Original research : utilization of a convolutional neural network for automated detection of lytic spinal lesions on body CTs
© 2023. The Author(s), under exclusive licence to International Skeletal Society (ISS)..
OBJECTIVE: To develop, train, and test a convolutional neural network (CNN) for detection of spinal lytic lesions in chest, abdomen, and pelvis CT scans.
MATERIALS AND METHODS: Cases of malignant spinal lytic lesions in CT scans were identified. Images were manually segmented for the following classes: (i) lesion, (ii) normal bone, (iii) background. If more than one lesion was on a single slice, all lesions were segmented. Images were stored as 128×128 pixel grayscale, with 10% segregated for testing. The training pipeline of the dataset included histogram equalization and data augmentation. A model was trained on Keras/Tensorflow using an 80/20 training/validation split, based on U-Net architecture. Additional testing of the model was performed on 1106 images of healthy controls. Global sensitivity measured detection of any lesion on a single image. Local sensitivity and positive predictive value (PPV) measured detection of all lesions on an image. Global specificity measured false positive rate in non-pathologic bone.
RESULTS: Six hundred images were obtained for model creation. The training set consisted of 540 images, which was augmented to 20,000. The test set consisted of 60 images. Model training was performed in triplicate. Mean Dice scores were 0.61 for lytic lesion, 0.95 for normal bone, and 0.99 for background. Mean global sensitivity was 90.6%, local sensitivity was 74.0%, local PPV was 78.3%, and global specificity was 63.3%. At least one false positive lesion was noted in 28.8-44.9% of control images.
CONCLUSION: A task-trained CNN showed good sensitivity in detecting spinal lytic lesions in axial CT images.
| Medienart: |
E-Artikel |
|---|
| Erscheinungsjahr: |
2023 |
|---|---|
| Erschienen: |
2023 |
| Enthalten in: |
Zur Gesamtaufnahme - volume:52 |
|---|---|
| Enthalten in: |
Skeletal radiology - 52(2023), 7 vom: 20. Juli, Seite 1377-1384 |
| Sprache: |
Englisch |
|---|
| Beteiligte Personen: |
Chang, Connie Y [VerfasserIn] |
|---|
| Links: |
|---|
| Themen: |
Computed tomography |
|---|
| Anmerkungen: |
Date Completed 18.05.2023 Date Revised 18.05.2023 published: Print-Electronic Citation Status MEDLINE |
|---|
| doi: |
10.1007/s00256-023-04283-x |
|---|
| funding: |
|
|---|---|
| Förderinstitution / Projekttitel: |
|
| PPN (Katalog-ID): |
NLM351627340 |
|---|
| LEADER | 01000naa a22002652 4500 | ||
|---|---|---|---|
| 001 | NLM351627340 | ||
| 003 | DE-627 | ||
| 005 | 20231226051837.0 | ||
| 007 | cr uuu---uuuuu | ||
| 008 | 231226s2023 xx |||||o 00| ||eng c | ||
| 024 | 7 | |a 10.1007/s00256-023-04283-x |2 doi | |
| 028 | 5 | 2 | |a pubmed24n1172.xml |
| 035 | |a (DE-627)NLM351627340 | ||
| 035 | |a (NLM)36651936 | ||
| 040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
| 041 | |a eng | ||
| 100 | 1 | |a Chang, Connie Y |e verfasserin |4 aut | |
| 245 | 1 | 0 | |a Original research |b utilization of a convolutional neural network for automated detection of lytic spinal lesions on body CTs |
| 264 | 1 | |c 2023 | |
| 336 | |a Text |b txt |2 rdacontent | ||
| 337 | |a ƒaComputermedien |b c |2 rdamedia | ||
| 338 | |a ƒa Online-Ressource |b cr |2 rdacarrier | ||
| 500 | |a Date Completed 18.05.2023 | ||
| 500 | |a Date Revised 18.05.2023 | ||
| 500 | |a published: Print-Electronic | ||
| 500 | |a Citation Status MEDLINE | ||
| 520 | |a © 2023. The Author(s), under exclusive licence to International Skeletal Society (ISS). | ||
| 520 | |a OBJECTIVE: To develop, train, and test a convolutional neural network (CNN) for detection of spinal lytic lesions in chest, abdomen, and pelvis CT scans | ||
| 520 | |a MATERIALS AND METHODS: Cases of malignant spinal lytic lesions in CT scans were identified. Images were manually segmented for the following classes: (i) lesion, (ii) normal bone, (iii) background. If more than one lesion was on a single slice, all lesions were segmented. Images were stored as 128×128 pixel grayscale, with 10% segregated for testing. The training pipeline of the dataset included histogram equalization and data augmentation. A model was trained on Keras/Tensorflow using an 80/20 training/validation split, based on U-Net architecture. Additional testing of the model was performed on 1106 images of healthy controls. Global sensitivity measured detection of any lesion on a single image. Local sensitivity and positive predictive value (PPV) measured detection of all lesions on an image. Global specificity measured false positive rate in non-pathologic bone | ||
| 520 | |a RESULTS: Six hundred images were obtained for model creation. The training set consisted of 540 images, which was augmented to 20,000. The test set consisted of 60 images. Model training was performed in triplicate. Mean Dice scores were 0.61 for lytic lesion, 0.95 for normal bone, and 0.99 for background. Mean global sensitivity was 90.6%, local sensitivity was 74.0%, local PPV was 78.3%, and global specificity was 63.3%. At least one false positive lesion was noted in 28.8-44.9% of control images | ||
| 520 | |a CONCLUSION: A task-trained CNN showed good sensitivity in detecting spinal lytic lesions in axial CT images | ||
| 650 | 4 | |a Journal Article | |
| 650 | 4 | |a Computed tomography | |
| 650 | 4 | |a Deep learning | |
| 650 | 4 | |a Osteolysis | |
| 650 | 4 | |a Vertebra | |
| 700 | 1 | |a Huber, Florian A |e verfasserin |4 aut | |
| 700 | 1 | |a Yeh, Kaitlyn J |e verfasserin |4 aut | |
| 700 | 1 | |a Buckless, Colleen |e verfasserin |4 aut | |
| 700 | 1 | |a Torriani, Martin |e verfasserin |4 aut | |
| 773 | 0 | 8 | |i Enthalten in |t Skeletal radiology |d 1993 |g 52(2023), 7 vom: 20. Juli, Seite 1377-1384 |w (DE-627)NLM001182935 |x 1432-2161 |7 nnns |
| 773 | 1 | 8 | |g volume:52 |g year:2023 |g number:7 |g day:20 |g month:07 |g pages:1377-1384 |
| 856 | 4 | 0 | |u http://dx.doi.org/10.1007/s00256-023-04283-x |3 Volltext |
| 912 | |a GBV_USEFLAG_A | ||
| 912 | |a GBV_NLM | ||
| 951 | |a AR | ||
| 952 | |d 52 |j 2023 |e 7 |b 20 |c 07 |h 1377-1384 | ||