in CT scans: First step in automated PET/CT-based 3D quantification of skeletal metastases",
abstract = "Purpose: The aim of this study was to develop a deep learning-based method for segmentation of bones in CT scans and test its accuracy compared to manual delineation, as a first step in the creation of an automated PET/CT-based method for quantifying skeletal tumour burden. Methods: Convolutional neural networks (CNNs) were trained to segment 49 bones using manual segmentations from 100 CT scans. After training, the CNN-based segmentation method was tested on 46 patients with prostate cancer, who had undergone 18 F-choline-PET/CT and 18 F-NaF PET/CT less than three weeks apart. Bone volumes were calculated from the segmentations. The network's performance was compared with manual segmentations of five bones made by an experienced physician. Accuracy of the spatial overlap between automated CNN-based and manual segmentations of these five bones was assessed using the S{\o}rensen-Dice index (SDI). Reproducibility was evaluated applying the Bland-Altman method. Results: The median (SD) volumes of the five selected bones were by CNN and manual segmentation: Th7 41 (3.8) and 36 (5.1), L3 76 (13) and 75 (9.2), sacrum 284 (40) and 283 (26), 7th rib 33 (3.9) and 31 (4.8), sternum 80 (11) and 72 (9.2), respectively. Median SDIs were 0.86 (Th7), 0.85 (L3), 0.88 (sacrum), 0.84 (7th rib) and 0.83 (sternum). The intraobserver volume difference was less with CNN-based than manual approach: Th7 2{\%} and 14{\%}, L3 7{\%} and 8{\%}, sacrum 1{\%} and 3{\%}, 7th rib 1{\%} and 6{\%}, sternum 3{\%} and 5{\%}, respectively. The average volume difference measured as ratio volume difference/mean volume between the two CNN-based segmentations was 5–6{\%} for the vertebral column and ribs and ≤3{\%} for other bones. Conclusion: The new deep learning-based method for automated segmentation of bones in CT scans provided highly accurate bone volumes in a fast and automated way and, thus, appears to be a valuable first step in the development of a clinical useful processing procedure providing reliable skeletal segmentation as a key part of quantification of skeletal metastases.",
keywords = "Artificial intelligence, Bone, Metastases, PET/CT, Prostate cancer",
author = "Belal, {Sarah Lindgren} and May Sadik and Reza Kaboteh and Olof Enqvist and Johannes Ul{\'e}n and Poulsen, {Mads Hvid} and Simonsen, {Jane Angel} and H{\o}ilund-Carlsen, {Poul Flemming} and Lars Edenbrandt and Elin Tr{\"a}g{\aa}rdh",
year = "2019",
doi = "10.1016/j.ejrad.2019.01.028",
language = "English",
volume = "113",
pages = "89--95",
journal = "European Journal of Radiology",
issn = "0720-048X",
publisher = "Elsevier Ireland Ltd",

}

Deep learning for segmentation of 49 selected bones in CT scans: First step in automated PET/CT-based 3D quantification of skeletal metastases. / Belal, Sarah Lindgren; Sadik, May; Kaboteh, Reza; Enqvist, Olof; Ulén, Johannes; Poulsen, Mads Hvid; Simonsen, Jane Angel; Høilund-Carlsen, Poul Flemming; Edenbrandt, Lars; Trägårdh, Elin.

I: European Journal of Radiology, Bind 113, 2019, s. 89-95.

Publikation: Bidrag til tidsskriftTidsskriftartikelForskningpeer review

TY - JOUR

T1 - Deep learning for segmentation of 49 selected bones in CT scans: First step in automated PET/CT-based 3D quantification of skeletal metastases

