Automatic classification of the vertebral endplate lesions in magnetic resonance imaging by deep learning model
IntroductionA novel classification scheme for endplate lesions, based on T2-weighted images from magnetic resonance imaging (MRI) scan, has been recently introduced and validated. The scheme categorizes intervertebral spaces as “normal,” “wavy/irregular,” “notched,” and “Schmorl's node.” These...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2023-06-01
|
| Series: | Frontiers in Surgery |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fsurg.2023.1172313/full |
| _version_ | 1797797737362620416 |
|---|---|
| author | Tito Bassani Andrea Cina Andrea Cina Fabio Galbusera Luca Maria Sconfienza Luca Maria Sconfienza Domenico Albano Federica Barcellona Alessandra Colombini Andrea Luca Marco Brayda-Bruno |
| author_facet | Tito Bassani Andrea Cina Andrea Cina Fabio Galbusera Luca Maria Sconfienza Luca Maria Sconfienza Domenico Albano Federica Barcellona Alessandra Colombini Andrea Luca Marco Brayda-Bruno |
| author_sort | Tito Bassani |
| collection | DOAJ |
| description | IntroductionA novel classification scheme for endplate lesions, based on T2-weighted images from magnetic resonance imaging (MRI) scan, has been recently introduced and validated. The scheme categorizes intervertebral spaces as “normal,” “wavy/irregular,” “notched,” and “Schmorl's node.” These lesions have been associated with spinal pathologies, including disc degeneration and low back pain. The exploitation of an automatic tool for the detection of the lesions would facilitate clinical practice by reducing the workload and the diagnosis time. The present work exploits a deep learning application based on convolutional neural networks to automatically classify the type of lesion.MethodsT2-weighted MRI scans of the sagittal lumbosacral spine of consecutive patients were retrospectively collected. The middle slice of each scan was manually processed to identify the intervertebral spaces from L1L2 to L5S1, and the corresponding lesion type was labeled. A total of 1,559 gradable discs were obtained, with the following types of distribution: “normal” (567 discs), “wavy/irregular” (485), “notched” (362), and “Schmorl's node” (145). The dataset was divided randomly into a training set and a validation set while preserving the original distribution of lesion types in each set. A pretrained network for image classification was utilized, and fine-tuning was performed using the training set. The retrained net was then applied to the validation set to evaluate the overall accuracy and accuracy for each specific lesion type.ResultsThe overall rate of accuracy was found equal to 88%. The accuracy for the specific lesion type was found as follows: 91% (normal), 82% (wavy/irregular), 93% (notched), and 83% (Schmorl's node).DiscussionThe results indicate that the deep learning approach achieved high accuracy for both overall classification and individual lesion types. In clinical applications, this implementation could be employed as part of an automatic detection tool for pathological conditions characterized by the presence of endplate lesions, such as spinal osteochondrosis. |
| first_indexed | 2024-03-13T03:52:53Z |
| format | Article |
| id | doaj.art-74f8bd313d9f4ef5926510e9df513337 |
| institution | Directory Open Access Journal |
| issn | 2296-875X |
| language | English |
| last_indexed | 2024-03-13T03:52:53Z |
| publishDate | 2023-06-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Surgery |
| spelling | doaj.art-74f8bd313d9f4ef5926510e9df5133372023-06-22T09:29:14ZengFrontiers Media S.A.Frontiers in Surgery2296-875X2023-06-011010.3389/fsurg.2023.11723131172313Automatic classification of the vertebral endplate lesions in magnetic resonance imaging by deep learning modelTito Bassani0Andrea Cina1Andrea Cina2Fabio Galbusera3Luca Maria Sconfienza4Luca Maria Sconfienza5Domenico Albano6Federica Barcellona7Alessandra Colombini8Andrea Luca9Marco Brayda-Bruno10IRCCS Istituto Ortopedico Galeazzi, Milan, ItalySpine Center, Schulthess Clinic, Zurich, SwitzerlandDepartment of Health Sciences and Technologies, ETH Zurich, Zurich, SwitzerlandSpine Center, Schulthess Clinic, Zurich, SwitzerlandIRCCS Istituto Ortopedico Galeazzi, Milan, ItalyDipartimento di Scienze Biomediche per la Salute, Università Degli Studi di Milano, Milan, ItalyIRCCS Istituto Ortopedico Galeazzi, Milan, ItalyComplex Unit of Radiology, Department of Diagnostic and Interventional Radiology, Azienda Socio Sanitaria Territoriale (ASST) Lodi, Lodi, ItalyIRCCS Istituto Ortopedico Galeazzi, Milan, ItalyIRCCS Istituto Ortopedico Galeazzi, Milan, ItalyIRCCS Istituto Ortopedico Galeazzi, Milan, ItalyIntroductionA novel classification scheme for endplate lesions, based on T2-weighted images from magnetic resonance imaging (MRI) scan, has