AU - Belal, Sarah Lindgren

AU - Sadik, May

AU - Kaboteh, Reza

AU - Enqvist, Olof

AU - Ulén, Johannes

AU - Poulsen, Mads Hvid

AU - Simonsen, Jane Angel

AU - Høilund-Carlsen, Poul Flemming

AU - Edenbrandt, Lars

AU - Trägårdh, Elin

PY - 2019

Y1 - 2019

N2 - Purpose: The aim of this study was to develop a deep learning-based method for segmentation of bones in CT scans and test its accuracy compared to manual delineation, as a first step in the creation of an automated PET/CT-based method for quantifying skeletal tumour burden. Methods: Convolutional neural networks (CNNs) were trained to segment 49 bones using manual segmentations from 100 CT scans. After training, the CNN-based segmentation method was tested on 46 patients with prostate cancer, who had undergone 18 F-choline-PET/CT and 18 F-NaF PET/CT less than three weeks apart. Bone volumes were calculated from the segmentations. The network's performance was compared with manual segmentations of five bones made by an experienced physician. Accuracy of the spatial overlap between automated CNN-based and manual segmentations of these five bones was assessed using the Sørensen-Dice index (SDI). Reproducibility was evaluated applying the Bland-Altman method. Results: The median (SD) volumes of the five selected bones were by CNN and manual segmentation: Th7 41 (3.8) and 36 (5.1), L3 76 (13) and 75 (9.2), sacrum 284 (40) and 283 (26), 7th rib 33 (3.9) and 31 (4.8), sternum 80 (11) and 72 (9.2), respectively. Median SDIs were 0.86 (Th7), 0.85 (L3), 0.88 (sacrum), 0.84 (7th rib) and 0.83 (sternum). The intraobserver volume difference was less with CNN-based than manual approach: Th7 2% and 14%, L3 7% and 8%, sacrum 1% and 3%, 7th rib 1% and 6%, sternum 3% and 5%, respectively. The average volume difference measured as ratio volume difference/mean volume between the two CNN-based segmentations was 5–6% for the vertebral column and ribs and ≤3% for other bones. Conclusion: The new deep learning-based method for automated segmentation of bones in CT scans provided highly accurate bone volumes in a fast and automated way and, thus, appears to be a valuable first step in the development of a clinical useful processing procedure providing reliable skeletal segmentation as a key part of quantification of skeletal metastases.

AB - Purpose: The aim of this study was to develop a deep learning-based method for segmentation of bones in CT scans and test its accuracy compared to manual delineation, as a first step in the creation of an automated PET/CT-based method for quantifying skeletal tumour burden. Methods: Convolutional neural networks (CNNs) were trained to segment 49 bones using manual segmentations from 100 CT scans. After training, the CNN-based segmentation method was tested on 46 patients with prostate cancer, who had undergone 18 F-choline-PET/CT and 18 F-NaF PET/CT less than three weeks apart. Bone volumes were calculated from the segmentations. The network's performance was compared with manual segmentations of five bones made by an experienced physician. Accuracy of the spatial overlap between automated CNN-based and manual segmentations of these five bones was assessed using the Sørensen-Dice index (SDI). Reproducibility was evaluated applying the Bland-Altman method. Results: The median (SD) volumes of the five selected bones were by CNN and manual segmentation: Th7 41 (3.8) and 36 (5.1), L3 76 (13) and 75 (9.2), sacrum 284 (40) and 283 (26), 7th rib 33 (3.9) and 31 (4.8), sternum 80 (11) and 72 (9.2), respectively. Median SDIs were 0.86 (Th7), 0.85 (L3), 0.88 (sacrum), 0.84 (7th rib) and 0.83 (sternum). The intraobserver volume difference was less with CNN-based than manual approach: Th7 2% and 14%, L3 7% and 8%, sacrum 1% and 3%, 7th rib 1% and 6%, sternum 3% and 5%, respectively. The average volume difference measured as ratio volume difference/mean volume between the two CNN-based segmentations was 5–6% for the vertebral column and ribs and ≤3% for other bones. Conclusion: The new deep learning-based method for automated segmentation of bones in CT scans provided highly accurate bone volumes in a fast and automated way and, thus, appears to be a valuable first step in the development of a clinical useful processing procedure providing reliable skeletal segmentation as a key part of quantification of skeletal metastases.

KW - Artificial intelligence

KW - Bone

KW - Metastases

KW - PET/CT

KW - Prostate cancer

U2 - 10.1016/j.ejrad.2019.01.028

DO - 10.1016/j.ejrad.2019.01.028

M3 - Journal article

C2 - 30927965

VL - 113

SP - 89

EP - 95

JO - European Journal of Radiology

JF - European Journal of Radiology

SN - 0720-048X

ER -