been recently introduced and validated. The scheme categorizes intervertebral spaces as “normal,” “wavy/irregular,” “notched,” and “Schmorl's node.” These lesions have been associated with spinal pathologies, including disc degeneration and low back pain. The exploitation of an automatic tool for the detection of the lesions would facilitate clinical practice by reducing the workload and the diagnosis time. The present work exploits a deep learning application based on convolutional neural networks to automatically classify the type of lesion.MethodsT2-weighted MRI scans of the sagittal lumbosacral spine of consecutive patients were retrospectively collected. The middle slice of each scan was manually processed to identify the intervertebral spaces from L1L2 to L5S1, and the corresponding lesion type was labeled. A total of 1,559 gradable discs were obtained, with the following types of distribution: “normal” (567 discs), “wavy/irregular” (485), “notched” (362), and “Schmorl's node” (145). The dataset was divided randomly into a training set and a validation set while preserving the original distribution of lesion types in each set. A pretrained network for image classification was utilized, and fine-tuning was performed using the training set. The retrained net was then applied to the validation set to evaluate the overall accuracy and accuracy for each specific lesion type.ResultsThe overall rate of accuracy was found equal to 88%. The accuracy for the specific lesion type was found as follows: 91% (normal), 82% (wavy/irregular), 93% (notched), and 83% (Schmorl's node).DiscussionThe results indicate that the deep learning approach achieved high accuracy for both overall classification and individual lesion types. In clinical applications, this implementation could be employed as part of an automatic detection tool for pathological conditions characterized by the presence of endplate lesions, such as spinal osteochondrosis.https://www.frontiersin.org/articles/10.3389/fsurg.2023.1172313/fullspineendplate lesionsosteochondrosisartificial intelligencedeep learningautomatic classification |
| spellingShingle | Tito Bassani Andrea Cina Andrea Cina Fabio Galbusera Luca Maria Sconfienza Luca Maria Sconfienza Domenico Albano Federica Barcellona Alessandra Colombini Andrea Luca Marco Brayda-Bruno Automatic classification of the vertebral endplate lesions in magnetic resonance imaging by deep learning model Frontiers in Surgery spine endplate lesions osteochondrosis artificial intelligence deep learning automatic classification |
| title | Automatic classification of the vertebral endplate lesions in magnetic resonance imaging by deep learning model |
| title_full | Automatic classification of the vertebral endplate lesions in magnetic resonance imaging by deep learning model |
| title_fullStr | Automatic classification of the vertebral endplate lesions in magnetic resonance imaging by deep learning model |
| title_full_unstemmed | Automatic classification of the vertebral endplate lesions in magnetic resonance imaging by deep learning model |
| title_short | Automatic classification of the vertebral endplate lesions in magnetic resonance imaging by deep learning model |
| title_sort | automatic classification of the vertebral endplate lesions in magnetic resonance imaging by deep learning model |
| topic | spine endplate lesions osteochondrosis artificial intelligence deep learning automatic classification |
| url | https://www.frontiersin.org/articles/10.3389/fsurg.2023.1172313/full |
| work_keys_str_mv | AT titobassani automaticclassificationofthevertebralendplatelesionsinmagneticresonanceimagingbydeeplearningmodel AT andreacina automaticclassificationofthevertebralendplatelesionsinmagneticresonanceimagingbydeeplearningmodel AT andreacina automaticclassificationofthevertebralendplatelesionsinmagneticresonanceimagingbydeeplearningmodel AT fabiogalbusera automaticclassificationofthevertebralendplatelesionsinmagneticresonanceimagingbydeeplearningmodel AT lucamariasconfienza automaticclassificationofthevertebralendplatelesionsinmagneticresonanceimagingbydeeplearningmodel AT lucamariasconfienza automaticclassificationofthevertebralendplatelesionsinmagneticresonanceimagingbydeeplearningmodel AT domenicoalbano automaticclassificationofthevertebralendplatelesionsinmagneticresonanceimagingbydeeplearningmodel AT federicabarcellona automaticclassificationofthevertebralendplatelesionsinmagneticresonanceimagingbydeeplearningmodel AT alessandracolombini automaticclassificationofthevertebralendplatelesionsinmagneticresonanceimagingbydeeplearningmodel AT andrealuca automaticclassificationofthevertebralendplatelesionsinmagneticresonanceimagingbydeeplearningmodel AT marcobraydabruno automaticclassificationofthevertebralendplatelesionsinmagneticresonanceimagingbydeeplearningmodel |